Introducing FCR6–Brain: Measuring fear of cancer recurrence in brain tumor patients and their caregivers

Abstract

Background

Fear of cancer recurrence (FCR) is a psychological consequence of cancer diagnosis that impacts quality of life in neuro-oncology. However, the instruments used to assess FCR have not been tested for validity in patients with brain tumors. The present study explored the psychometric properties of a brief FCR scale in patients with primary brain tumor (PBT) and their caregivers.

Methods

Adult patients with PBT (n = 165) and their caregivers (n = 117) completed the FCR–7-item scale (FCR7) and measures of psychological functioning. Exploratory factor analyses (EFA) were conducted for both patient and caregiver FCR7. Convergent validity, prevalence, the difference between FCR in patients and caregivers, and relationships with relevant medical and demographic variables were explored.

Results

EFAs revealed a single factor with one item demonstrating poor loading for both patients and caregivers. Removal of the item measuring hypervigilance symptoms (checking for physical signs of tumor) greatly improved the single factor metrics. The amended scale (FCR6-Brain) demonstrated good convergent validity. Caregiver FCR was significantly higher than patient. Clinical guidance to identify clinically significant FCR was introduced. Age, gender, and time since diagnosis were related to FCR, with higher FCR in younger women more recently diagnosed.

Conclusions

The FCR6-Brain is the first validated instrument to assess FCR in this population and should be used to identify individuals at risk for FCR and guide development of future psychotherapeutic interventions. This study highlights the distinct characteristics of FCR in neuro-oncology. Symptoms of hypervigilance in PBT patients need further investigation.

A cancer diagnosis can prompt a variety of distress responses across the treatment trajectory. Individuals with cancer often meet criteria for diagnosable psychiatric disorders, such as depression and anxiety¹; others may experience a constellation of symptoms that are unique to the cancer experience and do not meet the criteria for a clinical diagnosis. This includes existential fears, such as death-related distress and fear of cancer recurrence.² Fear of cancer recurrence (FCR) refers to a type of cancer-specific distress comprised of fear, worry, or concern that one’s cancer may return or progress.³ Theoretically, FCR includes four key components: (1) high levels of preoccupation, (2) high levels of worry, (3) both of which are persistent, and (4) hypervigilance to bodily symptoms.⁴

Approximately 50% of all cancer patients experience some level of FCR, though rates of actual recurrence can vary greatly across cancer types.^5,6 FCR, in this way, may be conceptualized on a continuum: while low levels of FCR are normative—especially around the time of scans⁷—excessive FCR can perpetuate dysfunctional behavior and significantly impair patient quality of life. Indeed, previous research has linked FCR to increased psychological distress as well as increased healthcare utilization.^8–10 Given the potential consequences of FCR and the fact that many patients identify FCR as a “major concern,” ⁶ more attention to this construct is warranted, including how FCR may manifest in oncology populations wherein the odds of recurrence are particularly high or certain.

Despite the foundational literature presented about FCR in various oncology populations, very little is known about FCR in neuro-oncology. A recent systematic review found that patients diagnosed with primary brain tumor (PBT) were significantly underrepresented in FCR research, with only 0.18% of all participants classified as having a brain tumor.² This is an almost eightfold underrepresentation, given that brain cancer comprises 1.4% of all diagnosed cancer types. This review identified six studies that included patients with PBT within their mixed cancer samples—preventing specific conclusions regarding FCR in PBT from being drawn.² Thus, a focused analysis of FCR in neuro-oncology is warranted given the extent of underrepresentation alone.

FCR may also manifest differently in patients with PBT, warranting further investigation. First, patients diagnosed with PBT have a higher certainty and frequency of tumor recurrence/progression, given the current lack of noncurative treatment options.^11,12 This, of course, translates to poorer prognostic factors: for example, glioblastoma, the most common type of malignant brain tumor, has a median overall survival rate of only 15 months¹¹ and even shorter progression-free survival with a range of 7–10 months.^13,14 Additionally, brain tumor patients often undergo aggressive and prolonged treatment regimens with more regular scans, which may exacerbate FCR.⁷ The location of cancer in the brain might also impact the experience of FCR: whereas FCR in other cancers may be accompanied by active self-monitoring behaviors (eg, checking breasts for new lumps or skin for new lesions),⁴ the indicators of a recurring brain tumor are often either diffuse—occurring in the form of seizures as well as cognitive and physical decline—or completely asymptomatic, without noticeable clinical manifestations. Nevertheless, the current theoretical understanding of FCR and the scales that measure it include items specific to “checking” behaviors and hypervigilance to physical symptoms, which are likely less relevant in PBT. Instead, PBT patients may be more likely to monitor neurological and cognitive signs, symptoms not typical of other cancers, and thus not included in the currently available scales of FCR.

Preliminary research suggests that FCR is not an experience limited to the patient; caregivers, who are often intricately involved in patient care, also report experiencing FCR.^6,15–17 In fact, studies suggest that caregivers endorse higher FCR than patients.^6,18 Perhaps not surprisingly, there is even less research that examines FCR in caregivers of patients diagnosed with PBT (only one study to date, from our group).¹⁸ The results of our previous study on FCR in PBT suggest the following: First, compared to other cancer populations, patients with PBT demonstrate a similar or higher prevalence of FCR. Second, their respective caregivers endorse an even higher prevalence of FCR. Last, there exists an interdependence between patients’ and caregivers’ FCR, as well as psychological distress more broadly. Therefore, FCR is pertinent to the psychosocial care of both patients and caregivers experiencing PBT, necessitating increased empirical attention. Yet, previous research in all cancer types has failed to validate a measure for use in caregivers; instead, these studies rely on modified scales (minor changes in item wording) or nonvalidated, single-item questions to assess FCR in caregivers.^16,19,20 It is critical that research establish the validity of an FCR measure designed for caregivers to open the door for further investigations of FCR in cancer caregivers of any diagnosis.

Several self-report measures of FCR have been developed in this burgeoning area of research.²¹ The only study to date of FCR in PBT¹⁸ used The Fear of Cancer Recurrence-7 (FCR7) and found adequate internal consistency in both patients (α = 0.89) and caregivers (α = 0.91). However, the full psychometric structure of this FCR measure, beyond internal consistency, was not examined. Moreover, across all cancer types, no measure of FCR has been explicitly validated for use in caregivers. Thus, the overarching aim of the present paper was to build upon our previous work and evaluate the psychometric properties of the Fear of Cancer Recurrence-7 (FCR7)²² in a sample of patients diagnosed with PBT and their respective caregivers. The specific aims of the present paper were as follows: (1) To determine an appropriate factor structure of the FCR7 in patients with PBT and their caregivers, (2) If necessary, to modify the FCR7 as indicated by factor analysis and examine the internal consistency and convergent validity of the resultant FCR measure, (3) To provide percentiles of FCR severity in patients and their caregivers for comparative guidance in future screening, and (4) To explore relationships with and differences between the modified patient FCR measure derived from factor analysis and caregiver FCR, medical variables, and demographics.

We hypothesized that exploratory factor analysis (EFA) would reveal a slightly different structure of the FCR7 in patients and caregivers affected by PBT, with a better model fit following the removal of the item assessing physical self-monitoring for recurrence (ie, item 6: “I examine myself to see if I have physical signs of cancer”). Moreover, we hypothesized that the adjusted FCR measure with item 6 removed would demonstrate adequate internal consistency and convergent validity with other measures of psychological distress (defined as symptoms of depression, anxiety, and death anxiety). We made no hypotheses regarding percentiles, as this aim was exploratory. Regarding relationships with and differences between patient FCR and caregiver FCR, we hypothesized that caregiver FCR would remain significantly higher than patient FCR with the final EFA-derived scale, per previous work in this area.¹⁸ We avoided hypotheses concerning the relationship of FCR with any medical and demographic variables due to the novelty of this measure and uniqueness of this cancer population in terms of prognosis, progression-free survival and treatment.

Methods

Participants and Procedure

In a cross-sectional design, patients with PBT and their caregivers were asked to complete surveys including measures of FCR and psychological functioning. Data were originally collected in-person during routine neuro-oncology clinic visits at an urban National Cancer Institute-designated cancer center (July 2018–March 2020). Patients and their caregivers were recruited, screened, and consented by trained research assistants. Participants completed surveys via iPad or paper in designated treatment rooms. Following the establishment of COVID-19 safety regulations, recruitment was paused (April 2020–June 2020). The second wave of data collection was converted to a virtual format (July 2020–February 2021), and participants were recruited from neuro-oncology social media support groups. A recruitment flyer was posted to these online platforms with a link to the screening questions, consent form, and study surveys. Screening questions were completed online via a confidential database (REDCap)^23,24 to determine eligibility. If found eligible, a full consent document followed. This document was signed electronically, with a required (Yes/No) indication of understanding the research purpose, the risks/benefits to participating, and the voluntariness of participation. Once signed, consenting participants were routed to the questionnaires electronically via REDCap. Inclusion criteria were as follows for patients: (1) age 18 or older, (2) PBT diagnosis, and (3) literate in English. To be included as a caregiver, the individual had to define themself as a family member, friend, or assigned caretaker to a patient with a PBT. Participants did not have to be a part of a dyad to be included; that is, patients and caregivers could consent to participate as individuals. Exclusion criteria for both patients and their caregivers included an inability to read or understand the English language or to provide informed consent.

One hundred forty-one patients were approached in-person during the study timeline, of which 25 patients declined participation (18%) and two were unable to provide consent (1%). In-person decline justification (if provided) included fatigue (n = 5; 20%), time constraints (n = 5; 20%), or topic disinterest (n = 7; 28%). Ninety in-person caregivers consented to participate. One hundred and forty-four patients and 71 caregivers selected the associated link for virtual participation. Seventy-two did not provide consent. Declined participation justification was not assessed during virtual recruitment. In total, 65 participants (42 patients: 16 in-person and 26 virtual; and 23 caregivers: nine in-person and 14 virtual) provided incomplete data. Thus, a total of 165 PBT patients and 117 caregivers were included in the final samples (there was overlap with the sample from our previous study of 80 PBT patients and 52 dyads; however, the present sample consisted of more participants resulting from further recruitment). A total of 68 dyads were available for all dyadic analyses. Please see Supplemental Figure for a detailed flow diagram. The present study was approved by the relevant internal review board (HM20013477). For in-person participants, demographic (eg, age, gender) and medical variables (eg, time since diagnosis, tumor type and grade, location of tumor) were extracted from medical records following informed consent. For virtual participants, demographic and medical variables were self-reported.

Measures

The following measures were selected to examine convergent validity as the relationship between depression, anxiety, and death distress with fear of cancer recurrence has been reliably demonstrated in the literature for both patients and caregivers.^18,25

Fear of Cancer Recurrence –7-item (FCR7) was used to assess FCR. It consists of six items rated on a 5-point Likert scale from “not at all” to “all the time.” ²² The final item is rated on a 10-point Likert scale from “not at all” to “all the time” making the range of possible scores 6–40. The FCR7 provides a total composite score, and previous literature provides comparison scores of ≥17 as higher than 60% of a mixed sample of cancer patients and ≥27 as higher than 90% of a mixed sample of cancer patients.²² Caregiver FCR7 questions were slightly altered to capture caregiver’s fear of their loved one’s tumor recurrence. For example, the patient-version “I am afraid that my cancer may recur” was changed to “I am afraid that my loved one’s cancer may recur.”

Psychological distress was defined as symptoms of depression, anxiety, and death anxiety as these clusters of symptoms. Patient Health Questionnaire – 9-item (PHQ-9) was used to assess depressive symptoms. It consists of nine items, each rated on a 4-point Likert scale from “not at all” to “nearly every day.” ²⁶ It produces a total score, and relevant cutoffs have been established in the literature for interpretation. Scores < 5 are considered minimal or subthreshold; scores from 5–9 are mild; 10–14 moderate; 15–19 moderately severe; and scores ≥20 are considered severe.²⁶ Patients and caregivers reported on their own depressive symptoms. The PHQ-9 is widely used and cross-validated in patient and healthy samples.²⁷ It demonstrated good internal reliability in the present sample (α = 0.87–0.88).

Generalized Anxiety Disorder –7-item (GAD-7) was used to assess generalized anxiety symptoms. It consists of seven items, on a 4-point Likert scale from “not at all” to “nearly every day.” ²⁸ The GAD-7 has established cutoffs for interpretation of composite scores such that scores <5 are considered minimal or subthreshold; scores from 5–9 are mild; 10–14 moderate; and ≥15 severe.²⁸ Patients and caregivers reported on their own generalized anxiety. The GAD-7 is cross-validated in patient and healthy samples²⁹ and demonstrated good internal reliability in the present sample (α = 0.92–0.94).

Death Distress Scale (DDS) was used to assess death-distress. It consists of 24 items, derived from three previously validated questionnaires: the Death Anxiety Scale,³⁰ Death Depression Scale,³¹ and Death Obsession Scale.³² The DDS retains this three-factor structure, with each domain containing eight statements that can be rated on a five-point scale ranging from 1 (no) to 5 (very much).³³ The score for each factor is based on the sum of all items within each domain yielding a death anxiety, death depression, and death obsession score. The following guidelines have been proposed to interpret the factor scores: 8–18 mild, 19–29 moderate, and 30–40 severe.³³ A total score can also be calculated to assess global existential distress (DDS-total), with a higher score indicative of greater distress; this was used in the current study. The DDS demonstrated good internal reliability in this sample (α = 0.93–0.95).

Demographic and Medical Characteristics were collected from patient medical records for in-person data and were patient- or caregiver-reported for virtual data. Time since diagnosis was established as date of survey completion minus date of diagnosis and then rounded to the nearest month. Tumor grade was investigated dichotomously with tumor grades I & II defined as low grade and grades III & IV defined as high grade. Location of tumor was investigated categorically with right hemisphere, left hemisphere, and bilateral as possible groups.

Data Analytic Plan

SPSS version 27 was used for all analyses. Study data were collected and managed using REDCap electronic data capture tools hosted at Virginia Commonwealth University.^23,24

Exploratory Factor Analyses (EFA), assuming no a priori factor structure, were performed using principal axis factoring and Promax rotations, including all seven items for the 165 patient responses to the FCR and separately for all 117 caregiver responses to the caregiver FCR. Factor fit was analyzed using scree plots,³⁴ cumulative variance,³⁵ and eigenvalues.

Internal reliability was determined via Cronbach’s alpha. Good internal reliability was defined as Cronbach’s alpha >.70.³⁶ Pearson’s correlations between the final FCR scale (determined via EFA) and measures of depressive symptoms, generalized anxiety, and death anxiety established convergent validity. Convergent validity was considered adequate if the correlation between measures was >.50.³⁷

To provide comparative guidance for the field on FCR severity in patients with PBT and their caregivers, 60th and 90th percentiles were reported.²² As a preliminary guide for interpreting the amended FCR scale, we recommend interpreting scores at or above the 90th percentile as “clinically significant” based on previous scale development research demonstrating adequate sensitivity and specificity of the 90th percentile in identifying the presence of a disorder or clinically relevant symptom.^38,39 Furthermore, we recommend interpreting scores at or above the 60th percentile as “clinically subthreshold” based on the understanding that those who report higher symptom burden than 60% of the sample should be screened further and made aware of supportive services.⁴⁰

The following analyses were carried out using the final FCR measure determined to be superior via EFA in patients with PBT and their caregivers. First, we investigated correlations and differences between patient and caregiver FCR using a paired-samples t-test. Then, separate hierarchical regressions explored relationships between FCR and age, gender, tumor grade, tumor location (left vs right; bilateral removed due to insufficient power), and time since diagnosis. Collinearity diagnostics were examined in the relevant regression analyses and based on established cutoffs (VIF > 10) were not found to be a concern (eg, age and tumor grade were not multicollinear).

Results

Full demographic and medical information for both patients and caregivers are provided in Table 1. Information is separated by data collection method (in-person/virtual).

Table 1.

Demographic and Medical Information for Patients and Caregivers by Data Collection Method

Demographic	Patients					Caregivers
	In-Person (n = 98)		Virtual (n = 67)		P	In-Person (n = 81)		Virtual (n = 36)		P
	Frequency	Mean (SD)/Percent	Frequency	Mean (SD)/Percent		Frequency	Mean (SD)/Percent	Frequency	Mean (SD)/Percent
Age	98	49.5 (15.1)	67	45.3 (11.1)	.035	81	54.5 (12.4)	36	49.1 (12.9)	.036
Gender
Male	44	44.9	25	37.3		29	23.6	6	6.5
Female	54	55.1	41	61.2	n.s.	53	43.1	30	32.6	.041
Transgender/Binary	0	0.0	1	1.5		0	0	0	0
Race
White	79	80.6	62	92.5		66	53.7	34	37.0
Black	17	17.3	1	1.5		13	10.6	1	1.1
Asian	0	0.0	2	2.9	.001	0	0	0	0	.041
Latinx	0	0.0	1	1.5		0	0	0	0
Other	2	2.0	1	1.5		1	0.8	1	1.1
Relationship to Patient
Spouse/ Partner	–	–	–	–		50	40.7	23	25.0
Family Member	–	–	–	–	–	29	23.6	12	13.0	n.s.
Friend	–	–	–	–		2	1.6	1	1.1
Time since Diagnosis	98	58.3 (67.6)	67	71.1 (72.5)	n.s.	79	53.8 (67.9)	36	47.31 (53.39)	n.s.
Tumor Type
Glioblastoma	31	31.6	14	20.9		31	37.8	20	55.6
Astrocytoma	18	18.4	17	25.4		14	17.1	5	13.9
Oligodendroglioma	22	22.5	13	19.4	n.s.	14	17.1	6	16.7	n.s.
Meningioma	12	12.3	12	17.9		9	11.0	2	5.6
Other	15	15.3	10	14.9		11	13.4	3	8.3
Tumor Grade
Low	48	49.0	24	35.8	.023	33	40.2	5	13.9	.006
High	50	51.0	32	47.8		45	54.9	28	77.8
Tumor Hemisphere
Left	50	51.0	24	35.8		42	51.2	17	47.2
Right	37	37.8	32	47.8	n.s.	29	35.4	16	44.4	n.s.
Bilateral	10	10.2	11	16.4		8	9.8	3	8.3
Treatment
Resection	74	75.5	54	80.6	n.s.	62	75.6	31	86.1	n.s.
Radiation	80	81.6	48	71.6	n.s.	67	81.7	32	88.9	n.s.
Chemotherapy	75	76.5	45	67.2	n.s.	66	80.5	31	86.1	n.s.
Psychological Functioning
GAD-7	96	5.1 (5.6)	63	8.8 (6.6)	<.001	80	5.9 (5.6)	35	10.4 (5.9)	<.001
PHQ-9	97	6.4 (6.0)	66	10.9 (7.1)	<.001	80	4.5 (4.9)	33	9.8 (6.1)	<.001
DDS	91	45.9 (15.9)	65	50.2 (16.5)	n.s.	72	47.1 (14.3)	30	70.5 (20.3)	<.001
FCR6-Brain	98	16.0 (8.2)	67	19.9 (8.1)	.003	81	19.7 (8.0)	26	24.0 (7.3)	.006

Note. SD, Standard Deviation; GAD-7, Generalized Anxiety Disorder – 7 Item; PHQ-9, Patient Health Questionnaire – 9 item; DDS, Death Distress Scale; FCR6-Brain, Fear of Cancer Recurrence – 6 item Brain.

Significant differences were found between data collection methods on the GAD-7 (P < .001), PHQ-9 (P < .001), and the final FCR measure found to be superior in factor analysis (P = .003), such that distress was higher in the virtual group. Age was significantly lower in the virtual group (P = .035). For caregivers, a similar pattern was identified—significant differences were found for GAD-7 (P < .001), PHQ-9 (P < .001), DDS (P < .001), and the FCR measure (P = .006), such that distress was higher in the virtual group. Caregiver age was also significantly lower in the virtual group (P = .036). Finally, there were significantly more female caregivers in the virtual group than in-person (P = .041).

Exploratory Factor Analysis

For patient FCR, all items were significantly intercorrelated (P < .001) with correlation coefficients ranging from .6 to .9. Although visual inspection of the scree plot³⁴ supported a two-factor model, only the one-factor model achieved an eigenvalue above 1.0 (5.3), and the one-factor model accounted for 75.0% of the variance. Six items correlated with the factor between .8 and .9, and the seventh item correlated with the single factor at .6 (“I examine myself to see if I have physical signs of the tumor”). Item-total correlations are presented in Table 2.

Table 2.

Patient FCR Item-Total Correlations for One-Factor Model (n = 165)

Item		Seven-Item Model	Six-Item Model
1.	“I am afraid that my tumor may reoccur”	.853	.864
2.	“I am worried or anxious about the possibility of tumor recurrence.”	.894	.903
3.	“How often have you worried about the possibility of getting a tumor again?”	.917	.928
4.	“I get waves of strong feelings about the tumor coming back.”	.892	.888
5.	“I think about the tumor returning when I didn’t mean to.”	.876	.861
6.	“I examine myself to see if I have physical signs of the tumor.”	.622	–
7.	“To what extent does worry about getting a tumor again spill over or intrude on your thoughts and activities”	.818	.806

Note. FCR, Fear of Cancer Recurrence.

Due to the weaker relationship between item six and the single factor, the EFA was re-run without item six. For patient FCR with item six removed, all items remained significantly intercorrelated (P < .001) and correlated between .7 and .9. Again, although visual inspection of the scree plot³² supported a two-factor model, only the one-factor model achieved an eigenvalue above 1.0 (4.8), and the one-factor model accounted for 80.5% of the variance. All six items correlated with the factor above .8, with items two and three correlating with the unitary factor above .9. Item-total correlations are presented in Table 2. The Cronbach’s Alpha was .91 for the 6-item measure.

An EFA was also run for caregiver FCR, with all items intercorrelated in both a six- and seven-item model (P < .001) with correlation coefficients ranging from .6 to .9 and from .5 to .9, respectively. Similarly, visual inspection of the scree plot³⁴ in both cases supported a two-factor model, though only the one-factor model achieved an eigenvalue above 1.0. The seven-item model achieved an eigenvalue of 5.2 and accounted for 74.1% of the variance, whereas the six-item model achieved an eigenvalue of 4.9, accounting for 80.8% of the variance. Item-total correlations are presented in Table 3 and again support a six-item model without question number six. The Cronbach’s Alpha was .90 for the 6-item measure.

Table 3.

Caregiver FCR Item-Total Correlations for One-Factor Model (n = 117)

Item		Seven-Item Model	Six-Item Model
1.	“I am afraid that my loved one’s tumor may reoccur”	.886	.891
2.	“I am worried or anxious about the possibility of my loved one’s tumor recurrence.”	.912	.912
3.	“How often have you worried about the possibility of your loved one getting a tumor again?”	.928	.927
4.	“I get waves of strong feelings about my loved one’s tumor coming back.”	.896	.901
5.	“I think about my loved one’s tumor returning when I didn’t mean to.”	.813	.803
6.	“I examine my loved one to see if they have physical signs of the tumor.”	.557	-----
7.	“To what extent does worry about your loved one getting a tumor again spill over or intrude on your thoughts and activities”	.825	.825

Note. FCR, Fear of Cancer Recurrence.

Convergent Validity

Using the sum total of the newly established FCR6—Brain Tumor scale (FCR6-Brain), patient FCR6-Brain was significantly correlated with all measures of patient-reported psychological distress, GAD-7 (r = .7, P < .001), PHQ-9 (r = .5, P < .001), and the DDS (r = .7, P < .001). Similarly, using the newly established FCR6 – Caregiver Brain Tumor scale (FCR6-CG-Brain), caregiver FCR was significantly correlated with all measures of caregiver-reported psychological distress, GAD-7 (r = .7, P < .001), PHQ-9 (r = .7, P < .001), and DDS (r = .7, P < .001).

Percentiles and Prevalence

The descriptive data on FCR6-Brain in patients with primary brain tumor and their caregivers is presented in Table 4. For reference, a study of breast and colorectal cancer patients using the original FCR 7-item (one more item than the FCR6-Brain) identified a mean score of 16.5 (SD = 7), where the 60th percentile score was 17 and the 90th percentile score was 27.²² Even with one less item, both patients and caregivers in our sample reported a higher mean than the comparison group.

Table 4.

Percentiles and Prevalence

	FCR6-Brain
	Patient	Caregiver
Range	5–35	5–35
Mean	17.6 (SD = 8.3)	21.0 (SD = 8.0)
Median	16	19
Mode	10	19
60th %	18	24
90th%	31	33

Note. FCR, Fear of Cancer Recurrence.

FCR-6 Brain Relationships and Differences

Patient and caregiver FCR6-Brain were both significantly correlated (r = .3, P = .04) and significantly different, where caregivers reported greater FCR (M = 19.2, SD = 7.9) than patients (M = 15.7, SD = 7.5), t(66) = −3.11, P = .003.

There was a significant difference found between in-person and virtual data collection of the FCR6-Brain; therefore, data collection method was entered as a covariate in all demographic and medical comparisons (see Tables 5 and 6). Younger female patients who were more recently diagnosed reported significantly higher FCR (P < .05; Table 5). For the caregiver sample, younger caregivers of patients with more recent diagnoses and right-sided tumors also reported higher FCR (P < .05; Table 6).

Table 5.

Separate Hierarchical Regressions between Patient FCR and Demographic/Medical Variables after Controlling for Data Collection

Predictor variables	B (SE)	t	Sig.	n
Data Collection	3.31 (1.28)	2.58	.011*	164
Age	−.13 (.05)	−2.78	.006*
Data Collection	3.71 (1.27)	2.91	.004*	164
Gender	−3.07 (1.23)	−2.49	.014*
Data Collection	4.58 (1.39)	3.31	.001*	153
Tumor Grade	−1.61 (1.34)	−1.20	.23
Data Collection	4.22 (1.41)	2.99	.003*	142
Tumor Location	2.10 (1.38)	1.52	.13
Data Collection	4.11 (1.28)	3.21	.002*	164
Time Since Diagnosis	−.02 (.01)	−2.27	.025*

* P < .05; B = unstandardized coefficient; SE = coefficient’s standard error; gender coded as 0 = female and 1 = male; tumor grade coded as 1 = low-grade and 2 = high-grade; tumor location coded as 0 = left hemisphere and 1 = right hemisphere.

Table 6.

Separate Hierarchical Regressions between Caregiver FCR and Demographic/Medical Variables after Controlling for Data Collection

Predictor Variables	B (SE)	t	Sig.	n
Data collection	3.47 (1.54)	2.25	.03*	115
Age	−.13 (.06)	−2.33	.02*
Data collection	4.37 (1.60)	2.74	.007*	116
Gender	.13 (1.61)	.08	.93
Data collection	4.47 (1.65)	2.71	.008*	109
Tumor grade	1.76 (1.59)	1.11	.27
Data collection	3.66 (1.62)	2.27	.03*	102
Tumor location	3.34 (1.52)	2.20	.03*
Data collection	4.40 (1.52)	2.90	.004*	113
Time since diagnosis	−.03 (.01)	−2.76	.007*

* P < .05; B = unstandardized coefficient; SE = coefficient’s standard error; gender coded as 0 = female and 1 = male; tumor grade coded as 1 = low-grade and 2 = high-grade; tumor location coded as 0 = left hemisphere and 1 = right hemisphere.

Discussion

The present study investigated the validity of the FCR-7 in patients with PBT and, using EFA, found a superior factor structure with the removal of the item pertaining to checking behaviors for physical signs of cancer (ie, item 6: “I examine myself to see if I have physical signs of cancer”). As hypothesized, with the removal of this item, the resultant FCR6-Brain demonstrated strong construct consistency and intercorrelation. In both patients and caregivers, the six-item version achieved superior validity over the seven-item scale and both demonstrated excellent internal reliability. Convergent validity was supported by significant positive relationships between the FCR6-Brain and measures of generalized anxiety, depression, and death-related distress for both patients and caregivers.

We used the FCR6-Brain to establish severity guidelines for future use as a screener in patients with PBT and their caregivers. We suggest interpretation of scores at or above the 60th percentile as “clinically subthreshold” FCR and those at or above the 90th percentile as “clinically significant”.²² Until future studies administer the FCR6-Brain alongside a structured interview of clinically significant FCR, these percentiles can guide FCR screening in patients with PBT and their caregivers. Visual comparison between the 60th and 90th percentiles on the FCR6-Brain and the FCR 7-item in a mixed-sample of breast and colorectal cancer patients²² demonstrated that FCR was higher in this sample of PBT patients, despite the removal of one item. This provides striking evidence of the severity of FCR in brain tumor patients when compared to other cancer samples.

Consistent with previous analyses of FCR in both neuro-oncology and beyond,^15,17,18 caregivers reported significantly higher levels of FCR compared to patients in our sample. This pattern of results follows other psychological determinants of health—such as depression and anxiety—in which caregivers consistently endorse greater emotional distress.⁴¹ Caregiver and patient FCR were also correlated in the current study. In the only previous investigation of caregivers in neuro-oncology,¹⁸ there was a significant partner effect between caregiver emotional distress and patient FCR, highlighting the dyadic stress model. The correlations between caregiver and patient distress should be kept in mind when designing interventions that target FCR—caregivers are an important component of patient well-being and should not be overlooked.

When examining age, we found that FCR was highest in younger patients. This is consistent with previous findings of negative relationships between age and FCR in other cancer groups.⁴² While the attenuating effect of FCR with age requires further investigation, a recent review outlines the possibility of health expectation at a younger age and the increased burden of unrealized years when diagnosed in youth as primary sources of distress.⁴³ Although the current study did not find this same risk factor in caregivers, previous research outside of neuro-oncology suggests there is significant partner effect between age and FCR, such that levels of FCR in one partner are higher when the other is younger.¹⁶ A future investigation of the actor–partner interdependence model using the new FCR6-Brain will be needed to understand these results in neuro-oncology.

In addition, when examining the effect of gender on patient FCR, women endorsed higher levels of FCR, consistent with findings from a recent meta-analysis which included over 30,000 oncology patients.⁴⁴ Researchers posit this discrepancy as multifactorial and related to well-documented gender differences in mental and psychological conditions, social inequities impacting women, and women’s tendency to both express and seek support for their problems. Conversely, this pattern was not observed among caregivers, with no significant difference in caregiver FCR by gender. This may be due in part to the small sample of men (n = 35), though it appears consistent with other nonsignificant findings between gender and FCR in oncology caregivers.⁴⁵

Both patient and caregiver FCR was elevated in those closer to diagnosis. This extends previous work in neuro-oncology using the FCR7¹⁸ and likely reflects a recency effect in which the diagnosis acts as a traumatic event, increasing worry, fear, and hypervigilance. Over time, patients and their caregivers may develop more acceptance of the realities of the diagnosis, especially in the case of recurrence and advancing disease. Alternatively, patients and their loved ones diagnosed initially with low-grade brain tumors may become hopeful following months and even years of progression-free survival and expect positive scans, thus decreasing worry and fear. Future longitudinal investigations of FCR in patients with brain tumors will be necessary to better understand how the disease and treatment trajectory affect these fears, and whether it is different for low- and high-grade patients.

When examining tumor location, it is well documented that the right hemisphere of the brain is associated with processing negative emotions, fear, and stress.^46–48 Patients with white matter lesions, hyperactivity, and/or stroke of the right hemisphere experience increased affective disorders such as anxiety and depression.^46,49,50 Interestingly, caregivers of patients with right-sided tumors also reported higher FCR, suggesting the possible implications of patient emotional well-being in caregiver distress. It may be that patients biologically predisposed to experience more negative emotion (ie, those with right-sided tumors) influence the caregiver response; such that the emotional distress is contagious, affecting the FCR of their caregivers as well. It is important to qualify these findings as preliminary and emphasize the need for replication in larger samples. Our findings demonstrate the possibility of this relationship and warrant future investigation of the role of tumor-hemisphere in FCR.

Study Limitations and Future Directions

This study is not without limitations. First, data collection methods were unavoidably inconsistent throughout the study, attributable to the restrictions imposed by the COVID-19 pandemic. This yielded observed sample differences. For example, patients and caregivers who completed the study online during the pandemic were not only younger but also reported greater emotional distress than those who completed the study in-person. While the change in data collection methods was not ideal, recruiting patients and caregivers both in-person and online allowed us to cast a wider net, potentially capturing more patients in need of support (eg, those from support groups, listservs, and social media platforms). Additionally, a downside of online recruitment included relying on self-reported medical variables. However, on a larger scale, this variation contributes to the generalizability of our results to the greater population of PBT within the historical context of the pandemic. On a similar note, this study’s cross-sectional data demonstrated heterogeneity, particularly in terms of time since diagnosis and type of treatment, with respect to FCR. Nevertheless, this is clinically relevant and promotes the universality of application to the general population. Longitudinal data are needed to map out possible trends in patient and caregiver FCR throughout the disease trajectory. Cutoff values to indicate clinically significant FCR have not been established in the literature and therefore could not be referenced and evaluated accordingly in this study. This remains a potential area for future work. Finally, this paper established the superior factor structure with removal of the item pertaining to checking behaviors for physical signs of cancer. Considering the key theoretical components of FCR,⁴ it may be that checking behaviors for signs of brain tumor recurrence are more subtle, for example, hypervigilance to headaches, aura, seizure symptoms, or cognitive changes. An investigation into alternate items to measure symptoms of hypervigilance in the neuro-oncology population was beyond the scope of the present study and represents a rich avenue open for future investigation.

Conclusion

The FCR6-Brain represents the first validated measure of FCR in patients with PBT. This study is also the first to validate a FCR measure for caregivers’ fear of their loved one’s cancer recurring. Benchmarks of 60th and 90th percentiles for both patients and caregivers provide guidance for identifying those who may be at risk for FCR to negatively impact quality of life, especially given strong correlations between FCR6-Brain and well-established measures of distress. The aims achieved in this study have important clinical and empirical implications. Specifically, the results of this study represent the critical first steps to understanding the longitudinal trajectory of FCR and to evaluating interventions designed to reduce FCR in this vulnerable cancer population.

Funding

(1) The project described was supported by CTSA award No. KL2TR002648 from the National Center for Advancing Translational Sciences. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the National Institutes of Health. (2) NCRR Award Number UL1TR002649.

Conflict of interest statement. The authors have no conflicts of interest to declare.

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