Burden of Comorbidities Is Higher Among Elderly Survivors of Oropharyngeal Cancer Compared With Controls

Abstract

BACKGROUND:

The prevalence of survivors of oropharyngeal cancer (OPC) is increasing due to improved survival for individuals with human papillomavirus (HPV)–related disease. Although elderly survivors of OPC are known to have a high burden of comorbidities, to the authors’ knowledge it is unknown how this compares with a similar cohort without a history of cancer.

METHODS:

The current retrospective, cross-sectional study included individuals with a first incident primary diagnosis of OPC from 2004 through 2011 from the Surveillance, Epidemiology, and End Results (SEER)–Medicare-linked databases and matched controls. The baseline prevalence and subsequent incidence of comorbid conditions were identified. The association between comorbidity and overall survival was evaluated.

RESULTS:

A total of 2497 eligible patients with OPC were matched to 4994 noncancer controls. Baseline comorbidity was higher in cases (Charlson Comorbidity Index >0 for 48.5% of cases vs 35.8% of controls). At 5 years, cases were more likely than controls to develop comorbidities. Survivors of OPC were at high risk (≥20% cumulative prevalence by 5 years) of developing several comorbidities, including cardiovascular diseases, cerebrovascular disease, chronic obstructive pulmonary disease, and tobacco abuse, and were at moderately high risk (10%-19% cumulative prevalence) of developing other conditions including carotid artery occlusive stroke, alcohol abuse, depression, and anxiety. In both cases and controls, the presence of the majority of comorbidities either at the time of diagnosis or during the follow-up period was associated with worse survival.

CONCLUSIONS:

Patients with OPC have a higher comorbidity burden compared with matched controls, both at baseline and during survivorship, the majority of which are associated with decreased survival. Oncologic surveillance of survivors of OPC should include screening for highly prevalent conditions.

INTRODUCTION

The incidence of oropharyngeal cancer (OPC) is rising in older age cohorts in the United States.^1–3 This increase is driven by human papillomavirus (HPV), with approximately 70% to 80% of OPC diagnoses in the United States estimated to be related to HPV, even among older individuals.^4–7 Indeed, the median age at the time of diagnosis of HPV-related OPC is rising.⁸

The combination of an increasing incidence of and good prognosis for HPV-related OPC^9–12 has led to an increased prevalence of long-term survivors.¹³ Between 2000 and 2012, the number of 5-year to 10-year survivors in the United States increased by an average of 115 per 100,000 individuals per year, with much of this increase occurring in patients aged ≥50 years.¹³ Therefore, the characteristics of this growing population of elderly survivors of OPC merits targeted study.

Both the prevalence and severity of comorbid conditions increase with age.¹⁴ The presence of comorbid disease impacts survival for patients with breast, lung, and colorectal cancers.¹⁵ For patients with head and neck cancer (HNC), those individuals with a higher comorbidity burden have lower overall survival (OS),^16,17 and the strength of this effect may vary by tumor site and HPV status.¹⁸

To the best of our knowledge, no study to date has examined how comorbidities and their prognostic impact differ between OPC cases and matched noncancer controls. The current study was designed to examine the prevalence of comorbid conditions at the time of OPC diagnosis, the incidence of new comorbid conditions during survivorship, and how each compares with controls. Comparing differences in comorbidity between survivors of OPC and noncancer controls is important to identify conditions for which survivors of OPC may benefit from additional focused surveillance beyond what currently is recommended.^19–21

MATERIALS AND METHODS

Study Design

The current retrospective study used the Surveillance, Epidemiology, and End Results (SEER)–Medicare-linked database to examine the prevalence of comorbid conditions for individuals with OPC as their first primary cancer at the time of diagnosis and during survivorship compared with matched noncancer controls. Medicare is a federal health insurance plan available to most Americans aged ≥65 years, and covers both inpatient (Part A) and outpatient (Part B) care. Data sources combine Medicare claims with data from the SEER registries (17 SEER registries covering a population-based sample of 26% of the US population, 23% with 16 of the 17 registries participating in the Medicare linkage). Data regarding noncancer controls were obtained from a 5% random sample of Medicare beneficiaries living in SEER regions.

Study Subjects

Squamous cell cancers of the base of the tongue (International Classification of Diseases for Oncology (ICD-O-3 code C019), palatine tonsil (C090-C099), oropharynx (C100-C109), and lingual tonsil (C024) were included. Cancer staging was assessed using the sixth edition guidelines of the American Joint Committee on Cancer. Additional eligibility criteria included age ≥66 years at the time of diagnosis and a first primary cancer diagnosis made between 2004 and 2011. Exclusion criteria included diagnosis by autopsy or death certificate, unknown race, unknown date of diagnosis, or receipt of palliative treatment only or no treatment administered within the first 6 months after diagnosis (see Supporting Table 1). It should be noted that HPV status was not available in the SEER-Medicare data during this time period.

Controls had to meet the same eligibility criteria except for the diagnosis of any cancer. Controls were matched 2:1 using age (within 3 years), race (white, black, or other), sex, and SEER region (combining Atlanta with rural Georgia and combining all California registries). A dummy diagnosis date equal to that of the matched case was used for all controls. Both cases and controls had to be continuously enrolled in fee-for-service Medicare Parts A and B from 1 year before diagnosis through the entire study period to December 31, 2013, or death. Any individual enrolled in managed care at any point during observation was excluded.²²

Statistical Analysis

Comorbid conditions of interest (see Supporting Table 2) were defined as prevalent if claims related to the diagnosis were present in the year before and up to 30 days after the cancer diagnosis. All comorbidities were analyzed individually using univariate models. Differences in prevalent diagnoses between cases and controls were assessed using conditional logistic regression modeling. Incident diagnoses were defined as conditions for which the first claim occurred after 30 days from diagnosis and were recorded until death or the end of the study. For those who were without the condition at the time of diagnosis, we calculated time from diagnosis to the date of the first claim for the comorbid condition, the end of the study, or death, whichever came first, as a time-to-event outcome. Participants with the condition at the time of diagnosis were included in the analysis as events and given a time-to-event value of 0. This approach accounted for variations in the prevalence of comorbidities at baseline and new diagnoses in the follow-up period while using all of the data. The cumulative probability of the diagnosis of each condition was estimated using competing risks regression, treating death as a competing event.²³ Cause-specific models, in which cases or controls who died without a diagnosis of comorbidity were censored at the time of death, were estimated as sensitivity analyses (see Supporting Table 3). The risk of developing comorbid disease was considered “high” for conditions with a ≥20% cumulative incidence by 5 years and “moderately high” for conditions with a 10% to 19% cumulative incidence by 5 years.

The median follow-up was calculated using the reverse Kaplan-Meier method. OS was calculated as the time from diagnosis to death or December 31, 2013, if the individual still was alive and was estimated using the Kaplan-Meier method. The association between prevalent comorbid conditions and death was assessed using Cox proportional hazards models. The association between the cumulative burden of comorbid diagnoses at the time of diagnosis and at follow-up and death was assessed by including the comorbid condition as a time-dependent covariate in Cox regression models. The consistency of the effect of the comorbidities on death between cases and controls was assessed using interaction analysis. All time-to-event analyses included a cluster term to identify the matched pairs for calculating robust standard errors and included the matched covariates of age, race, sex, and SEER region. Analyses were completed using SAS (version 9.4) and R (version 3.4.2) statistical software.²⁴

RESULTS

The study population (N = 7491) was comprised of 2497 OPC cases and 4994 matched noncancer controls. The median age was 72 years, and the majority were male (75%) and white (88%) (Table 1). The majority of cases had an advanced overall stage of disease (stage IV in 64.8%) and were treated with radiotherapy (93.6%). The median follow-up was 5.4 years and 5.7 years, respectively, for cases and controls.

TABLE 1.

Demographic and Clinical Characteristics of OPC Cases and Matched Noncancer Controls

Characteristic	OPC Cases N = 2497	Noncancer Controls N = 4994
Median age at diagnosis (Q1-Q3)	72 (69-77)	72 (68-77)
Sex, no. (%)
Male	1864 (74.6)	3728 (74.6)
Female	633 (25.4)	1266 (25.4)
Race, no. (%)
White (Caucasian)	2196 (87.9)	4392 (87.9)
Black (African American)	198 (7.9)	396 (7.9)
Other	103 (4.1)	206 (4.1)
SEER region, no. (%)
California	797 (31.9)	1594 (31.9)
Connecticut	153 (6.1)	306 (6.1)
Detroit	148 (5.9)	296 (5.9)
Hawaii	33 (1.3)	66 (1.3)
Iowa	106 (4.2)	212 (4.2)
New Mexico	61 (2.4)	122 (2.4)
Seattle	138 (5.5)	276 (5.5)
Utah	47 (1.9)	94 (1.9)
Georgia	334 (13.4)	668 (13.4)
Kentucky	185 (7.4)	370 (7.4)
Louisiana	167 (6.7)	334 (6.7)
New Jersey	328 (13.1)	656 (13.1)
Tumor location, no. (%)
Base of tongue	1265 (50.7)
Tonsil	952 (38.1)
Oropharynx	257 (10.3)
Lingual tonsil	23 (0.9)
Y of oropharyngeal cancer diagnosis, no. (%)
2004-2005	564 (22.6)
2006-2007	591 (23.7)
2008-2009	649 (26.0)
2010-2011	693 (27.8)
AJCC 6th edition stage of disease, no. (%)
0	18 (0.8)
I	119 (5.5)
II	185 (8.7)
III	433 (20.2)
IV	1392 (64.8)
Unknown	350
Treatment modality, no. (%)
Radiation only	470 (18.8)
Chemotherapy plus radiation	1341 (53.7)
Surgery only	159 (6.4)
Surgery plus radiation	165 (6.6)
Surgery plus chemotherapy with or without radiation	362 (14.5)

Abbreviations: AJCC, American Joint Committee on Cancer; OPC, oropharyngeal cancer; Q, quartile; SEER, Surveillance, Epidemiology, and End Results.

Baseline Comorbidities

At baseline, approximately half (51.5%) of cases had a Charlson Comorbidity Index of 0 compared with 64.2% of controls (P < .001). The majority of comorbidities were more prevalent in cases than controls (Table 2). Hypertension was the most common comorbidity for both, but its prevalence was higher among cases (57.3% vs 46.3%; P < .001). Other high-prevalence (≥20%) comorbidities that were more common in cases than controls included hyperlipidemia (32.8% vs 29.1%; P = .001) and chronic obstructive pulmonary disease (COPD) (21.0% vs 10.1%; P < .001). Moderately high-prevalence (10%-19%) conditions for cases included anemia (14.4% vs 9.0%; P < .001), tobacco abuse (13.2% vs 3.4%; P < .001), and cardiac disease (11.6% vs 8.9%; P < .001).

TABLE 2.

Prevalence of Comorbidities at the Time of Diagnosis Among OPC Cases and Matched Noncancer Controls

Condition of Interest No. (%)	OPC Cases N = 2497	Noncancer Controls N = 4994	P
Charlson Comorbidity Index
0	1287 (51.5)	3205 (64.2)	<.001
1	653 (26.2)	1056 (21.1)
≥2	557 (22.3)	733 (14.7)
Hypertension	1432 (57.3)	2314 (46.3)	<.001
Hyperlipidemia	818 (32.8)	1455 (29.1)	.001
COPD	524 (21.0)	502 (10.1)	<.001
Diabetes	509 (20.4)	1012 (20.3)	.90
Anemia	360 (14.4)	450 (9.0)	<.001
Tobacco abuse	329 (13.2)	169 (3.4)	<.001
Cardiac diseasea	289 (11.6)	444 (8.9)	<.001
Dysphagia	257 (10.3)	58 (1.2)	<.001
Cerebrovascular disease	219 (8.8)	236 (4.7)	<.001
Congestive heart failure	184 (7.4)	298 (6.0)	.02
Peripheral vascular disease	176 (7.0)	189 (3.8)	<.001
Weight loss	166 (6.6)	61 (1.2)	<.001
Chronic renal failure	145 (5.8)	198 (4.0)	<.001
Depression	112 (4.5)	159 (3.2)	.005
Alcohol abuse	109 (4.4)	26 (0.5)	<.001
Carotid stroke	108 (4.3)	101 (2.0)	<.001
Gastrointestinal disease	105 (4.2)	129 (2.6)	<.001
Myocardial infarction	95 (3.8)	133 (2.7)	.008
Pneumonia	93 (3.7)	122 (2.4)	.002
Malnutrition	88 (3.5)	33 (0.7)	<.001
Anxiety	73 (2.9)	114 (2.3)	.09
Angina	61 (2.4)	118 (2.4)	.83
Rheumatologic disease	45 (1.8)	85 (1.7)	.75
Dementia	35 (1.4)	75 (1.5)	.73
Liver disease	32 (1.3)	14 (0.3)	<.001
Dental disease	32 (1.3)	–b	<.001
Dysphonia	27 (1.1)	–b	<.001
Hip fracture	17 (0.7)	22 (0.4)	.18
Paralysis	–	15 (0.3)	.49
Disturbance in smell and/or taste	–b	–b	.24
Thrombolic event	–b	69 (1.4)	<.001
AIDS	–b	–b	.07
Tongue abscess	–b	–b	.37

Abbreviations: COPD, chronic obstructive pulmonary disease; OPC, oropharyngeal cancer.

^aComposite variable including congestive heart failure, myocardial infarction, and angina.

^bSuppressed to protect confidentiality due to the small cell size.

Presence of Baseline Comorbidities Is Associated With Decreased OS

The median OS was 3.8 years for cases and was not reached for controls (see Supporting Fig. 1). The majority of comorbidities at baseline conferred a worse OS for both populations (see Supporting Fig. 2), which is consistent with a previous analysis of cases.²⁵ Among the conditions that were found to be associated with the highest risk of death, 4, including malnutrition, weight loss, pneumonia, and dysphagia, were potentially related to the diagnosis of HNC.

Incident Comorbidities at 5 Years

For cases, the majority of incident comorbidities were diagnosed within the first year (median, 3 new diagnoses; range, 0-15 new diagnoses). The median overall number of comorbidities for cases over the study period was 4 (range, 0-18 comorbidities). Controls were diagnosed with fewer comorbidities, with a median of 1 comorbidity in the first year and 2 overall in the study period (range, 0-18 comorbidities).

The probability of being diagnosed with cardiovascular precursor conditions (hypertension, hyperlipidemia, or diabetes)²⁶ by year 5 was high (≥20%) (Fig. 1) for both cases and controls. For example, by year 5, hypertension was the most commonly diagnosed comorbidity for both populations, affecting 80% of cases and 78% of controls (hazard ratio [HR], 1.12; 95% CI, 1.08-1.18 [P < .001]). The risk of tobacco abuse was high for cases and moderately high for controls (46% vs 18%: HR, 3.10 [95% CI, 2.85-3.37; P < .001]). Cases were found to be at a reduced risk of hyperlipidemia and diabetes compared with controls (hyperlipidemia: 56% vs 64% [HR, 0.87; 95% CI, 0.82-0.92 (P < .001)]; and diabetes: 30% vs 36% [HR, 0.85; 95% CI, 0.78-0.92 (P < .001)]).

Figure 1.

Cumulative probability of being diagnosed with a comorbid condition by 5 years in cases with oropharyngeal cancer (OPC) and matched controls. COPD indicates chronic obstructive pulmonary disease.

The incidence of cardiovascular comorbidities among cases by 5 years was high for congestive heart failure (CHF) and moderately high for myocardial infarction (MI). The risk of CHF was similar for cases and controls (21% vs 19% [HR, 1.09; 95% CI, 0.99-1.21 (P = .08)]), whereas MI was more common among cases (15% vs 12% [HR, 1.23; 95% CI, 1.09-1.40 (P < .001)]).

Cases were more likely than controls to develop other high-risk comorbid conditions by year 5 including COPD (47% vs 26% [HR, 2.08; 95% CI, 1.93-2.24 (P < .001)]) and cerebrovascular disease (25% vs 21% [HR, 1.20; 95% CI, 1.09-1.32 (P < .001)]) (Fig. 2A). Cases were at moderately high risk of carotid artery occlusive stroke (14%) (Fig. 2B), depression (19%), anxiety (17%), and alcohol abuse (10%), and were 1.3-fold to 3.8-fold more likely than controls to develop these conditions. Both cases and controls were found to be at moderately high risk of chronic renal failure (17% in both groups; P = .29).

Figure 2.

Cumulative incidence curves of (A) cerebrovascular disease and (B) carotid artery occlusive stroke in cases with oropharyngeal cancer (OPC) compared with matched, noncancer controls.

Cases were significantly more likely to develop treatment-related comorbidities than controls. These included anemia (58% vs 34% [HR, 2.07; 95% CI, 1.93-2.22 (P < .001)]), dysphagia (63% vs 8% [HR, 11.07; 95% CI, 10.03-12.23 (P < .001)]), weight loss (46% vs 11% [HR, 5.28; 95% CI, 4.79-5.81 (P < .001)]), malnutrition (31% vs 5% [HR, 6.23; 95% CI, 5.47-7.10 (P < .001)]), and pneumonia (33% vs 13% [HR, 2.64; 95% CI, 2.39-2.90 (P < .001)]).

Comorbidities During Survivorship Are Associated With Decreased OS

The association between comorbid conditions and survival during the follow-up period was examined using a time-varying covariate analysis (Fig. 3). The majority of comorbid diagnoses were found to be associated with worse survival for both populations.

Figure 3.

Forest plots summarizing hazard ratios for the association between time-varying diagnoses of comorbid conditions and death, separately for oropharyngeal cancer (OPC) cases and controls. The size of the point was scaled to the number of cases.

Among the cardiovascular disease (CVD) precursors and conditions identified as having a high incidence among cases, tobacco abuse conferred a nearly 2-fold increased risk of death (HR, 1.87; 95% CI, 1.66-2.11), whereas the highly prevalent hypertension was found to have a modest effect on survival (HR, 1.15; 95% CI, 1.00-1.32). CHF (HR, 2.14; 95% CI, 1.88-2.45) and MI (HR, 1.71; 95% CI, 1.43-2.04) each were found to be associated with worse survival, and diabetes trended toward worse survival (HR, 1.13; 95% CI, 0.99-1.27). In contrast, hyperlipidemia (HR, 0.65; 95% CI, 0.58-0.73) conferred a reduced risk of death.

Other comorbidities that cases were found to be at high risk of developing during survivorship also significantly increased the risk of death, including COPD (HR, 2.18; 95% CI, 1.95-2.45) and cerebrovascular disease (HR, 1.31; 95% CI, 1.16-1.49). Of the treatment-related comorbidities, pneumonia (HR, 3.72; 95% CI, 3.27-4.23), malnutrition (HR, 2.96; 95% CI, 2.6-3.37), and weight loss (HR, 2.48; 95% CI, 2.21-2.8) were found to have the largest effects on survival. Diagnoses of anemia (HR, 1.68; 95% CI, 1.5-1.89) or dysphagia (HR, 1.54; 95% CI, 1.36-1.73) also significantly reduced survival.

Among the moderately high-risk comorbidities, alcohol abuse (HR, 3.3; 95% CI, 2.75-3.95), depression (HR, 1.63; 95% CI, 1.41-1.88), anxiety (HR, 1.5; 95% CI, 1.28-1.76), gastrointestinal disease (HR, 1.56; 95% CI, 1.34-1.81), and chronic renal failure (HR, 1.57; 95% CI, 1.35-1.83) were associated with worse survival for cases. For the majority of conditions, the increased risk of death conferred by a comorbid diagnosis was lower among cases compared with controls.

DISCUSSION

The growing population of survivors of OPC is at high risk of developing co morbid conditions.^13,27,28 To the best of our knowledge, the current analysis is the first to demonstrate that the burden of comorbid conditions during survivorship is significantly higher for patients with OPC compared with noncancer controls, and these conditions confer negative prognostic implications. Not only are comorbidities more likely at baseline, but the cumulative risk of developing the majority (74%) of conditions analyzed is moderately high (10%-19%) or high (≥20%) in the 5 years after diagnosis. Present guidelines offered by the National Comprehensive Cancer Network (NCCN), American Society of Clinical Oncology, and American Cancer Society include long-term, targeted oncologic surveillance and screening for treatment effects.^19–21 However, these guidelines offer no specific recommendations regarding the evaluation of medical comorbidities for which survivors are at increased risk during early survivorship. The results of the current analysis have provided important new data to guide the surveillance efforts of multidisciplinary head and neck oncologic and primary care teams for older survivors of OPC.

Survivors of OPC are at high risk of developing 13 conditions within the first 5 years after a diagnosis of malignancy, each of which is significantly associated with OS. These include CHF, COPD, cerebrovascular disease, CVD precursors (hypertension, hyperlipidemia, tobacco abuse, and diabetes), and treatment-related comorbidities (dysphagia, weight loss, malnutrition, anemia, and pneumonia). Survivors have a moderately high risk of developing 7 additional conditions, including MI, carotid artery occlusive stroke, depression, anxiety, alcohol abuse, gastrointestinal disease, and chronic renal failure.

Given the significant reductions in survival imparted by the majority of these moderately high-prevalence and high-prevalence conditions, additional screening measures and risk reduction efforts targeting these comorbidities should be considered. It is interesting to note that the majority of comorbid diagnoses were found to be associated with decreased OS for both cases and controls, but the magnitude of the effect was larger among controls; this may represent an ascertainment bias due to increased health care use by cases, with the resultant earlier detection of comorbidities among cases compared with controls.^29,30 This suggests that even earlier detection of these conditions could improve survival. Therefore, expanding surveillance efforts to include screening for these comorbidities and initiating appropriate treatments may reduce harm and improve survival. Although the current data set focused on older individuals, the prevention of these diagnoses among younger survivors of OPC who are likely to be at least at a similar risk of these conditions should be considered.

The high risk of developing CVD and its precursors is notable. Studies of survivors of other cancers, including breast, testicular, and prostate cancer as well as lymphoma, have demonstrated a similarly increased risk of CVD.³¹ Although there is overlap in risk factors for HNC and CVD, including older age and tobacco abuse, cardiotoxic chemotherapy treatments such as cisplatin may amplify this risk in survivors of OPC.³² There are well-described risk factors for CVD including tobacco abuse, hyperlipidemia, hypertension, diabetes, and alcohol use.³³ In the current analysis, survivors of OPC were found to be at high or moderately high risk of developing these conditions and, with the exception of diabetes and hyperlipidemia, each condition was substantially more common among cases. These high-prevalence and moderately high-prevalence comorbidities contribute to CVD, are modifiable, and can be treated medically or with lifestyle interventions. Therefore, the identification of these conditions should be considered as an important screening priority and a medical counseling opportunity that medical providers may be missing, according to the literature.³⁴

To the best of our knowledge, the current analysis is the first to demonstrate that the risk of carotid artery occlusive stroke among survivors with OPC significantly exceeds that of matched controls. OPC cases were found to be at moderately high risk of developing carotid stroke, with a cumulative prevalence of approximately 14% at 5 years. To our knowledge, there currently is no formal recommendation for carotid artery screening among survivors of HNC, although carotid stroke is a known sequela of head and neck irradiation. Carotid screening in asymptomatic survivors of HNC who were treated with radiation is controversial, without consensus from neuroradiology and vascular surgery guidelines regarding whether and when to screen.^35,36 However, data have suggested that it is cost-effective to screen for asymptomatic carotid artery stenosis (CAS) in populations with high (≥20%) or intermediate (5%-20%) prevalence if postoperative complications are low (<5%).^35,37 Asymptomatic CAS has been estimated to have an incidence of 12% to 23% among patients with irradiated HNC at 2 to 5 years of follow-up. In addition, data have suggested that CAS progression is accelerated compared with nonirradiated, noncancer controls.^38,39 Interventions including carotid angioplasty and stenting are safe and effective in reducing transient ischemic attack and carotid stroke in patients with HNC who were irradiated.^38,40 Therefore, implementing routine carotid artery screening as early as year 2 or 3 of survivorship may both prove cost-effective and reduce the incidence of stroke.

The risk of cerebrovascular disease among cases in the current study also exceeded that of controls. This finding highlights the question of whether survivors of OPC may benefit from more stringent targets for cholesterol management and other targeted stroke reduction interventions, as has been recommended in other high-risk populations, such as those individuals with diabetes.⁴¹ This represents an area of future investigation.

COPD also presents a screening opportunity. COPD is a highly prevalent condition that is related to smoking and old age that confers significant morbidity and mortality,^42,43 and the results of the current analysis demonstrated that survivors of OPC had twice the risk of developing COPD compared with controls. NCCN guidelines recommend computed tomography scans of the chest as clinically indicated during surveillance for patients with a smoking history.^42,44 In addition, screening with spirometry is inexpensive and easily performed in a primary care setting. As such, the routine screening of survivors of OPC for COPD during surveillance may be beneficial for early diagnosis and management.

Finally, the results of the current analysis have reinforced the importance of existing NCCN recommended supportive care and rehabilitation guidelines, including nutritional and swallowing support, smoking and alcohol cessation counseling, and depression screening. To the best of our knowledge, the current study is the first to demonstrate that survivors of OPC have between 1.5 times and 3.8 times the risk of experiencing tobacco or alcohol abuse, depression, and anxiety compared with controls, each of which significantly increases mortality risk. Patients with HPV-related OPC, who likely comprised the majority of the analytic population in the current study, are known to have lower rates of tobacco abuse compared with patients with traditional tobacco-related HNC. However, the risk of OPC is highest in current and former smokers, and the results of the current study have demonstrated that nearly one-half of survivors are diagnosed with tobacco abuse by year 5, which is known to increase the risk of CVD, stroke, COPD, tumor recurrence, and the development of new malignancies.^9,45–49

In addition, survivors are at moderately high risk of developing depression, anxiety, and alcohol abuse. Data have shown that the risk of suicide among survivors of HNC is nearly twice that of other cancer survivors, and that rate has increased since 2000.⁵⁰ The high prevalence and negative impact of developing substance abuse or psychosocial comorbidity taken together with the availability of office-based screening and treatment indicates that these conditions are important surveillance priorities.

There are several limitations to the current study. SEER-Medicare data include only those individuals aged ≥ 66 years, which may limit generalizability to younger populations. Furthermore, given that SEER-Medicare data did not provide information regarding the HPV status of tumors during the studied interval, we were unable to examine differences between HPV-related and non–HPV-related survivor populations. In addition, there is likely to be miscoding or underreporting of conditions, especially for behavioral diagnoses such as tobacco or alcohol use, which often are coded only when needed to justify care and do not quantify the degree of substance abuse. In addition, although the burden of each individual comorbid condition was explored as the outcome of interest in the current analysis, it is likely that some comorbid conditions are correlated. For example, tobacco smoking and cardiovascular conditions are known to be associated. Therefore, the presence of multiple comorbidities may impose a higher risk of death that was not assessed herein. Finally, although the current analysis highlighted comorbid conditions with prognostic significance and made the case for additional nononcologic surveillance of survivors of OPC, it is important to note that prior to instituting any population health initiatives, rigorous criteria must be met.⁵¹ Identification of the most appropriate screening methods and cost utility analyses, among other criteria, will be necessary.

These limitations notwithstanding, the current study addressed an important gap in our understanding and to the best of our knowledge is the first analysis to date to provide population-level estimates of the risk of comorbid conditions among survivors of OPC compared with population controls. Among the conditions studied, there were 20 comorbid conditions for which OPC cases were considered to be at high or moderately high risk of developing during the first 5 years (Fig. 4). Surveillance should not be limited to oncologic-related outcomes, but rather should include partnerships between head and neck oncologists and primary care physicians to provide appropriate medical management throughout survivorship.

Figure 4.

Comorbidities that reached or exceeded a cumulative prevalence of 10% for survivors of oropharyngeal cancer. COPD indicates chronic obstructive pulmonary disease.