Characterizing head and neck cancer survivors’ discontinuation of survivorship care

Abstract

Background:

Little is known about cancer survivors who discontinue survivorship care. Our objective was to characterize head and neck cancer patients who discontinue survivorship care with their treating institution and identify factors associated with discontinuation.

Methods:

We conducted a retrospective cohort study of patients diagnosed with head and neck cancer between January 1, 2014, and December 31, 2016, who received cancer-directed therapy at the University of Iowa Hospitals and Clinics (UIHC). Eligible patients achieved cancer-free status after curative-intent treatment and made at least one visit 90+ days after treatment completion. The primary outcome was discontinuation of survivorship care, defined as a still-living survivor who has not returned to a UIHC cancer clinic for twice the expected interval. Demographic and oncologic factors were examined to identify associations with discontinuation.

Results:

Ninety-seven (22.8%) of the 426 eligible patients discontinued survivorship care at UIHC during the study period. The mean time in follow-up for those discontinued treatment was 14.9 months. Factors associated with discontinuation of care included unmarried status (p=0.036), longer driving distance to facility (p=0.0031), and single-modality cancer treatment (p=<0.0001). Rurality was not associated with discontinuation (24.3% vs 21.6% for urban residence, p=0.52), nor was age, gender, or payor status.

Conclusions:

Study results indicate that a sizeable percentage of head and neck cancer survivors discontinue care with their treating institution. Both demographic and oncologic factors were associated with discontinuation at the treating institution, pointing to potential clinical and care delivery interventions.

Precis:

Using oncology registry and chart review of 426 head and neck cancer patients, we documented that 22.8% discontinued survivorship care at the treating institution. Demographic and oncologic factors associated with whether a patient continued survivorship care—distance to facility, marital status, and treatment modality—indicate potential loci for survivorship care optimization.

Introduction

Survival for people with a head and neck cancer (HNC) history has increased in the past two decades, the result of both improved therapeutic intervention and a marked increase in human papillomavirus-related oropharyngeal cancers, which have better prognoses.^1,2 As of 2017 the estimated 480,000 HNC survivors composed approximately 3% of all cancer survivors,³ and as more people with a HNC history are living and living longer, survivorship has become an increasingly important phase of HNC care delivery. Survivors can face a range of physical, psychological, and social challenges as a result of complex, often multimodal therapy involving surgery, radiation, and/or chemotherapy, including speech and swallowing difficulties; facial and shoulder pain, paralysis, and weakness; depression, anxiety, and diminished quality of life; and financial and employment challenges.^4-9 Along with persistent physical issues as a result of treatment and long-lasting psychosocial and financial impacts, HNC survivors’ risk for recurrence and second primary cancers engender the need for continued cancer surveillance and screening,^10-12 all of which reinforces the need for effective, long-term survivorship care.^13-15

Recent research has examined HNC survivors’ utilization of post-treatment follow-up cancer care, demonstrating that patient engagement in follow-up care decreases over time.^16,17 One group about whom little is known, however, is those HNC survivors who do not continue survivorship care with the oncology team at the institution where they received treatment. Recommendations stress the importance of multidisciplinary, team-based HNC survivorship care, which most often is provided at the treating institution.^13,15 Clinical experience indicates, however, that some percentage of post-treatment HNC patients, for various and often unknown reasons, do not continue with their survivorship care at the treating institution. Identifying those who do not continue follow up and understanding why they discontinue care is critical to addressing gaps in the care delivery system and potential disparities in HNC survivorship care. As a first step, in the current study, we have sought to characterize patients who, after being treated at our tertiary care academic medical center offering multidisciplinary HNC care, discontinued follow-up care at our center for any reason and to identify factors associated with loss to follow-up.

Methods

A retrospective cohort study was conducted in order to identify those who discontinued survivorship care at the University of Iowa Hospitals and Clinics (UIHC) and determine associated factors. The study received IRB approval from the University of Iowa’s Institutional Review Board (#201909737).

UIHC HNC Survivorship Care

Cancer survivorship care is variably defined, depending on the audience and the purpose.¹⁸ Within the UIHC Head and Neck Oncology program, survivorship care as it impacts clinical practice is defined as a distinct phase of cancer care that begins once treatment with curative intent has ended successfully. Multidisciplinary HNC survivorship care at UIHC is managed by a multidisciplinary team which can involve radiation oncology, medical oncology, speech pathology, medical nutrition, psychology, and social work. Survivors may be followed by a surgeon, radiation oncologist, medical oncologist, or a survivorship clinic, and they may be followed by more than one of those clinics. The Departments of Otolaryngology – Head and Neck Surgery and Radiation Oncology and the Clinical Cancer Center have dedicated survivorship clinics staffed by mid-level providers. At 2 years after treatment completion, patients, at the recommendation of their treating physician, may transition to survivorship care through the respective department’s survivorship clinic; those who do not transition to the survivorship clinics continue to be seen through the physicians’ clinics.

The follow-up care protocol of our Head and Neck Oncology program is based upon the National Comprehensive Cancer Network’s recommendations for follow-up examinations.¹⁹ Our multidisciplinary HNC team, including traditional and survivorship clinics, routinely recommends follow-up appointments every 3 months for 1 year after treatment, every 4 months for the second year, every 6 months for the third through fifth year, and annually after five years. The same schedule generally is used for patients with in situ and invasive cancer. Patients typically receive imaging to determine their cancer status at their first post-treatment visit at 3 months; additional surveillance and screening imaging is on an individualized schedule based upon risk factors, symptoms, and physical findings. Patients and their primary care physician (if they have one) receive a survivorship care plan, ideally three months post-treatment; the plan is delivered to the patient in person by the Otolaryngology – Head and Neck Surgery survivorship clinic advanced practice provider or an RN, who walks the patient through the information it contains.

Patients

Eligible subjects were patients who were seen through the Departments of Otolaryngology – Head and Neck Surgery, Radiation Oncology, or Clinical Cancer Center for in situ or invasive cancer of the nasal cavity, oral cavity, oropharynx, hypopharynx, nasopharynx, larynx, or major salivary gland, with date of diagnosis between January 1, 2014, and December 31, 2016. We excluded cases in which the index tumor was lymphoma or melanoma as well as lip cancers managed only in the Department of Dermatology. Patients whose diagnosis was made elsewhere were included, but patients who did not receive any cancer-directed therapy at UIHC were excluded. We also excluded patients referred to University of Iowa from Department of Veterans Affairs Medical Centers specifically for radiotherapy or surgical modalities not offered there, as their cancer care including some surgery, systemic therapy, and follow-up care was based outside our institution. We included only patients who achieved cancer-free status after curative-intent treatment confirmed during at least one visit 90 days after treatment completion. The observed study period ended on October 1, 2019. If patients had experienced cancer recurrence or second primary in the head and neck, we included them if all recurrences or second primary tumors had been treated with curative intent, and we defined the date of last treatment by the treatment of the most recent recurrence or primary.

Data Collection

Eligible subjects were identified through the institutional oncology registry, which houses patient-level data on all patients diagnosed or treated and is maintained by the Holden Comprehensive Cancer Center. Demographic and oncologic data were obtained for eligible subjects. A manual review of electronic health record information was conducted for each patient to verify information regarding diagnosis, treatment modality and end date, occurrence of recurrence, and post-treatment follow-up. Vital status was drawn from the oncology registry, which actively tracks vital status by multiple methods including patient charts, Social Security Death Index, and contact with physicians’ offices.

Primary Outcome: Discontinuation of Care

As described, HNC patients at UIHC who have undergone therapy with curative intent are considered survivors upon completion of their treatment. Discontinuation of care was determined based on the follow-up care protocol of our Head and Neck Oncology program. For this study, we defined discontinuation of survivorship care as having occurred if a survivor was alive, had achieved cancer free status, and had subsequently not returned to a UIHC cancer clinic in more than twice the interval expected for the length of time since treatment ended. Patients who left UIHC before a 90-day post-treatment follow up visit were excluded from the study cohort. The 90-day post-treatment follow up was included as a criterion both to delineate survivorship care from acute post-treatment care and to verify cancer free status. Across the range of patient experiences, many visits within the period one to three months after cancer treatment are part of acute care rather than survivorship care, e.g., visits for management of pain from acute oral mucositis, returns to the infusion suite for IV fluids, or returns to the surgeon’s clinic for removal of feeding tubes and management of wounds. In addition, verification of a patients’ cancer free status at the 90-day visit is important because those who do not achieve cancer-free status are managed differently. We defined discontinuation as occurring at their last clinic examination. For a patient whose last visit was at 12 months after treatment completion, for example, a status of discontinuation would be triggered at 20 months and 1 day post-treatment, given that they should have been seen at least every 4 months during the second post-treatment year; and the start date of their discontinuation would be their 12-month post-treatment examination.

Covariates

For each patient, demographic factors were recorded, including age, sex, race, marital and insurance status at diagnosis, receipt of radiotherapy and surgery, and driving distance between residence ZIP code and our institution. Cancer site, behavior (in situ or invasive), and receipt of radiation therapy and surgery were recorded. Patient residence addresses were used to calculate rurality using RUCA (Rural-Urban Commuting Area) codes, defined by census tract based on United States Census definitions and work commute patterns.²⁰ Patients were categorized as having rural residence if they resided in a census tract from which <30% of residents commuted to a census-defined urbanized area. Driving distance was operationalized as a dichotomous variable: <100 miles and ≥ 100 miles. We also conducted a sensitivity analysis using the thresholds <50 miles, 51-100 miles, >100 miles.

Data Analysis

Data were collected in REDCap and analyzed using SAS 9.4. Univariate assessment of discontinuation of follow-up by independent variables was done using logistic regression. To assess and compare duration of follow-up while accounting for potentially informative censoring at death, we used a competing risks model, computing cumulative incidence functions of discontinuation of follow-up, with study end set at October 1, 2019. We assessed the univariate effect of each covariate using competing risk proportional hazard regression according to the method of Fine and Gray.²¹ We then entered all covariates with univariate p-value < 0.1 into a multivariate Cox competing risk proportional hazard model. For all analyses, we used two-sided hypothesis tests and considered p<0.05 statistically significant.

Results

Patient Characteristics

Between January 2014 and December 2016, 564 patients were identified as having been treated at least partially at our institution with curative intent for new cancers at the specified primary sites (see Figure 1). Of those, 436 achieved cancer-free status and did not recur during the study period, while an additional 46 patients experienced recurrences which were treated with curative intent. A total of 426 patients from those 482 remained in follow-up at our institution at least 90 days after last cancer therapy; these 426 patients compose the study cohort. Over the course of follow-up, 42 of the 426 patients died.

Figure 1:

Flow Diagram of the Cohort

Within the study cohort (see Table 1), the majority of patients had been diagnosed with cancers of the oral cavity (41.3%), oropharynx (23.7%), or larynx (19.5%). The cohort was mostly white (95.8%), married or partnered (61.7%), and insured (private, Medicare, or Medicaid, 97.9%). Age and residence were more evenly split, with 53.3% of patients over the age of 60 and 42.5% of patients in areas designated as rural.

Table 1:

Characteristics of the cohort, N=426

Age at diagnosis	≤ 60	199 (46.7%)
	> 60	227 (53.3%)
Sex	Female	128 (30.1%)
	Male	298 (69.9%)
Race	White	408 (95.8%)
	Black	6 (1.4%)
	Other/Unknown	12 (2.8%)
Site	Oral cavity	176 (41.3%)
	Oropharynx	101 (23.7%)
	Salivary	33 (7.8%)
	Larynx	83 (19.5%)
	Other	33 (7.8%)
Histologic type	SCC or variant	367 (86.2%)
	Sarcoma/neurogenic	5 (1.2%)
	Other histology	54 (12.7%)
Behavior	In situ	32 (7.5%)
	Invasive	394 (92.5%)
Treatment (some had >1)	Surgery	342 (80.3%)
	Radiotherapy	242 (56.8%)
	Systemic therapy	132 (31.0%)
Clinical stage	0-2	197 (46.2%)
	3-4	197 (46.2%)
	Unknown	32 (7.6%)
Multidisciplinary treatment	Single modality	196 (46.0%)
	Multimodality	230 (54.0%)
Residence type	Urban	245 (57.5%)
	Rural	181 (42.5%)
Driving distance	< 100 miles	278 (65.3%)
	≥ 100 miles	148 (34.7%)
Marital status	Married/partnered	263 (61.7%)
	Single/separated/divorced/widowed	124 (29.1%)
	Unknown	39 (9.2%)
Insurance status	Private or NOS	216 (50.7%)
	Medicare/Tricare/VA	170 (39.9%)
	Medicaid	31 (7.3%)
	No insurance	9 (2.1%)
Treatment site	All Tx at UIHC	359 (84.3%)
	Some Tx elsewhere	67 (15.7%)
Survivorship care plan In the first year	Yes	95 (22.3%)
	No	331 (77.7%)

Factors Associated with Discontinuation of Survivorship Care

Of the entire cohort, 97 patients (22.8%) discontinued survivorship care at UIHC prior to death or the end of the observation period (see Table 2). Marital status, driving distance to facility, and treatment modality (single vs multimodality) were associated with loss to follow-up. Unmarried patients (including single, divorced, widowed, and separated) were more likely to discontinue survivorship care at UIHC (29.8%) than were married or partnered patients (18.6%, OR 1.86 [1.13 – 3.05, p=0.036]). Those whose driving distance to our institution was 100 miles or more were more likely to discontinue, compared to those who lived closer, 31.1% vs 18.4%, OR 2.01 (1.27 – 3.19, p=0.0031). When we used a logistic regression model with distance thresholds at both 50 and 100 miles, there was no difference between discontinuation rates of patients who drove fewer than 50 miles and 51-100 miles (OR 1.16 [0.58 – 2.32, p=0.28]) while there was a difference between the <50 mile and ≥100 mile groups (OR 2.30 [1.15 – 4.59, p=0.0025]). Therefore, we used only the 100-mile threshold in the analysis. Patients who received multimodality cancer treatment were also less likely to discontinue their survivorship care at UIHC compared to patients who underwent single-modality therapy (surgery or radiation; 32.7% vs 14.4%, OR 0.35 [0.22 – 0.56, p=<0.0001]); similarly, patients with AJCC Stage 3-4 were less likely to discontinue than patients with AJCC Stage 0-2 (16.8% vs 28.9%, OR 0.49 [0.30 – 0.80, p=0.017]).

Table 2:

Patient factors and discontinuation of survivorship care at UIHC, n=97

Characteristic	Factor	n (%) discontinuing follow up	Odds ratio (CI)	p
Age at diagnosis	> 60	50 (22.0)	0.91 (0.58 – 1.44)	0.70
	≤ 60	47 (23.6)	Reference
Sex	Female	31 (24.2)	1.12 (0.69 – 1.83)	0.64
	Male	66 (22.2)	Reference
Race	Black	1 (16.7)	0.69 (0.08 – 5.96)	0.64
	Other/Unknown	4 (33.3)	1.72 (0.51 – 5.84)
	White	92 (22.6)	Reference
Site	Oral cavity	46 (26.1)	3.54 (1.03 – 12.14)	0.34
	Oropharynx	21 (20.8)	2.62 (0.73 – 9.44)
	Salivary	8 (24.2)	3.20 (0.77 – 13.35)
	Larynx	19 (22.9)	2.97 (0.82 – 10.81)
	Other	3 (9.1)	Reference
Histologic type	SCC/variant	83 (22.6)	0.84 (0.43 – 1.61)	0.86
	Sarcoma/neurogenic	0 (0)	–
	Other histology	14 (25.9)	Reference
Behavior	In situ	11 (34.4)	1.88 (0.87 – 4.04)	0.11
	Invasive	86 (21.8)	Reference
Multidisciplinary treatment	Multimodality	33 (14.4)	0.35 (0.22 – 0.56)	<0.0001
	Single modality	64 (32.7)	Reference
Clinical stage AJCC 7th Ed.	3-4	33 (16.8)	0.49 (0.30 – 0.80)	0.017
	Unknown	7 (21.9)	0.69 (0.28 – 1.68)
	0-2	57 (28.9)	Reference
Residence type	Rural	44 (24.3)	1.16 (0.74 – 1.84)	0.52
	Urban	53 (21.6)	Reference
Driving distance	≥ 100 miles	46 (31.1)	2.01 (1.27 – 3.19)	0.0031
	< 100 miles	51 (18.4)	Reference
Marital status	Single/separated/div/wid	37 (29.8)	1.86 (1.13 – 3.05)	0.036
	Unknown	11 (28.2)	1.72 (0.80 – 3.68)
	Married/partner	49 (18.6)	Reference
Insurance status	Medicare, Tricare, VA	37 (21.8)	0.92 (0.57 – 1.50)	0.88
	Medicaid	7 (22.6)	0.97 (0.39 – 2.38)
	No insurance	3 (33.3)	1.66 (0.40 – 6.88)
	Private or NOS	50 (23.2)	Reference
Treatment site	Some Tx elsewhere	14 (20.9)	0.88 (0.46 – 1.66)	0.69
	All Tx at UIHC	83 (23.1)	Reference
Survivorship care plan in first year	Yes	22 (23.2)	1.03 (0.60 – 1.77)	0.92
	No	75 (22.7)	Reference

p-value corresponds to Wald chi-square for univariate logistic regression

Several factors were not significantly associated with discontinuation. Of note, patients with rural residence were as likely to discontinue survivorship care at UIHC as those with urban residence, 24.3% vs 21.6%, OR 1.16 (0.74 – 1.84, p=0.52). Patients over age 60 at diagnosis were as likely as younger patients to discontinue survivorship care, 22.0% vs 23.6%, OR 0.91 (0.58 – 1.44, p=0.70). Women and men were also equally likely to discontinue their UIHC-based survivorship care, 24.2% vs 22.2% respectively, OR 1.12 (0.69 – 1.83, p=0.70). Payor source was not significantly associated with discontinuation of care at UIHC, although uninsured survivors (n=3) comprised only 3.1% of those who discontinued. Patients who received a portion of their cancer treatment at another facility were equally likely to discontinue as those who received all treatment at our facility, 20.9% vs 23.1%, OR 0.88 (0.46 – 1.66). There was no difference in discontinuation rates in patients who received a survivorship care plan within the first year after treatment completion and those who did not (23.2% vs 22.7%), OR 1.03 (0.60 – 1.77).

Time to Discontinuation

The mean time in post-treatment follow-up care for the cohort was 32.7 months, while the 97 patients who discontinued their survivorship care at UIHC were in follow-up for a mean 15.4 months from completion of treatment; the median was not reached (Figure 2). On univariate competing risks analysis, multidisciplinary treatment was associated with lower hazard of discontinuing care at UIHC (HR 0.38 [0.25 – 0.58], p<0.0001); see Table 3 and Figure 3 for factors associated with length of time in survivorship care. Driving distance of 100 miles or more was associated with higher hazard of discontinuation (HR 1.81 [1.22 – 2.70], p=0.0034). Marital status was associated with time in survivorship care at UIHC, with single patients having a higher hazard of discontinuation than married or partnered patients (HR 1.70 [1.11 – 2.61], p=0.041). AJCC stage 3-4 was associated with a lower hazard of discontinuation than AJCC stages 0-2 (HR 0.52 [0.34 – 0.79], p=0.010). Again, several factors were not associated with longer times in survivorship care at UIHC. Patients from urban areas did not continue with their survivorship care at UIHC longer than those from rural areas (HR 1.15 [0.77 – 1.71], p=0.49). Age over 60 was not associated with time in survivorship care (HR 0.91 [0.61 – 1.36], p=0.65), nor was payer status (p=0.76).

Figure 2:

Discontinuation of Survivorship Care Over Time

Table 3:

Proportional hazard competing risks model of discontinuation of follow up

Variable	Level	Univariate		Multivariate
		Hazard Ratio (95% conf int)	p	Hazard Ratio (95% conf int)	p
Age at diagnosis	> 60	0.91 (0.61 – 1.36)	0.65
	≤ 60	Reference
Sex	Female	1.13 (0.74 – 1.73)	0.59
	Male	Reference
Race	Black	0.67 (0.11 – 4.23)	0.52
	Other/Unknown	1.67 (0.64 – 4.32)
	White	Reference
Primary site	Oral cavity	3.38 (1.06 – 10.76)	0.29
	Oropharynx	2.51 (0.75 – 8.34)
	Salivary	2.84 (0.76 – 10.51)
	Larynx	2.77 (0.83 – 9.27)
	Other	Reference
Histologic type	SCC or variant	0.90 (0.52 – 1.56)	0.72
	Sarcoma/neurogenic	–
	Other histology	Reference
Behavior	In situ	1.10 (0.91 – 3.20)	0.098	1.23 (0.65 – 2.31)	0.53
	Invasive	Reference		Reference
Multidisciplinary Treatment	Multimodality	0.38 (0.25 – 0.58)	<0.0001	0.36 (0.17 – 0.73)	0.0050
	Single modality	Reference		Reference
AJCC Stage	3-4	0.52 (0.34 – 0.79)	0.010	1.10 (0.53 – 2.28)	0.91
	Unknown	0.75 (0.34 – 1.63)		0.90 (0.41 – 1.98)
	0-2	Reference		Reference
Residence type	Rural	1.15 (0.77 – 1.71)	0.49
	Urban	Reference
Driving distance	≥ 100 miles	1.81 (1.22 – 2.70)	0.0034	1.90 (1.26 – 2.85)	0.0021
	< 100 miles	Reference		Reference
Marital status	Single/separated/div/wid	1.70 (1.11 – 2.61)	0.041	1.56 (1.01 – 2.41)	0.13
	Unknown	1.54 (0.80 – 2.97)		1.35 (0.70 – 2.62)
	Married/partner	Reference		Reference
Insurance status	Medicare/Tricare/VA	0.90 (0.59 – 1.38)	0.76
	Medicaid	0.97 (0.44 – 2.12)
	No insurance	1.68 (0.53 – 5.30)
	Private / NOS	Reference
Treatment site	Some Tx elsewhere	0.88 (0.50 – 1.56)	0.66
	All Tx at UIHC	Reference
Survivorship care plan 1st yr	Yes	1.01 (0.63 – 1.62)	0.96
	No	Reference

Figure 3:

Discontinuation of Survivorship Care Over Time, by Marital Status, Driving Distance, and Treatment Complexity

The multivariate proportional hazard analysis included behavior (in situ / invasive), multidisciplinary treatment (single vs. multimodality), driving distance ≥ 100 miles, AJCC stage, and marital status. In the multivariate analysis, driving distance ≥ 100 miles or more was associated with a higher hazard of discontinuation, (HR 1.90 [1.26 - 2.85], p=0.0021), as was multimodality treatment compared to single modality (HR 0.36 [0.17 - 0.73], p=0.0050). Marital status was not associated with an overall statistically significant hazard of discontinuation in the multivariate analysis (p=0.13), but unmarried survivors still had a higher hazard of discontinuation than married survivors (HR 1.56 [1.01 – 2.41]).

Discussion

Our data characterize patients who did not continue their survivorship care in an HNC clinic at our institution. We found that marital status, driving distance, and treatment modality were associated with utilization patterns of UIHC-based survivorship care, specifically with continuation or discontinuation. We also found that, among those who discontinue survivorship care, the mean time to discontinuation was just over one year from treatment completion. To the best of our knowledge, discontinuation as an outcome has not been examined. Other research has examined utilization of survivorship care, identifying demographic, oncologic, and care delivery factors associated with guideline-adherent cancer follow-up, as well as health care utilization for other acute and chronic conditions and preventive care.^22,23 However, as Kenzik noted in a recent review of health care utilization among cancer survivors, the majority of this research has been done among breast, prostate, and colorectal survivors. Recent work on HNC survivorship utilization, specifically, has shown that care utilization can vary widely between patients, can be bolstered by a survivor’s involvement (cancer-related or otherwise) with primary care, and frequently decreases over time.^16,17 This research reveals the patterns of those who engage in survivorship care but cannot provide insight into the experience of those who, for whatever reason, do not. As a complement to work on HNC survivorship care utilization, our findings are a first step toward a detailed understanding of how best to meet the needs of all HNC survivors.

Survivors in our cohort who had received multimodality cancer treatment remained in survivorship care longer than those who received single-modality treatment. This finding is concordant with that reported by others. Massa et al, in a prospective cross-sectional survey study, identified multidisciplinary care as a factor associated with desire for more frequent follow-up visits,²⁴ and Yao et al, in a SEER-Medicare-based retrospective study, found that treatment received was associated with HNC patients’ follow-up care utilization.¹⁶ Multimodality cancer treatment correlates with higher stage cancer, greater treatment complexity, and more severe treatment toxicity—all of which are factors that reasonably might result in survivor needs that could be addressed through continued survivorship care. Indeed, we also found clinical stage to have a similar effect on discontinuation as multidisciplinary care, which is unsurprising given the high degree of correlation between the two.

Our finding that married and partnered patients are more likely to continue survivorship care and remain in it longer than unmarried, widowed, or divorced patients provides another indication that spousal support is an important factor in patients’ engagement in survivorship care. Several studies have found better survival outcomes for married patients with head and neck cancer, and cancer in general, than for unmarried patients.^25-28 Among head and neck cancer survivors, in particular, perceived and actual social support also may help to counter psychosocial effects of depression and social isolation that survivors and their caregivers can experience.^29-32 It also is important to note that spousal support affects decision-making processes across their cancer care, from diagnosis to treatment and through survivorship, and that earlier decisions—treatment decisions, in particular—might mediate the relationship between marital status and survival.^25,33-35

Survivors in our study who traveled more than 100 miles to reach UIHC were more likely to discontinue their survivorship. This extends the finding of Massa et al that travel and work absence expenses were important sources of stress for survivors attending follow-up appointments.²⁴ This also aligns with other work that has shown the impact of increased travel on, especially, cancer diagnosis and treatment, as well as the potential burden imposed by the centralization of cancer care,^36-38 and importantly adds to the scant research on travel burden in the context of survivorship. Our findings point to the need for future research on the effect of travel burden on long-term survivorship care utilization patterns.

Although we found driving distance was associated with survivorship care discontinuation, we did not find an association between rurality and discontinuation at UIHC. This is in keeping with other research examining both rurality and travel distance,^39,40 and several possible explanations exist for this difference. One, our study’s cohort comprised 82 Iowa and Illinois counties of residence, which are characterized by small to mid-sized metropolitan areas interspersed with rural communities of varying size. In patients’ lives, this geography—along with the centralization of HNC specialty care—means that patients considered metropolitan by urban-rural classification systems may well have greater travel resource burdens than closer, yet rural-classified patients. Two, in contrast to other cancers such as rectal cancer in which an urban-rural divide is marked by differences in hospital and surgeon volume,^40-42 the treatment and care of HNC is characterized by strong centralization. Three, a rural-urban divide in HNC might be evident at earlier phases of the cancer care continuum,⁴³ such that our institution’s survivor population is already a select group. This finding, along with our research elsewhere indicating that rural HNC survivors report differences in psychosocial symptoms,⁴⁴ indicates the need to better understand the multilevel factors underlying rurality and their effects on HNC survivorship.⁴⁵

Our finding that the mean time to discontinuation was just over a year (15.4 months) for those who discontinued is noteworthy given that the risk of recurrence or metastatic disease for HNC survivors remains highest within the first two years.^10,46 Our work elsewhere has shown that surveillance for recurrence remains HNC survivors’ a top post-treatment priority long after that two year threshold.⁴ HNC survivors experience high levels of fear of recurrence, which impacts their quality of life and psychological health.^47,48 Understanding why some patients discontinue when their recurrence risk is at its highest is important to developing survivorship care that addresses their care needs and goals.

Our study has the advantages of a large sample size, a consecutive patient design, and verification of clinical data with detailed chart review. There are also limitations. The fact that this was a single-site study with a relatively homogenous patient population limits its generalizability. Additional limitations arise from the limits of administrative and chart data available for retrospective review. We categorized all patient visits to a cancer-related clinic as scheduled and included them in our analysis as such. It is likely that some visits were unscheduled and prompted by patient concerns; however, given that many patients have persistent symptoms that require intensive follow up, distinguishing clearly between scheduled and unscheduled visits was not possible. We also were not able to account for the clinical interventions, if any, that may have replaced survivorship care at our institution for each survivor. Some patients in our sample received a portion of their cancer care elsewhere, mostly at community hospitals nearer to their homes, and it is likely that some survivors chose follow-up with those cancer specialists or with community otolaryngologists. Although our previous work found it to be uncommon,⁴ a portion of our cohort may have conducted survivorship care with a primary care provider. Others may have felt comfortable foregoing further HNC care unless they experienced further symptoms. Another group, however, may still have survivorship care needs but are unable to address them, for example, because they have other health problems that are more pressing and have deprioritized head and neck cancer survivorship care. As part of our ongoing research, we plan to qualitatively investigate the cancer care activities of the cohort who have ceased follow-up at our institution.

Despite limitations, our study operationalizes HNC survivorship care discontinuation and demonstrates the potential of identifying and examining patients who discontinue survivorship care. For patients with a history of HNC, the research on ideal timing of follow-up care for improving detection of and survival outcomes from recurrence and second primaries has been mixed,⁴⁹ while research has shown that HNC survivors and their caregivers often have unmet care needs that survivorship care has the potential to address.^50,51 Increasingly there is a sense that individualized approaches to HNC survivorship care may be more beneficial.¹⁷ However, as HNC survivorship care research to date has focused on those survivors engaged in continued survivorship care, there is scant data on the range of patients who discontinue at the treating institution and their outcomes. This paucity limits the evidentiary base for survivorship care recommendations and models. By offering an explicit definition of discontinuation in HNC, our study also provides a starting point for conceptualizing survivorship care discontinuation for other cancer types by identifying design considerations and challenges. Alongside other research on survivorship care utilization patterns, a characterization of survivors who discontinue is important for understanding the needs and outcomes of those patients and provide new insights on what survivorship care means to all patients.

Conclusion

Almost one-fourth of head and neck cancer patients treated at UIHC discontinued their survivorship care with us, many within the first 15 months. Married patients, those with less than 100 mile driving distance from our institution, and those who had multidisciplinary cancer care stayed in follow-up longer than other patients. As survivorship care delivery interventions continue to be development and survivorship programs are implemented broadly, these findings highlight the need to consider those who do not or cannot continue their cancer care, the reasons they do so, and the alternatives they may seek out.