Self‐Reported Cancer‐Related Cognitive Impairment and Psychosocial Symptom Burden in Survivors of Head and Neck Cancer

ABSTRACT

Background

Cancer‐related cognitive impairment (CRCI) is an increasingly recognized survivorship concern, yet its prevalence and associated clinical and psychosocial factors among survivors of head and neck cancer (HNC) remain poorly characterized.

Methods

This cross‐sectional study recruited survivors of HNC, who were at least 1‐year post‐diagnosis, from an institutional tumor registry to complete a comprehensive survey. Self‐reported cognitive function was assessed using the PROMIS Cognitive Function Short Form 8a, with a T‐score < 40 indicating cognitive impairment. Symptom burden was assessed using the EORTC QLQ‐HN43, PROMIS short forms, and the Insomnia Severity Index. Social role functioning was assessed using the PROMIS Satisfaction with Participation in Social Roles short form. Multivariable linear and logistic regression examined factors independently associated with self‐reported cognitive impairment.

Results

Of 337 survey respondents, 41 survivors (12.2%) met criteria for self‐reported moderate to severe cognitive impairment. Cognitively impaired survivors reported significantly higher symptom burden and lower social role functioning than survivors within normal limits. In multivariable linear regression (R ² = 0.520), poorer cognitive function was significantly associated with anxiety (β = −0.362, p < 0.001), fatigue (β = −0.223, p < 0.001), sleep disturbance (β = −0.261, p < 0.001), and social isolation (β = −0.178, p = 0.002). In logistic regression, anxiety (OR = 1.07, p = 0.035), fatigue (OR = 1.08, p = 0.021), and high‐intensity treatment (OR = 2.62, p = 0.028) were independently associated with impairment.

Conclusions

Approximately one in eight HNC survivors reported moderate‐to‐severe cognitive impairment. Self‐reported cognitive difficulty was associated with psychosocial symptom burden and multimodal treatment, and co‐occurred with poorer social role functioning. These findings support concurrent assessment of cognitive and psychosocial symptoms in HNC survivorship care.

1. Introduction

Cancer‐related cognitive impairment (CRCI) describes deficits in memory, attention, processing speed, and executive function that arise during or following cancer and its treatment [1, 2]. Its etiology is multifactorial, encompassing both direct neurotoxic effects of cancer treatments as well as indirect contributors, most prominently neuroinflammation [3, 4]. Although early research focused primarily on chemotherapy, other anti‐cancer treatments, including radiotherapy, targeted therapy, and immunotherapy, are now recognized as contributors to cognitive dysfunction through these shared neuroinflammatory pathways and modality‐specific mechanisms [3, 5, 6]. Beyond treatment‐related neurotoxicity, CRCI risk is further shaped by clinical factors such as aging, comorbidities, and tobacco and alcohol use [5, 7, 8, 9, 10], and psychosocial symptoms, including depression and fatigue [6, 11, 12].

CRCI can be assessed through both objective neuropsychological testing and patient‐reported measures, which are only modestly correlated and appear to capture distinct constructs [13]. Objective performance is most strongly associated with treatment‐related and demographic factors, while self‐reported cognitive difficulties tend to align more closely with mood, fatigue, and sleep patterns [14, 15]. This distinction is clinically important because CRCI is associated with reduced quality of life, diminished work performance, and social withdrawal, and represents one of the most distressing and burdensome concerns reported by cancer survivors, with symptoms persisting for years following treatment in a meaningful subset of individuals [6, 11, 16].

Despite a robust literature on CRCI, survivors of head and neck cancer (HNC) have received comparatively little attention, with most evidence derived from breast cancer or mixed‐cancer populations [6, 11, 16, 17]. However, the population of survivors of HNC continues to grow. Incidence of oral cavity and pharyngeal cancers has increased approximately 1% annually over the past decade, driven largely by HPV‐associated disease, while 5‐year relative survival has improved from 54% in the mid‐1970s to more than 75%, resulting in more than half a million survivors of oral cavity, pharynx and larynx cancers currently in the U.S. [18, 19] Characterizing CRCI in this growing population of survivors is therefore an emerging priority [20].

Several features of HNC and its treatment may make CRCI in this population distinctive. Treatment is frequently intensive and multimodal, incorporating surgery, radiation, chemotherapy, targeted therapy, and immunotherapy, including the recent adoption of pembrolizumab as standard first‐line treatment for recurrent or metastatic disease [21, 22, 23]. Radiation fields may lie in close proximity to neural structures involved in cognition. Survivors also contend with disease‐ and treatment‐related sequelae, including dysphagia, chronic pain, nutritional compromise, and changes in speech and appearance, which may further compound cognitive vulnerability and affect the functional consequences of cognitive symptoms in ways not fully captured in studies of other cancer populations [10]. Limited prospective evidence in patients with HNC documents significant post‐treatment declines in cognitive function, with rates of impairment as high as 38% at 2 years compared with 0% in controls [24]. Cognitive impairment has also been independently associated with increased anxiety, fatigue, and insomnia and reduced quality of life [10, 25].

To address these gaps, the present study characterized the prevalence of self‐reported cognitive impairment and associated sociodemographic, clinical, and symptom factors in a sample of HNC survivors. We hypothesized that self‐reported cognitive impairment would be present in a clinically meaningful subset of survivors and would be associated with greater psychosocial symptom burden, more intensive treatment exposure, and higher burden of comorbidity.

2. Materials and Methods

2.1. Study Design and Participants

This cross‐sectional study recruited survivors of HNC from an institutional tumor registry at an academic quaternary care center in 2019 to 2020. Eligible participants were adults (≥ 18 years) with a documented diagnosis of HNC who were between 1 and 5 years post‐diagnosis and able to complete the survey in English. The study was reviewed and approved by the University of California, Los Angeles Institutional Review Board (IRB‐19‐0798). All participants provided their consent upon survey submission and received a gift card incentive for their time and effort.

2.2. Data Collection

Participants were invited via electronic mail, when available, or by postal mail with two follow‐up contact attempts as needed. Individuals who chose to participate were sent a unique survey link and entered responses directly into Research Electronic Data Capture (REDCap) [26, 27], a secure, web‐based application, or a paper version when preferred. The comprehensive survey required approximately 60 min to complete and consisted of several validated questionnaires and project‐specific items.

2.3. Measures

2.3.1. Cognitive Function

Self‐reported cognitive function was assessed using the Patient‐Reported Outcomes Measurement Information System (PROMIS) Cognitive Function Short Form 8a (v2.0), a validated 8‐item measure assessing perceived difficulties with memory, attention, concentration, and mental processing speed in everyday life [28, 29, 30]. This instrument is recommended by the Cancer Neuroscience Initiative working group for measuring self‐reported cognitive impairment in cancer populations, and has demonstrated strong internal consistency and convergent validity with other patient‐reported cognitive measures in cancer survivors, including comparable performance to the widely used Functional Assessment of Cancer Therapy‐Cognitive Function Perceived Cognitive Impairment subscale [30, 31]. Items are scored on a 5‐point Likert scale and converted to T‐scores (population mean = 50, SD = 10), with higher scores indicating better cognitive function. A T‐score cutoff was used to classify patients as cognitively impaired (T < 40) versus within normal limits (T ≥ 40). This threshold represents one standard deviation below the population mean and corresponds to moderate‐to‐severe self‐reported cognitive impairment according to PROMIS severity thresholds proposed by in oncology populations [32].

2.3.2. HNC‐Specific Symptoms

HNC‐specific symptom burden was assessed using the European Organization for Research and Treatment of Cancer Head and Neck Cancer Module (EORTC QLQ‐HN43), a validated 43‐item questionnaire capturing 19 disease‐specific symptoms and functional domains [33]. Items are scored on a 4‐point Likert scale and transformed to a 0–100 scale per standard EORTC scoring guidelines; higher scores indicate greater symptom burden. Key domains examined include pain, swallowing, fear of recurrence, dry mouth, speech, social eating, coughing, neurological problems, and sensory problems including taste and smell.

2.3.3. Psychosocial Symptoms and Social Functioning

Anxiety, fatigue, depression, and social isolation were assessed using validated PROMIS short forms. PROMIS Anxiety (4‐item), PROMIS Fatigue (4‐item), PROMIS Depression (4‐item), and PROMIS Social Isolation (4‐item) short forms were scored using T‐scores (mean = 50, SD = 10 in the US general population), with higher scores indicative of greater symptom burden [29, 34, 35, 36]. Clinically elevated scores were defined as T ≥ 55 (one‐half SD above the population mean), corresponding to the operational threshold for clinically meaningful (mild or greater) symptoms on PROMIS measures [37]. Sleep disturbance was assessed using the Insomnia Severity Index (ISI), a validated 7‐item self‐report measure of sleep disturbance [38, 39]. Items are rated on a 5‐point Likert scale with total scores ranging from 0 to 28; higher scores indicate greater insomnia severity. Scores of 8–14 indicate subthreshold insomnia and scores ≥ 15 indicate clinical insomnia. Social role functioning was assessed using the PROMIS Satisfaction with Participation in Social Roles Short Form v1.0 4a, which measures self‐reported satisfaction with performing usual social roles and participating in daily activities, scored as a T‐score (mean = 50, SD = 10 in the US general population), with higher scores indicating greater satisfaction [40].

2.3.4. Sociodemographic and Clinical Characteristics

Sociodemographic information collected included age, gender, race/ethnicity, marital status, annual household income, employment status, and caregiver status. Clinical information was obtained from the tumor registry and self‐report and included primary cancer site, cancer stage, human papillomavirus (HPV) status, Charlson Comorbidity Index score [41, 42], and treatment modalities received (surgery, radiation, chemotherapy, immunotherapy, targeted therapy). Treatment data were additionally categorized into combination groups, including surgery only, surgery with radiotherapy, chemoradiation without surgery, and chemoradiation with surgery. High‐intensity treatment was defined as receipt of both chemotherapy and radiotherapy, with or without additional treatment modalities. Polypharmacy was defined as concurrent use of ≥ 5 prescribed medications per day [43].

2.4. Data Analysis

Descriptive statistics were used to characterize the analytic sample and compare sociodemographic, clinical, and symptom characteristics between those with and without self‐reported cognitive impairment (PROMIS Cognitive Function T‐score < 40). Continuous variables were compared using independent samples t‐tests or Wilcoxon rank‐sum tests, as appropriate, and categorical variables were compared using chi‐square tests or Fisher's exact tests. To account for multiple testing, Benjamini‐Hochberg false discovery rate correction was applied to bivariate comparisons of symptom domains and treatment modalities [44]. Differences in mean PROMIS Cognitive Function T‐scores across primary cancer sites and treatment groups were evaluated using one‐way analysis of variance (ANOVA).

All analyses used complete cases; missing data were not imputed. Cancer stage was unavailable for 52 participants (15.4%). Because cognitive function scores did not differ between survivors with known and unknown stage, stage was retained in multivariable analyses as a 3‐level variable (not advanced, advanced, unknown) rather than excluding participants with missing data.

Two multivariable regression models examined factors independently associated with cognitive function. A linear regression model used continuous PROMIS cognitive function T‐scores as the outcome, and a logistic regression model examined factors associated with moderate‐to‐severe self‐reported cognitive impairment (T < 40). Both models were adjusted for age, gender, Charlson Comorbidity Index score, and high‐intensity treatment exposure; the linear model additionally included time since diagnosis and cancer stage. Sensitivity analyses examined the independent contributions of depression and HNC‐specific symptom burden to self‐reported cognitive function, with HNC‐specific symptom burden operationalized as the number of EORTC QLQ‐HN43 symptom domains endorsed at any level. Statistical significance was defined as p < 0.05. All analyses were conducted in R (version 4.3.1) [45].

3. Results

3.1. Sample Characteristics

Among 1089 eligible registry‐identified individuals, 347 responded to the survey, yielding a 31.0% response rate. Of these, 337 (97.1%) provided complete cognitive function data and comprised the analytic sample. Compared to respondents, non‐respondents were more likely to identify as non‐White (p < 0.001) or Hispanic (p = 0.002) and be current smokers (p = 0.002); no significant differences were observed for other available demographic or clinical characteristics.

The analytic sample was primarily male (n = 245, 72.7%), White (n = 275, 82.6%), married (n = 240, 71.4%), and among those with available income data (n = 299), 52.8% reported annual household income ≥ $100,000. On average, patients were 61.9 years (SD = 11.8) at HNC diagnosis and 44.4 months (SD = 28.3) since diagnosis. The most common primary sites were the oropharynx (n = 186, 55.2%) and oral cavity (n = 95, 28.2%). Most survivors received surgery (n = 268, 80.0%) and radiotherapy (n = 265, 79.1%), and nearly half received chemotherapy (n = 159, 47.5%). HPV‐associated disease was present in 192 survivors (57.0%). Additional characteristics are presented in Table 1.

TABLE 1.

Sample characteristics by cognitive function status.

Variable	Overall	Normal	Impaired	p b
	(n = 337) a	(T ≥ 40) a	(T < 40) a
Gender
Male	245 (73%)	217 (73%)	28 (68%)	0.625
Female	92 (27%)	79 (27%)	13 (32%)
Marital status
Married	240 (71%)	213 (72%)	27 (66%)	0.51
Not Married	96 (29%)	82 (28%)	14 (34%)
Annual household income c
< $100K	141 (47%)	120 (46%)	21 (58%)	0.21
≥ $100 K	158 (53%)	143 (54%)	15 (42%)
Age at diagnosis (years)	61.9 (11.8)	61.9 (11.5)	61.6 (14.1)	0.885
Time Since Diagnosis (months)	44.4 (28.3)	44.5 (28.7)	43.9 (26.0)	0.898
Charlson comorbidity index (CCI)	0.51 (0.88)	0.48 (0.87)	0.73 (0.92)	0.11
Comorbidity Burden
None	222 (66%)	202 (68%)	20 (49%)	0.021
Any comorbidity present (CCI ≥ 1)	115 (34%)	94 (32%)	21 (51%)
Polypharmacy
< 5 Medications daily	256 (76%)	230 (78%)	26 (63%)	0.07
≥ 5 Medications daily	81 (24%)	66 (22%)	15 (37%)
Primary cancer site
Oral cavity	95 (28%)	87 (29%)	8 (20%)	0.210
Oropharynx	186 (55%)	161 (54%)	25 (61%)
Salivary gland	35 (10%)	32 (11%)	3 (7.3%)
Other pharynx	21 (6.2%)	16 (5.4%)	5 (12%)
HPV‐Related HNC
No	114 (34%)	101 (34%)	13 (32%)	0.247
Yes	192 (57%)	165 (56%)	27 (66%)
Not available	31 (9.2%)	30 (10%)	1 (2.4%)
Cancer stage
Not advanced (I‐II)	187 (55%)	162 (55%)	25 (61%)	0.802
Advanced (III‐IV)	98 (29%)	87 (29%)	11 (27%)
Unknown	52 (15%)	47 (16%)	5 (12%)
Treatment exposure (yes/no) d
Surgery	268 (80%)	236 (80%)	32 (80%)	> 0.999
Chemotherapy	159 (47%)	134 (45%)	25 (63%)	0.063
Radiotherapy	265 (79%)	229 (78%)	36 (88%)	0.209
Immunotherapy	37 (11%)	29 (9.8%)	8 (20%)	0.104
Targeted therapy	25 (7.4%)	21 (7.1%)	4 (9.8%)	0.525
Treatment intensity grouping d
Surgery only	58 (17%)	55 (19%)	3 (7.5%)	0.117
High‐intensity treatment	152 (45%)	127 (43%)	25 (63%)	0.032

^a Continuous: mean (SD); Categorical: n (%).

^b Pearson's Chi‐squared test; Welch Two Sample t‐test; Fisher's exact test.

^c Percentages for household income reflect available data (n = 299, 88.7%).

^d Treatment exposure variables are not mutually exclusive. Treatment intensity groupings are individually mutually exclusive comparison groups (Surgery Only; High‐Intensity: chemotherapy and radiotherapy, with or without additional modalities). Treatment information was missing for 2–4 participants depending on the variable.

3.2. Prevalence of Self‐Reported Cognitive Impairment

The mean PROMIS Cognitive Function T‐score for the overall sample was 51.3 (SD = 10.0), and 41 patients (12.2%) met criteria for self‐reported moderate to severe cognitive impairment (T < 40). T‐scores ranged from 22.4 to 63.5; the distribution of cognitive function T‐scores is presented in Figure 1.

FIGURE 1.

Distribution of PROMIS Cognitive Function T‐scores in survivors of HNC (n = 337). T‐scores reflect self‐reported cognitive function; higher scores indicate better function (population mean = 50, SD = 10; range = 22.4–63.5). The dashed line marks the T = 40 threshold, one standard deviation below the mean, corresponding to moderate to severe cognitive impairment.

3.3. Sociodemographic and Clinical Characteristics by Cognitive Function Group

Survivors with self‐reported cognitive impairment were more likely to report any comorbidity than survivors without cognitive impairment (51% vs. 32%, p = 0.021). Sociodemographic and clinical characteristics were otherwise comparable between groups, with no statistically significant differences observed by gender (p = 0.625), marital status (p = 0.510), income (p = 0.210), age at diagnosis (p = 0.885), time since diagnosis (p = 0.898), polypharmacy (p = 0.070), primary cancer site (p = 0.210), HPV status (p = 0.247), or cancer stage (p = 0.802) (Table 1).

3.4. Treatment Exposures by Cognitive Function Group

Rates of self‐reported cognitive impairment varied across treatment combination groups, ranging from 5.2% among survivors treated with surgery alone to 18.5% among those receiving chemoradiation plus surgery (Figure 2). Although differences across treatment groups did not reach statistical significance (χ ² p = 0.073), self‐reported cognitive impairment was significantly more prevalent among survivors who received high‐intensity treatment than among those who did not (16.4% vs. 8.2%, p = 0.032). Thirty‐seven survivors (12%) received immunotherapy, among whom 21.6% reported cognitive impairment compared with 11% of those who did not receive immunotherapy, although this difference was not statistically significant (p = 0.104). When examined individually, no treatment modality was significantly associated with self‐reported cognitive impairment after correction for multiple comparisons; the highest unadjusted proportion was among survivors who received chemotherapy (15.7% vs. 8.5%; p = 0.045, q = 0.225). Because radiation fields vary by anatomic subsite, mean cognitive function T‐scores were also compared across primary cancer sites; no significant differences were observed (ANOVA p = 0.806).

FIGURE 2.

Percentage of survivors reporting cognitive impairment (PROMIS Cognitive Function T‐score < 40) by treatment combination group. Mean PROMIS Cognitive Function T‐scores are displayed within each bar. 22 survivors receiving other treatment combinations not shown (n = 315); chi‐square p = 0.073.

3.5. Symptom Burden by Cognitive Function Group

Compared with survivors reporting cognitive function within normal limits, those with self‐reported cognitive impairment reported significantly higher symptom severity across all HNC‐specific, psychosocial, and behavioral symptom domains (Table 2). Among HNC‐specific physical symptoms, mean EORTC subscale scores were significantly higher in cognitively impaired patients, including mouth pain (45.7 vs. 23.8, p < 0.001), swallowing difficulties (45.2 vs. 21.7, p < 0.001), dry mouth (65.0 vs. 47.8, p < 0.001), chemosensory dysfunction including taste and smell (47.5 vs. 24.7, p < 0.001), and coughing (42.3 vs. 21.9, p < 0.001). Speech problems (40.2 vs. 22.6, p = 0.002), social eating restrictions (41.3 vs. 25.1, p = 0.002), neurological problems (48.3 vs. 27.6, p = 0.004), and fear of recurrence (61.7 vs. 37.1, p < 0.001) were also significantly elevated in the cognitively impaired group.

TABLE 2.

Symptoms and complications by cognitive function status.

Variable	Overall	Normal	Impaired	p b
	(n = 337) a	(T ≥ 40) a	(T < 40) a
HNC‐Specific Symptoms
Fear of recurrence	40.0 (31.9)	37.1 (30.8)	61.7 (31.6)	< 0.001
Swallowing difficulties	24.5 (28.3)	21.7 (26.9)	45.2 (30.1)	< 0.001
Mouth pain	26.5 (27.7)	23.8 (26.5)	45.7 (28.9)	< 0.001
Sensory problems (Taste/Smell)	27.4 (29.6)	24.7 (28.8)	47.5 (27.9)	< 0.001
Coughing	24.4 (28.6)	21.9 (27.0)	42.3 (33.4)	< 0.001
Speech problems	24.5 (27.4)	22.6 (26.5)	40.2 (29.7)	0.002
Social eating restrictions	27.0 (30.4)	25.1 (30.1)	41.3 (29.9)	0.002
Neurological problems	30.1 (35.7)	27.6 (34.2)	48.3 (41.3)	0.004
Dry mouth	49.9 (33.5)	47.8 (33.8)	65.0 (26.9)	< 0.001
Psychosocial and behavioral symptoms and social functioning
Anxiety T‐score	49.7 (9.3)	48.2 (8.4)	60.3 (8.6)	< 0.001
Fatigue T‐score	49.3 (8.9)	47.9 (8.2)	58.8 (7.9)	< 0.001
Depression T‐score	48.8 (8.6)	47.5 (7.8)	58.4 (8.5)	< 0.001
Social isolation T‐score	44.0 (9.3)	42.8 (8.7)	53.0 (8.8)	< 0.001
Sleep disturbance (ISI) score	8.2 (6.6)	7.3 (6.0)	15.1 (6.6)	< 0.001
Social role functioning T‐score	50.8 (10.2)	52.1 (9.8)	41.3 (8.5)	< 0.001

^a Mean (SD); n(%).

^b Welch 2 Sample t‐test; Pearson's Chi‐squared test.

Across psychosocial domains, mean PROMIS T‐scores were significantly higher in cognitively impaired patients compared to those with normal cognitive function, including anxiety (60.3 vs. 48.2, p < 0.001), fatigue (58.8 vs. 47.9, p < 0.001), social isolation (53.0 vs. 42.8, p < 0.001), and depression (58.4 vs. 47.5, p < 0.001). Sleep disturbance scores were also significantly higher in the cognitively impaired group (ISI: 15.1 vs. 7.3, p < 0.001). Survivors reporting cognitive impairment were substantially more likely to endorse psychosocial symptoms at clinically elevated levels (T ≥ 55), including anxiety (80.5% vs. 25.1%), depression (63.4% vs. 19.0%), fatigue (63.4% vs. 20.0%), and social isolation (45.0% vs. 12.2%) (all p < 0.001), with symptom comparisons remaining significant after Benjamini‐Hochberg correction (all q ≤ 0.004). Social role functioning was also significantly poorer among survivors reporting cognitive impairment than those within normal limits (mean T‐score 41.3 vs. 52.1, p < 0.001).

3.6. Factors Associated With Self‐Reported Cognitive Impairment

In the multivariable linear regression model, PROMIS cognitive function T‐scores were independently associated with psychosocial symptoms after controlling for age, gender, time since diagnosis, cancer stage, comorbidity score, and high‐intensity treatment exposure (Table 3). Higher anxiety (β = −0.362, p < 0.001), fatigue (β = −0.223, p < 0.001), sleep disturbance (β = −0.261, p < 0.001), and social isolation (β = −0.178, p = 0.002) scores were each independently associated with lower PROMIS Cognitive Function T‐scores. High‐intensity treatment was also independently associated with lower cognitive function (β = −2.022, p = 0.016). The model explained 52.0% of variance in cognitive function T‐scores (R ² = 0.520, Adjusted R ² = 0.503, p < 0.001).

TABLE 3.

Multivariable linear regression of factors associated with cognitive function T‐score ^a .

Characteristic	β	95% Confidence interval	p
Anxiety T‐score	−0.362	−0.484, −0.241	< 0.001
Fatigue T‐score	−0.223	−0.341, −0.105	< 0.001
Sleep disturbance (ISI) score	−0.261	−0.414, −0.107	< 0.001
Social Isolation T‐score	−0.178	−0.288, −0.068	0.002
Age (years)	−0.056	−0.125, 0.013	0.113
Gender (male vs. female)	−0.881	−2.689, 0.928	0.339
Time since diagnosis (months)	−0.005	−0.032, 0.023	0.740
Cancer Stage
Not advanced	Reference
Advanced	1.250	−0.655, 3.154	0.198
Unknown	0.554	−1.660, 2.767	0.623
Charlson comorbidity index	−0.909	−1.819, 0.002	0.051
High intensity treatment	−2.022	−3.668, −0.376	0.016

^a Model adjusted for age, gender, time since diagnosis, cancer stage (advanced, not advanced, or unknown), comorbidity score, and high‐intensity treatment; R ² = 0.520; Adjusted R ² = 0.503; N = 326.

In the logistic regression model, anxiety, fatigue, and high‐intensity treatment were independently associated with moderate to severe self‐reported cognitive impairment (T < 40) (Table 4). Each one‐point increase in anxiety T‐score was associated with 7% higher odds of cognitive impairment (OR = 1.07, 95% CI: 1.01–1.15, p = 0.035), and each one‐point increase in fatigue T‐score was associated with 8% higher odds (OR = 1.08, 95% CI: 1.01–1.16, p = 0.021). Sleep disturbance and social isolation were not independently associated with moderate‐to‐severe self‐reported cognitive impairment in the adjusted model (p = 0.159 and p = 0.217, respectively). High‐intensity treatment was associated with over two‐fold higher odds of cognitive impairment (OR = 2.62, 95% CI: 1.13–6.36, p = 0.028).

TABLE 4.

Logistic regression of factors associated with self‐reported cognitive impairment (T < 40) ^a .

Characteristic	Odds ratio	95% Confidence interval	p
Anxiety T‐score	1.073	1.005, 1.146	0.035
Fatigue T‐score	1.082	1.014, 1.159	0.021
Sleep disturbance (ISI Score)	1.052	0.981, 1.132	0.159
Social isolation T‐score	1.035	0.981, 1.094	0.217
Age (years)	1.005	0.971, 1.041	0.768
Gender (male vs. female)	1.277	0.508, 3.416	0.612
Charlson comorbidity index	1.120	0.719, 1.697	0.600
High‐intensity treatment	2.616	1.129, 6.364	0.028

^a Cognitive impairment defined as PROMIS Cognitive Function T‐score < 40. Model adjusted for age, gender, comorbidity score, and high‐intensity treatment exposure. Cancer stage and time since diagnosis not included given limited number of events (n = 41). N = 326.

In sensitivity analyses to the primary linear model, depression was not independently associated with self‐reported cognitive function when modeled alongside anxiety, fatigue, sleep disturbance, and social isolation (β = −0.009, 95% CI: −0.164 to 0.146, p = 0.909; Table S1). Variance inflation factors (VIFs) were highest for anxiety (3.26) and depression (2.98), whereas all other variables in the model had VIFs below 2.0. Similarly, HNC‐specific symptom burden, defined as the number of EORTC QLQ‐HN43 symptom domains endorsed at any level, was not independently associated with cognitive function after adjustment for covariates (β = 0.023, 95% CI: −0.193 to 0.240, p = 0.832; Table S2).

4. Discussion

Cognitive impairment is an increasingly recognized and clinically significant survivorship concern among patients with HNC, with implications for quality of life, functional status, and long‐term recovery. In this cross‐sectional study of 337 HNC survivors, approximately one in eight individuals reported moderate‐to‐severe cognitive impairment. Lower cognitive function scores were independently associated with anxiety, fatigue, sleep disturbance, and social isolation, while only anxiety and fatigue were also associated with meeting the threshold for moderate‐to‐severe impairment. High‐intensity treatment was independently associated with poorer self‐reported cognitive outcomes in both models.

4.1. Prevalence and Measurement Considerations

The observed prevalence of self‐reported cognitive impairment (12.2%) is comparable to prior HNC‐specific estimates of 14.8% [20] and 11% [46] among survivors of oropharyngeal cancer, and lower than estimates observed in cancer survivor populations more broadly [6, 47], survivors of breast cancer [48], and hematologic malignancy populations [49], and similar to colorectal cancer survivors [9], although direct comparison is limited by variation in measurement approaches.

An important methodological consideration is the absence of a consensus impairment threshold for the PROMIS Cognitive Function measure. In this study, we applied a threshold of T < 40, corresponding to the moderate‐to‐severe range developed through clinician‐rated standard setting in oncology populations [32]. The more stringent, ROC‐derived cut‐off proposed more recently by Henneghan and colleagues [30] (T < 34) would have yielded a substantially lower estimate in this sample (5.3% vs. 12.2%). Of note, the empirically derived threshold falls close to the boundary for severe impairment (T < 35) in Rothrock et al.'s framework [32], despite arising from unrelated methods. This proximity may suggest some convergence in the mid‐30s range, although both thresholds are anchored to subjective rather than objective criteria, and neither has been validated against objective neuropsychological performance. In addition, the threshold proposed by Henneghan et al. [31] was derived in a predominantly female cohort of breast cancer survivors, whereas the present sample consisted primarily of male survivors of HNC. Therefore, threshold selection carries important implications for prevalence estimates and identification of survivors if incorporated in routine survivorship care.

4.2. Psychosocial Symptoms and Social Participation

The association between psychosocial symptoms and self‐reported cognitive impairment is consistently reported across cancer populations [50]. The relationship between cognitive concerns and social participation has received less attention, particularly among HNC survivors. In cancer survivors more broadly, perceived cognitive‐communication difficulties have been linked to reduced social engagement and lower satisfaction with social participation [51]. Among HNC survivors, communicative participation, or the ability to engage in everyday communication activities, appears to be shaped by a combination of cognitive, speech‐related, and psychosocial factors, with self‐reported cognitive function among the strongest predictors [52, 53]. Few studies, however, have jointly examined cognitive and psychosocial contributors to social participation in this population.

Contemporary models conceptualize cognitive and psychosocial symptoms as closely interrelated components of CRCI [12]. In the present sample, survivors reporting cognitive impairment endorsed higher psychosocial symptom burden and markedly lower satisfaction with participation in social roles (mean T‐score, 41.3 versus 52.1). Cognitive function in this study was more consistently associated with psychosocial symptoms rather than clinical or treatment factors, a pattern that mirrors findings reported in broader CRCI studies [50].

These findings may be particularly relevant in the context of HNC survivorship where cognitive difficulties frequently co‐occur with speech and communication challenges, swallowing dysfunction, altered eating, and reduced social engagement [54, 55, 56]. Cognitive symptoms implicated in CRCI, such as word finding difficulties, forgetfulness, and slowed processing, may therefore compound communication and social participation challenges already commonly faced by this population, with serious implications for quality of life and daily functioning.

HNC survivors experience depression at rates substantially higher than many other cancer populations [57], and baseline depression has been identified as a predictor of subsequent cognitive decline using objective measures [24]. In the present analysis, however, depression was not independently associated with cognitive function once anxiety, fatigue, sleep disturbance, and social isolation were modeled simultaneously, with VIFs suggesting moderate overlap among psychosocial domains. Findings across cancer populations vary in whether depression retains independent association with cognitive symptoms when co‐occurring psychosocial symptoms are modeled together, consistent with a psychoneurological symptom cluster [12, 50, 58]. However, disentangling unique from shared contributions will require longitudinal designs and analytic approaches suited to interrelated symptoms, which remain limited in CRCI research [12].

4.3. Treatment Exposures and Self‐Reported Cognitive Outcomes

High‐intensity treatment was independently associated with poorer cognitive outcomes, with more than 2‐fold higher odds of impairment, while individual treatment modalities were not. Cumulative treatment burden may be more relevant to self‐reported cognitive outcomes than any single exposure, although the single modality findings may also reflect reduced statistical power and collinearity. Because the high‐intensity category was defined by receipt of chemotherapy and radiotherapy with or without additional modalities, this contrast reflects aggregate burden and remains subject to confounding, as more intensive treatment is often given for more advanced disease.

Research in HNC has focused largely on radiotherapy, given the proximity of treatment fields to structures involved in memory and cognition [20], and to some extent, cisplatin, a commonly used chemotherapy agent in HNC [59]. Notably, much of this evidence derives predominantly from objective neuropsychological testing rather than patient‐reported outcomes [17, 24]. Recent work suggests that dosimetry‐based predictors of objective cognitive impairment differ from predictors of self‐reported cognitive impairment, with subjective difficulties relating more closely to fatigue and mood symptoms [17, 20]. The present analysis extends this literature to self‐reported cognitive function in a contemporary, multi‐subsite HNC sample, in contrast to prior studies with predominantly nasopharyngeal populations.

Among the 37 survivors who received immunotherapy, exposure was not significantly associated with cognitive outcomes, although self‐reported impairment was more common than among those who did not receive immunotherapy (21.6% vs. 11%). Interpretation is limited by the small number of exposed participants and by potential confounding, as immunotherapy during the study period was primarily reserved for recurrent or metastatic disease. As checkpoint inhibitors move into earlier lines of HNC treatment and become relevant to a substantially larger survivor population, their potential cognitive effects warrant continued investigation.

4.4. Clinical Implications

Cognitive concerns remain inconsistently addressed in HNC survivorship care. While anxiety, depression, and fatigue are each addressed by dedicated clinical practice guidelines [60, 61], guidance for CRCI remains less prescriptive, without a validated brief screening instrument, consensus impairment threshold, or clear criteria for formal evaluation. As a result, cognitive difficulties may go unrecognized. The NCCN Survivorship guidelines recommend periodic assessment of cognitive concerns by using patient report, with a positive screen prompting evaluation of potentially treatable contributors, including mood, fatigue, sleep disturbance, pain, medications, and comorbidities [62]. In the present cohort, anxiety and fatigue were most consistently associated with impairment, making them potentially high‐yield targets for initial evaluation. Furthermore, in a large HNC cohort, treatment of anxiety and depressive disorders was associated with reduced mortality, underscoring the broader benefit of recognizing and treating psychosocial symptoms beyond cognition [63].

According to the NCCN guidelines, first‐line interventions for CRCI include neuropsychological evaluation, cognitive rehabilitation, psychotherapy, and physical activity [62]. Given well‐documented barriers to formal neuropsychological evaluation among cancer survivors [64, 65], a pragmatic approach may prioritize more accessible components, such as cognitive rehabilitation through occupational or speech therapy, alongside education and self‐management programs, while developing tiered pathways to formal testing for survivors with functionally significant or persistent impairment.

These considerations are particularly relevant in HNC, where cognitive concerns co‐occur with speech and swallowing difficulties. Interdisciplinary HNC survivorship models with embedded rehabilitation specialists in routine follow‐up have substantially improved identification of functional impairments, increasing referral to supportive services from 3% to 63% in one implementation [66]. Although cognitive function was not among the domains screened, extending existing multidomain assessments to include a brief cognitive measure may provide a feasible starting point.

5. Limitations

Several limitations must be considered in light of these findings. The cross‐sectional design precludes inference about directionality among cognitive, psychosocial, and clinical variables. Cognitive function was assessed by self‐report alone, although subjective and objective cognition measures are only modestly correlated. Therefore, PROMIS scores reflect survivors' experience of difficulty rather than measured performance. In addition, prevalence estimates varied substantially when different thresholds were applied, and no proposed cut‐off has been validated against objective performance in HNC survivors.

Detailed treatment data were unavailable, limiting evaluation of individual exposures, sequencing, and dose. The high‐intensity variable reflects aggregate burden and remains susceptible to confounding. The relatively small number of survivors meeting impairment criteria also limited statistical power for threshold‐specific analyses. Cancer stage was unavailable for 15.4% of participants, and other potential contributors to cognitive function, such as specific medications and baseline cognitive status, were not fully captured.

The 31% response rate and resulting sample composition may have introduced selection bias. Non‐respondents were more likely to be non‐White, Hispanic, and current smokers, whereas the analytic sample was predominantly White, married, and higher income. These groups are underrepresented in survivorship research and may experience greater symptom burden, and the observed prevalence may underestimate the burden in more socioeconomically and racially diverse HNC populations.

Despite these limitations, strengths include a relatively large, registry‐based sample spanning multiple anatomic subsites and contemporary treatment combinations, and concurrent assessment of HNC‐specific symptoms, psychosocial factors, and social participation using validated instruments.

6. Conclusion

Approximately one in eight HNC survivors reported moderate‐to‐severe cognitive impairment. Anxiety and fatigue were associated with both lower cognitive function scores and greater odds of impairment, whereas sleep disturbance and social isolation were associated with lower scores. These findings are consistent with emerging models of CRCI that conceptualize cognitive impairment as a component of a broader psychoneurological symptom cluster. In HNC survivors, this symptom experience occurs alongside speech, swallowing, and communication challenges that are largely absent from the populations on which most CRCI research is based.

Systematic assessment of cognitive concerns within multidisciplinary survivorship visits may help validate survivors' experiences, identify potentially modifiable co‐occurring symptoms, and determine when formal evaluation is warranted. Longitudinal studies combining patient reported and objective cognitive measures with inflammatory biomarkers and detailed treatment data are needed to further clarify mechanisms and refine impairment thresholds. Linking these measures to functional outcomes, such as communicative participation and social functioning, will inform more tailored survivorship care for the growing population of HNC survivors.

Author Contributions

Matthew E. Lin: formal analysis, writing – original draft, writing – review and editing. Jake Stenzel: writing – original draft, writing – review and editing. Desiree Delavary: writing – original draft, writing – review and editing. Maie A. St. John: writing – review and editing, supervision. Deborah J. Wong: writing – review and editing, supervision. Patricia A. Ganz: conceptualization, writing – review and editing, funding acquisition, project administration, methodology. Eden R. Brauer: conceptualization, funding acquisition, writing – original draft, writing – review and editing, project administration, formal analysis, methodology.

Funding

This work was supported by Jonsson Comprehensive Cancer Center. National Cancer Institute, K08CA252615.

Disclosure

Self‐reported cognitive impairment affects approximately one in eight head and neck cancer survivors. Anxiety and fatigue were independently associated with both poorer cognitive function and greater odds of impairment, while sleep disturbance and social isolation were associated with lower cognitive function scores. These findings support routine cognitive screening in head and neck cancer survivorship care and highlight psychosocial symptom management as a potential intervention target.

Ethics Statement

This study was reviewed and approved by the University of California, Los Angeles Institutional Review Board (IRB‐19‐0798) under expedited review.

Consent

Written consent was waived by the UCLA IRB. Completion and submission of the survey was considered consent to participate.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Sensitivity Analysis: Primary Linear Regression Model with Inclusion of Depression T‐Score¹.

Table S2: Sensitivity Analysis: Primary Linear Regression Model with Inclusion of HNC‐Specific Symptom Burden².

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.