Validity Testing of the Head and Neck Lymphedema and Fibrosis Symptom Inventory

Abstract

Background:

Lack of reliable and valid tools significantly impacts early identification and timely treatment of lymphedema and fibrosis (LEF) in the head and neck cancer population. To address this need, we developed and reported a patient-reported outcome measure (Head and Neck Lymphedema and Fibrosis Symptom Inventory [HN-LEF SI]). This article reports the construct validity (convergent and divergent validity) testing of the tool.

Materials and Methods:

A prospective, longitudinal, instrument validation study was conducted in patients with a newly diagnosed oral cavity or oropharyngeal cancer. Participants completed the HN-LEF SI and six carefully selected self-report measures at pretreatment, end-of-treatment, and every 3 months up to 12 months after treatment. Spearman correlations were used.

Results:

A total of 117 patients completed the study. Patterns of correlations of the HN-LEF SI scores with the established self-report measure scores were consistent with expected convergent and divergent validity.

Conclusion:

Evidence from this work supports the construct validity of the HN-LEF SI.

Introduction

Recent literature has indicated that head and neck cancer (HNC) patients are at a high risk for developing lymphedema and fibrosis (LEF), which causes substantial symptom burden, functional impairments, and decreased quality of life.^1,2 Early identification in conjunction with timely and skilled lymphedema therapy is critical for mitigation of the negative impact of LEF and to slow down its progression of fibrosis.^3–5 Lack of reliable and valid measures is a significant barrier to early identification of LEF and assessment of therapeutic efficacy.^6,7 Furthermore, lack of measurement tools that are sensitive to changes in LEF-associated clinical burden over time has contributed to the sparsity of clinical trials evaluating novel treatment strategies for the management of LEF.

To fill in this gap, we developed a patient-reported outcome measure designed to capture symptom burden and functional loss related to LEF in the HNC population.^8,9 We subsequently conducted a prospective, longitudinal study (R01 DE024982) directed at optimizing measurement tools for patients with HNC-associated LEF with an emphasis on oral cavity and oropharyngeal cancer.¹⁰ Our first step was to minimize patient burden by reducing the number of component items.¹¹ That process resulted in a shortened tool referred to as the Head and Neck Lymphedema and Fibrosis Symptom Inventory (HN-LEF SI).¹¹ The HN-LEF SI contains 33 items with 7 subscales, all of which demonstrated content validity and good reliability.¹¹

In this article, we aim to present the findings of our subsequent work assessing the construct validity (convergent and divergent) of the HN-LEF SI using multiple, well-established measures of quality of life, symptom burden, and functionality. In general, we hypothesized that specific symptoms captured on the HN-LEF SI would demonstrate a positive correlation with similar symptoms captured on established measures. For example, the jaw and oral dysfunction subscale of the HN-LEF SI was expected to have a relatively high correlation with the trismus score on the Vanderbilt Head and Neck Symptom Survey (VHNSS), thus demonstrating convergent validity. Conversely, we hypothesized that correlations between dissimilar symptoms would be low. For example, the mucosal irritation subscale of the HN-LEF SI was expected to have a relatively low correlation with the neck range of motion score from the VHNSS thus demonstrating divergent validity.

Materials and Methods

Trial design

Before the initiation of the prospective, longitudinal trial (ClinicalTrials.gov identifier: NCT02412241),¹⁰ we received approval from the Institutional Review Board and Scientific Review Committee at Vanderbilt University. The trial was conducted in line with the ethical principles of the Declaration of Helsinki and the International Conference on Harmonization guidelines for Good Clinical Practice. All participants signed a written informed consent form before they engaged in any study-related activities.

Sample and settings

Eligibility for participation included the following: (1) having a newly diagnosed, histologically proven cancer arising from the oral cavity and oropharynx; (2) having tumor stage II or greater (American Joint Committee on Cancer seventh edition); (3) ≥21 years of age; (4) being willing and able to undergo study assessment; and (5) ability of speaking and reading English and understanding informed consent. Patients were excluded if they met one of the following criteria: (1) having medical record documentation of cognitive impairment that would preclude the ability to provide informed consent; (2) being unwilling to undergo routine follow-up; (3) having recurrent cancer; or (4) having any other active cancer. Patients were recruited from the Head and Neck Cancer Clinics at Vanderbilt-Ingram Cancer Center (Nashville, TN).

Data collection procedures

Participants completed a demographic survey at baseline and self-report measures at baseline, at the end of treatment, and every 3 months thereafter for up to 12 months post-HNC treatment. The battery of six self-reported measures included the HN-LEF SI. Study staff performed a head and neck examination that included skin and soft tissues. External manifestations of LEF were assessed and documented using the HN-LEF Assessment Criteria.^12,13 Internal lymphedema was evaluated by an endoscopic examination and documented using the Modified Patterson Scale.¹⁴ Tumor and treatment-related data were obtained by study staff through electronic medical record review. All survey responses and extracted medical chart review data were entered into a Research Electronic Data Capture (REDCap) database.^15,16

Measures

Only the measures that are directly related to this report are included.

Demographic form

The demographic form captures basic patient characteristics such as age, gender, ethnicity, and highest education level.

Medical information form

The medical information form included primary tumor site (oral cavity or oropharynx), tumor stage, histology, and cancer treatment-related information.

Head and Neck Lymphedema and Fibrosis Symptom Inventory

This tool contains 33 items (symptoms), which are answered in a “yes” or “no” format. If patients answer “yes” to any given item, they are then asked to rate the intensity of the respective symptom on a five-point scales. Each item then receives a score ranging from “0” (Does not experience the symptom) to “5” (Experiences the symptom with severe intensity). Previous work demonstrated that the symptoms fall into one of seven subscales: (1) soft tissues and neurologic toxicity, (2) systemic symptoms and social functioning, (3) jaw and oral dysfunction, (4) swallowing and taste changes, (5) body image and sexuality, (6) communication, and (7) mucosal irritation.¹¹

Two scores may be generated for each subscale: the average intensity rating for all symptoms in the subscale and the maximum intensity rating for any single symptom with the subscale. All seven subscales with the associated items and Cronbach's alpha reliability coefficients from that previous work are given in Table 1.

Table 1.

Head and Neck Lymphedema and Fibrosis Symptom Inventory

Subscale	Reliability excluding baseline (Cronbach's alpha)
Soft tissue and neurologic toxicity	0.86 (12 months) to 0.90 (3 months)
Feeling uncomfortable head and neck region
Tightness
Firmness or hardness of skin
Stiffness
Tingling
Pain on movement of the head and neck
Numbness
Tenderness
Limited head and neck movement
Change in skin texture
Systemic symptoms and social functioning	0.69 (3 months) to 0.78 (12 months)
Feeling anxious
Feeling tired
Problems sleeping
Decreased social activity
Inability complete hobbies
Jaw and oral dysfunction	0.59 (6 months) to 0.76 (12 months)
Problems chewing
Hard to open mouth
Hard to move tongue
Swallowing and taste changes	0.69 (3 months) to 0.80 (EOT, 12 months)
Problems swallowing solids
Problems swallowing soft solids
Something stuck in throat
Taste change
Body image and sexuality	0.75 (EOT) to 0.90 (9 months)
Concerns about looks
Feeling unattractive
Decrease sexual activity
Feeling less sexually attractive
Lack interest in sex
Communication	0.71 (6 months) to 0.80 (EOT)
Problems talking
Voice changes
Hard to be understood
Mucosal irritation	0.49 (12 months) to 0.73 (EOT)
Excess mucous production
Hoarseness
Sore throat

EOT, end of treatment.

European Organization for Research and Treatment of Cancer Quality-of-Life Questionnaire (EORTC-QLQ-C30) global health status

This is a validated 30-item cancer-specific questionnaire that includes a global health-related quality-of-life scale, 5 functional subscales (physical, role, emotional, cognitive, and social), 3 symptom scales (e.g., fatigue), and 6 single items (e.g., loss of appetite). Each item has four response alternatives on a four-point scales (i.e., “not at all,” “a little,” “quite a bit,” and “very much”), except the two items of the global health status scale, which have response options ranging from (1) “very poor” to (7) “excellent.”

Scores for all scales and single items are linearly converted to range from 0 to 100. High scores of functional scales characterize healthy functioning. Similarly, a high score for global health status signifies a higher quality of life. The tool has been extensively used in cancer (e.g., HNC) patients.¹⁷ Cronbach's alpha reliability coefficients for the global and five subscales in this study ranged from 0.70 to 0.95.

Short-Form McGill Pain Questionnaire

This is a widely used and validated assessment tool for measuring pain.¹⁸ It contains 15 words that describe the sensory (11 words) and affective (4 words) dimensions of pain. These words are rated on an intensity scale as 0 = none, 1 = mild, 2 = moderate, and 3 = severe. The sensory score ranges from 0 to 33, and the affective score ranges from 0 to 12. In addition, there are two measurements of pain intensity: a present pain intensity scale (numerical rating scale from 0 [no pain] to 5 [excruciating]) and a 10 cm visual analogue scale (score from 0 to 100). The tool has been used in lymphedema populations to evaluate pain or discomfort.^19,20 In this study, the Cronbach's alpha coefficient for sensory and affective Short-Form McGill Pain Questionnaire (SF-MPQ) scores ranged from 0.75 to 0.90.

Neck Disability Index

The Neck Disability Index (NDI) (10-item) is a validated tool that measures pain severity and its impact on daily functions personal care, lifting, reaching, headache, concentration, work, driving, sleeping, and recreation.²¹ The tool uses a Gutman-style response format with six levels of increasing disability degrees. Each of the 10 items is scored from 0 to 5. The maximum score is 50 (0–4 = no disability; 5–14 = mild disability; 15–24 = moderate disability; 25–34 = severe disability; above 34 = complete disability). The internal consistency coefficients for the scores in this study ranged from 0.81 to 0.86 (Cronbach's alpha).

VHNSS version 2.0 and General Symptom Survey

VHNSS version 2.0 (50-item) is a validated tool to measure physical symptom burden and functional deficits related to HNC and its treatment. It includes 10 domains and 3 single items: mouth pain (6 items), general pain (3 items), swallowing solids (8 items), swallowing liquids (2 items), nutrition (4 items), mucus (4 items), dry mouth (5 items), taste/smell (6 items), voice/communication (3 items), teeth (4 items), hearing (1 item), trismus (1 item), and neck range of motion (1 item).^22,23

The General Symptom Survey (GSS) includes 11 additional items and this subscale was specifically developed to assess the systemic symptoms associated with HNC and its therapy.²⁴ Each item is scored from 0 (none) to 10 (a lot). Mean scores on each VHNSS subscale can range from 0 (none) to 10 (a lot). Studies have reported evidence for validity with both patient-reported and objective outcome measures.^22–24 The Cronbach's alpha reliability coefficients for the scores in this study ranged from 0.75 to 0.90.

Hospital Anxiety and Depression Scale

The 14-item Hospital Anxiety and Depression Scale (HADS) is a validated tool for measuring symptoms of anxiety and depression among nonpsychiatric populations.²⁵ The HADS has seven items for depression subscale and seven items for anxiety subscale. Scoring for each item ranges from 0 to 3, with three denoting highest anxiety or depression level. The maximum score of each subscale is 21 (0–7 = normal; 8–10 = mild; 11–14 = moderate; 15–21 = severe). Scores for the HADS-anxiety, HADS-depression, and HADS-total scales in this study demonstrated reliability coefficients ranging from 0.80 to 0.90.

Marlowe–Crowne Social Desirability Scale—Short Form C

The 13-item Marlowe–Crowne Social Desirability Scale—Short Form C (MC-C) has been demonstrated to be a valid and reliable measure of socially desirability in diverse settings and populations.²⁶ Participants respond to each item by indicating whether it is true or false. The total score of the MC-C ranges from 0 to 13. High scores indicate that a respondent is presenting him/herself in an unrealistically favorable manner. The Cronbach's alpha coefficients for the scores in this study ranged from 0.60 to 0.70.

Statistical analysis

Data were analyzed using IBM SPSS Statistics (version 26) and SAS (version 9.4). Descriptive statistics were used to describe the participant demographic and treatment characteristics. Owing to skewness of many of the data distributions, Spearman correlations of scores from HN-LEF SI and each of the scores from the validated measures were used for evidence of construct validity.

As noted in the description of the scoring of the HN-LEF SI, two scores may be generated for each subscale at each time of assessment. The maximum value for any one symptom comprising a subscale may be a useful clinical tool; the average value for all the symptoms comprising a subscale may be of more use for research purposes.¹¹ We generated correlations of each type of score (maximum and average) of the HN-LEF SI with scores from the validated measures. Furthermore, assessments used in this study were conducted at multiple time points.

Correlations are based partly on the actual covariation of two variables and partially on the amount of variability demonstrated by the two variables. Because presence and severity of symptoms vary dramatically after the course of cancer treatment and with the types of symptoms, the maximum observed correlation between a given HN-LEF SI score and validated measure score observed from all posttreatment assessments are given in this article and used as evidence for convergent/divergent validity (or lack thereof). Owing to a wide range of measurement characteristics of the established measures used in this study, relative standards for demonstrations of “good” compared with “poor” construct validity were used in our evaluations of the correlations.

Statistical significance is of little use for studies of validity given that we want much larger indications of shared variability between two scores than that required for simply statistical significance. Sample sizes varied slightly among time periods; regardless of any correlation >0.33 was statistically significant (p < 0.05). Thus, correlations lesser than ±0.33 were deemed to be poor in that they represented <10% shared variability between the two scores (divergent validity). We chose to highlight correlations greater than ±0.45 (at least 20% shared variability) and gave an even greater focus to correlations greater than ±0.55 (at least 30% shared variability) for evidence of convergent validity.

Results

Description of the sample

Among 120 patients enrolled, 117 patients completed the study. The participants had a mean age of 59 years (standard deviation = 10.6; range 27–82), were primarily White (n = 110, 94%), mostly male (n = 84, 72%), married or living with a partner (74%), and had some college or higher education (n = 71, 61%) (Table 2). Most participants had oropharyngeal cancer, 81% of them had HPV-positive tumor, and 68% of them had advanced stage disease (III/IV). Most participants (66%) received at least two modalities of cancer treatment (e.g., surgery and radiation) (Table 2).

Table 2.

Sample Characteristics (N = 117)

Characteristics	N	Mean (SD)
Age, years	117	59.0 (10.6)
		n (%)
Gender	117
Female		33 (28.2)
Male		84 (71.8)
Race	117
White		110 (94.0)
Black or African American		4 (3.4)
Other		3 (2.6)
Marital status	117
Single/widowed/other		30 (25.6)
Married/single, living with partner		87 (74.4)
Highest education level
Less than or equal to HS	117	46 (39.3)
Undergraduate		56 (47.9)
Graduate		15 (12.8)
Employment status	117
Employed		69 (59.0)
Not employed		48 (41.0)
Area of residence	117
City		33 (28.2)
Country		51 (43.6)
Suburb		33 (28.2)
Primary tumor location	117
Oral cavity		45 (38.5)
Oropharynx		72 (61.5)
Virus	73
No known virus		14 (19.2)
HPV		59 (80.8)
Cancer stage	117
I		4 (3.4)
II		33 (28.2)
III		30 (25.6)
IV		50 (42.8)
Complete cancer treatment received	111
Surgery only		35 (31.5)
Radiation only		2 (1.8)
Surgery and radiation		4 (3.6)
Chemo and radiation		54 (48.7)
Surgery, chemo, and radiation		16 (14.4)

SD, standard deviation.

Associations of HN-LEF SI with selected established measures

Summaries of the correlations derived from the average HN-LEF SI subscale scores are given in Table 3; those from the maximum (highest single item) scores are summarized in Table 4. As shown, the patterns of correlations for the two types of scores were quite similar.

Table 3.

Maximum Correlations of the Average Head and Neck Lymphedema and Fibrosis Symptom Inventory Scores with Established Self-Report Measure Scores

Established measures	HN-LEF Symptom Inventory
	Soft issues/neurologic toxicity	Systemic symptoms/social functioning	Jaw and oral dysfunction	Swallowing and taste changes	Body image and sexuality	Communication	Mucosal irritation
EORTC-QLQ Global Health	−0.47	−0.58	−0.45	−0.41	−0.34	−0.44	−0.48
Neck Disability Index	0.57	0.72	0.48	0.39	0.47	0.49	0.49
Short-Form McGill Pain Questionnaire
Sensory	0.52	0.53	0.53	0.53	0.44	0.52	0.50
Affective	0.50	0.64	0.43	0.54	0.49	0.49	0.54
Present pain intensity	0.54	0.56	0.50	0.47	0.31	0.43	0.42
Visual analogue	0.56	0.63	0.38	0.49	0.34	0.44	0.44
VHNSS
Mouth pain	0.32	0.44	0.30	0.60	0.45	0.29	0.39
General pain	0.56	0.56	0.45	0.52	0.34	0.49	0.58
Swallowing solids	0.23	0.45	0.46	0.61	0.43	0.40	0.47
Swallowing liquids	0.16	0.37	0.19	0.44	0.31	0.34	0.50
Nutrition	0.16	0.52	0.33	0.61	0.37	0.24	0.47
Mucus	0.38	0.49	0.36	0.51	0.36	0.45	0.60
Dry mouth	0.24	0.59	0.19	0.70	0.40	0.34	0.41
Taste/smell	0.24	0.45	0.27	0.64	0.48	0.26	0.36
Voice/communication	0.51	0.56	0.53	0.53	0.58	0.76	0.62
Teeth	0.40	0.27	0.38	0.32	0.45	0.32	0.25
Hearing	0.31	0.48	0.44	0.33	0.24	0.40	0.52
Trismus	0.41	0.32	0.70	0.37	0.36	0.41	0.28
Neck range of motion	0.57	0.36	0.45	0.37	0.50	0.48	0.26
Hospital Anxiety and Depression Scale
Total	0.46	0.73	0.42	0.45	0.56	0.54	0.54
Anxiety	0.37	0.64	0.34	0.32	0.46	0.43	0.49
Depression	0.53	0.71	0.46	0.53	0.54	0.59	0.51
VHNSS-GSS
Unexplained fatigue	0.47	0.71	0.37	0.44	0.37	0.50	0.53
Fatigue limits activity	0.42	0.71	0.37	0.51	0.37	0.50	0.57
Falling asleep	0.49	0.64	0.43	0.43	0.29	0.26	0.45
Staying asleep	0.42	0.66	0.30	0.45	0.39	0.33	0.49
Unexplained sweating	0.42	0.47	0.28	0.19	0.25	0.41	0.49
Cold	0.27	0.51	0.23	0.55	0.33	0.33	0.35
Hot	0.22	0.46	0.30	0.30	0.28	0.32	0.37
Memory	0.47	0.41	0.35	0.34	0.31	0.48	0.37
Joint pain/aches	0.46	0.48	0.36	0.35	0.29	0.43	0.45
Sad/depressed	0.31	0.57	0.31	0.43	0.48	0.40	0.41
Anxious	0.27	0.51	0.40	0.34	0.40	0.35	0.45
Marlowe–Crowne Social Desirability Scale—Short Form C	−0.28	−0.32	−0.14	−0.20	−0.38	−0.19	−0.25

Values in italics indicate at least 20% shared variability; values in bold indicate at least 30% shared variance.

Maximum observed correlation from the posttreatment assessment time points.

Average of all symptom (item) values comprising each HN-LEF Symptom Inventory subscale.

EORTC-QLQ, European Organization for Research and Treatment of Cancer Quality of Life Questionnaire; HN-LEF, Head and Neck Lymphedema and Fibrosis; VHNSS-GSS, Vanderbilt Head and Neck Symptom Survey—General Symptom Survey.

Table 4.

Maximum Correlations Using the Maximum Head and Neck Lymphedema and Fibrosis Symptom Inventory Scores with Established Self-Report Measure Score

Established measures	HN-LEF Symptom Inventory
	Soft issues/neurologic toxicity	Systemic symptoms/social functioning	Jaw and oral dysfunction	Swallowing and taste changes	Body image and sexuality	Communication	Mucosal irritation
EORTC-QLQ Global Health	−0.25	−0.52	−0.27	−0.33	−0.34	−0.38	−0.49
Neck Disability Index	0.48	0.66	0.35	0.39	0.43	0.51	0.52
Short-Form McGill Pain Questionnaire
Sensory	0.45	0.50	0.45	0.42	0.43	0.51	0.47
Affective	0.47	0.62	0.49	0.38	0.44	0.47	0.50
Present pain intensity	0.56	0.53	0.37	0.40	0.31	0.41	0.45
Visual analogue	0.46	0.59	0.40	0.35	0.32	0.45	0.48
VHNSS
Mouth pain	0.34	0.30	0.40	0.40	0.42	0.43	0.37
General pain	0.49	0.49	0.49	0.55	0.37	0.49	0.60
Swallowing solids	0.40	0.51	0.43	0.56	0.39	0.45	0.42
Swallowing liquids	0.27	0.44	0.41	0.53	0.36	0.33	0.47
Nutrition	0.33	0.45	0.33	0.52	0.37	0.42	0.47
Mucus	0.22	0.49	0.34	0.60	0.36	0.40	0.57
Dry mouth	0.21	0.45	0.26	0.63	0.37	0.29	0.44
Taste/smell	0.35	0.57	0.53	0.60	0.46	0.71	0.62
Voice/communication	0.42	0.57	0.52	0.63	0.60	0.70	0.59
Teeth	0.50	0.49	0.39	0.36	0.40	0.73	0.56
Hearing	0.33	0.49	0.57	0.26	0.23	0.37	0.44
Trismus	0.43	0.29	0.66	0.40	0.28	0.35	0.29
Neck range of motion	0.48	0.36	0.56	0.38	0.49	0.26	0.23
Hospital Anxiety and Depression Scale
Total	0.36	0.62	0.40	0.40	0.52	0.46	0.57
Anxiety	0.45	0.62	0.47	0.44	0.44	0.57	0.51
Depression	0.46	0.65	0.43	0.49	0.54	0.47	0.53
VHNSS–GSS
Unexplained fatigue	0.46	0.63	0.38	0.42	0.35	0.52	0.50
Fatigue limits activity	0.45	0.64	0.46	0.50	0.38	0.49	0.59
Falling asleep	0.41	0.61	0.30	0.39	0.31	0.31	0.39
Staying asleep	0.32	0.66	0.29	0.37	0.43	0.44	0.49
Unexplained sweating	0.33	0.58	0.26	0.41	0.37	0.30	0.43
Cold	0.25	0.50	0.32	0.52	0.36	0.33	0.36
Hot	0.24	0.48	0.24	0.31	0.29	0.51	0.38
Memory	0.46	0.43	0.35	0.28	0.28	0.42	0.44
Joint pain/aches	0.34	0.51	0.28	0.42	0.47	0.40	0.40
Sad/depressed	0.45	0.47	0.35	0.41	0.48	0.39	0.43
Anxious	0.18	0.45	0.42	0.31	0.41	0.34	0.47
Marlowe–Crowne Social Desirability Scale—Short Form C	−0.25	−0.23	−0.05	−0.21	−0.37	−0.19	−0.25

Values in italics indicate at least 20% shared variability; values in bold indicate at least 30% shared variance.

Maximum observed correlation from the posttreatment assessment time points.

Highest single symptom (item) value comprising each HN-LEF Symptom Inventory subscale.

Convergent and divergent validity for each HN-LEF SI subscale

The HN-LEF SI soft tissue and neurologic toxicity average and maximum scores were most strongly correlated with the more general symptomatic validated measures (NDI, SF-MPQ, HADS, and with the VHNSS), general pain and neck range of motion scores (r_s ≥ 0.45), thus demonstrating convergent validity. Furthermore, the average score demonstrated a meaningful inverse correlation with the EORTC-QLQ global health score (r_s = −0.47). The relatively low correlations of both the average and maximum scores with dissimilar VHNSS symptom scores such as dry mouth (r_s < 0.33) provided evidence of divergent validity (Tables 3 and 4).

HN-LEF SI systemic symptoms and social functioning scores demonstrated strong positive correlations (convergence) with the measures assessing fatigue, mood disorders, and general function (VHNSS-GSS fatigue/asleep, NDI, SF-MPQ affective, HADS, r_s ≥ 0.60; EORTC-QLQ global health, r_s ∼ −0.55). Because of the more general nature of this subscale, it was at least moderately correlated with the other measures with the exception of social desirability (MC-C: r_s = −0.32 average scores, −0.23 maximum scores) providing evidence of divergent validity (Tables 3 and 4).

The HN-LEF SI jaw and oral dysfunction scores correlated most strongly with the VHNSS trismus scores (r_s > 0.65). Other than the positive correlations with the previously validated measures specific to oral functioning, very low correlations were observed for both the average and maximum scores with most of the other symptom scores (r_s < 0.33) demonstrating divergent validity for this subscale (Tables 3 and 4).

HN-LEF SI swallowing and taste changes average and maximum scores demonstrated the strongest correlations (convergence) with the respective VHNSS symptom scores reflecting swallowing difficulties and taste dysfunction. Correlations of the maximum Swallowing/Taste scores with all of those VHNSS symptoms were >0.50; the respective average Swallowing/Taste correlations also were >0.50 with the exception of the VHNSS swallowing liquids score (r_s = 0.44). Divergent validity was demonstrated by low correlations of this subscale score with other measures not specific to swallowing and taste (e.g., hearing, r_s = 0.33 average scores, 0.26 maximum scores) (Tables 3 and 4).

HN-LEF SI body image and sexuality scores were specifically associated with general and regional body-related validated measures (NDI, HADS, VHNSS voice/communication, taste/smell, and neck range of motion, r_s ≥ 0.45). The HN-LEF SI communication subscale average and particularly the maximum scores were most strongly correlated with the respective VHNSS communication-related symptom scores (voice/communication, teeth, taste/smell, r_s ≥ 0.70) (Tables 3 and 4). Very low correlations were observed with social desirability (MC-C: r_s = −0.19 both average and maximum scores) demonstrating divergent validity for this subscale (Tables 3 and 4).

Finally, the HN-LEF SI mucosal irritation scores were meaningfully positively correlated with most of the expected VHNSS symptom scores. The strongest correlations were observed with the VHNSS mucus (r_s = 0.60), voice/communication (r_s = 0.62), general pain (r_s = 0.60), and taste/smell (r_s = 0.62) scores. Divergent validity for the scores was demonstrated by very low correlations with VHNSS trismus and neck range of motion scores (r_s < 0.30) (Tables 3 and 4).

Finally, the MC-C was mentioned previously in terms of its role in demonstrating divergent validity for the general HN-LEF SI systemic symptoms/social functioning and communication subscales. It was included to assess the extent to which patient responses to the HN-LEF SI might be driven by social desirability. The low correlations of the MC-C scores with both sets of HN-LEF SI scores provide very little evidence for social desirability driving those responses (Tables 3 and 4).

Discussion

We have previously reported on the content validity and reliability of the HN-LEF SI.¹¹ In this article, we present data assessing the construct validity. We correlated patient responses to the HN-LEF SI with six carefully selected, validated measures that encompass a wide array of biopsychosocial outcomes. The results demonstrate expected correlations supporting convergent and divergent validity of the HN-LEF SI.

Construct validity of soft tissue and neurologic toxicity subscale

Studies have investigated the loss of function associated with soft tissue damage related to LEF; however, little attention has been given to the sensory symptoms associated with the tissue damage itself. The soft tissue/neurologic toxicity subscale contains items that describe the HNC patient's symptom experience related to the soft tissues involved by LEF. Because of the unique content of this subscale, the ability to validate the subscale using established patient-reported outcome measures is limited.

However, as anticipated, there is high correlation between the soft tissue/neurological toxicity subscale and the NDI, the SF-MPQ, and the general pain subscale and the neck range of motion item of the VHNSS. The low correlation with dissimilar symptoms (e.g., dry mouth) demonstrated the divergent validity of the soft tissue/neurologic toxicity subscale. These findings confirmed our hypotheses and supported the construct validity of the soft tissue/neurologic toxicity subscale.

The soft tissue/neurologic toxicity subscale contains items that represent distinct neurosensory symptoms, such as, pain, tenderness, numbness, and tightness. Unfortunately, the pathophysiology of neurosensory symptoms in HNC patients is both complex and poorly characterized. Acutely, the vast majority of HNC patients will experience pain or dysesthesia related to tissue effects of tumor or treatment that result in activation of peripheral nociceptors and mechanoreceptors followed by propagation of the stimulus to second- and third-order neurons.²⁷

In addition, tumor and surgery may damage nerves resulting in a peripheral neuropathy.²⁸ Of note, peripheral nerves are relatively radiation resistant in the acute setting;²⁹ thus, radiation-induced peripheral neuropathy does not contribute significantly to sensory symptomatology during and immediately after completion of therapy. Forty percent of HNC survivors will go on to experience chronic pain³⁰ and the percentage of patients with dysesthesia may be substantially higher. The high rate of long-term sensory symptoms may be related to a number of factors.

First, if a noxious stimulus (e.g., lymphedema) is either intense or persists for a protracted period of time, peripheral and central sensitization may occur. In this setting, symptoms may persist despite the resolution of the inciting agent.³¹ In addition, late tissue damage with resultant atrophy, lymphedema, fibrosis, and vascular insufficiency affect a broad array of tissues including nerves, skin, adipose tissue, muscles, and bone.^32–34 This results in a broad range of neurosensory syndromes. For example, direct axonal damage with demyelination, fibrosis within and around nerve trucks, and ischemic injury to the vascular supply of nerves may result in peripheral neuropathic symptoms.³² Denervation of muscles secondary to damage anywhere along the nerve segment may lead to muscular spasticity with cramping.

Lymphedema, fibrosis, and altered posture may result in musculoskeletal impairment and associated pain and dysesthesia. A better understanding of the pathophysiology and clinical manifestations of neurosensory symptoms may enhance our ability to develop targeted intervention strategies.

Construct validity of systemic symptoms and social functioning subscale

Severe or protracted systemic inflammation may result in a neuroinflammatory state manifested by systemic symptoms such as anxiety, depression, chronic fatigue, and impaired sleep.^24,35 Often referred to as Central Sensitivity Syndrome,^24,35,36 this symptom complex is seen in numerous disease processes associated with acute or chronic inflammation.^36–38 LEF are associated with elevated levels of inflammatory mediators;³⁴ thus, it is not surprising that patients with LEF manifest high levels of systemic symptoms.

The systemic symptoms and social functioning subscale of the HN-LEF SI captures three cardinal symptoms that comprise the syndrome: anxiety, fatigue, and sleep disturbance. As expected, this subscale correlates with the HADS (total score, anxiety subscale, and depression subscale), the affective items of the SF-MPQ and items from the GSS subscale of VHNSS, as well as the NDI (total score). The low correlation with the MC-C demonstrated the divergent validity of the systemic symptoms and social functioning subscale. These findings verified our hypotheses and supported the construct validity of the systemic symptoms and social functioning subscale.

The literature indicates that systemic symptoms are associated with decreased levels of social functioning.^24,35 Two items from the systemic symptom and social functioning subscale of the HN-LEF SI, “inability to participate in hobbies” and “decreased social activity,” capture the social impact of the symptom burden experienced by HNC patients with LEF. Although these types of activities are not necessary for survival, they bring joy and meaning to life. When patients experience chronic systemic symptoms, hobbies and socializing are often the first activities to be eliminated thus decreasing quality of life. Of note, the systemic symptom and social functioning subscale of the HN-LEF SI had the strongest negative correlation with the EORTC-QLQ global health item.

Construct validity of jaw and oral dysfunction subscale

The jaw and oral dysfunction subscale captures manifestations of LEF specific to the oral cavity. Loss of plasticity and range of motion of anatomical structures in the oral cavity secondary to LEF can have a devastating adverse impact on mastication, swallowing, and speech.^34,39 As expected, there was a strong correlation between the items of the jaw and oral dysfunction subscale and the trismus item and the swallowing solid subscale of the VHNSS, which confirmed the convergent validity of the jaw and oral dysfunction subscale. The relatively low correlation with the MC-C and other dissimilar symptoms (e.g., dry mouth) demonstrated the divergent validity of the jaw and oral dysfunction subscale.

Of interest, there was also a correlation with pain items on both the SF-MPQ and the VHNSS. This is an important observation that should be further explored. It may be hypothesized that movement of oral cavity structures affected by LEF may be a pain generator. Unfortunately, stretching and strengthening exercises are critical to prevent function loss. If pain prevents a patient from implementing these exercises, patients may find themselves in a self-perpetuating cycle of continue loss of function.

Construct validity of swallowing and taste changes subscale

LEF are critical contributing factors to long-term swallowing difficulties in HNC survivors.^7,9,39 Instrumental assessment of swallow function with either a modified barium swallow or a fiberoptic endoscopic evaluation of swallowing is the gold standard for swallow evaluation and is particularly important in the setting of suspected aspiration.^40,41 However, both are expensive, time consuming, and require skilled operators. A number of self-report measures have demonstrated good correlation with objective measures of swallow function and, therefore, may be able to serve as an effective screening tool. The HN-LEF SI has three items that capture swallow function: difficulty swallowing solids, difficulty swallowing soft solids, and the sensation of something being stuck in the throat. Taste abnormality clustered with the three swallow dysfunction items. This may be due to a common underlying pathophysiology of mucosal damage.

As anticipated, the swallowing and taste changes subscale of the HN-LEF SI correlated strongly with the swallowing and taste dysfunction subscales of the VHNSS. The low correlation with the MC-C and other dissimilar symptoms (e.g., hearing, neck range of motion) demonstrated the divergent validity of the swallowing and taste changes subscale. These findings proved the construct validity of the swallowing and taste changes subscale.

Of note, there were also strong correlations between the swallowing and taste changes subscale with the multiple subscales from the VHNSS including mouth pain, mucus, nutrition, dry mouth, and voice/communication; the SF-MPQ (reflecting multiple items with higher scores), and the HADS depression subscale. These findings are not surprising because pain/mucosal pain, excess mucous, and oral dryness may interfere with normal swallowing or exacerbate already impaired swallow function, leading to nutrition and weight problems.

In addition, studies conducted in the HNC population have shown that the severity of dysphagia (swallowing difficulty) correlated with voice problems, pain, and depression.^42–44 These findings indicate that patients with moderate-to-severe dysphagia would require an interdisciplinary team approach involving nutritional support, physical therapy, speech rehabilitation, pain management, and psychological counseling.^24,44

Construct validity of body image and sexuality subscale

Studies capturing the impact of HNC therapy on body image, sexuality, and intimacy have been limited.^45,46 This is in part owing to the lack of well-validated tools that specifically address the unique issues faced by HNC patients. An earlier study indicated that HNC patients postcancer treatment frequently experienced body image disturbance, problems with sexual interest and enjoyment, and problems with intimacy; and altered body image postsurgery was strongly linked to intimacy.⁴⁷

A few studies demonstrate a correlation between body image and depression in HNC patients.⁴⁸ The strength of this relationship is confirmed in this study as the body image and sexuality subscale of the HN-LEF SI strongly correlates with the HADS depression subscale. In addition, a recent literature review shows that body image disturbance has an impact on patients' daily functioning.⁴⁵ Our findings from this study confirmed this observation. Voice, taste/smell, and neck range of motion subscales of the VHNSS were among the strongest correlations with the body image and sexuality subscale of the HN-LEF SI. These findings supported the convergent validity of the body image and sexuality subscale. The low correlations with the MC-C demonstrated the divergent validity of the body image and sexuality subscale.

As expected, the remaining two subscales (communication and mucosal irritation) of the HN-LEF SI behaved. The scores for the communication subscale correlated most strongly with the voice/communication subscale from the VHNSS. The mucosal irritation subscale, which includes mucous production, sore throat, and hoarseness correlated most strongly with mucus and voice/communication subscales of the VHNSS. Again, the low correlations with the MC-C demonstrated the divergent validity of these two subscales.

Strengths and limitation

The strength of this study is that the HN-LEF SI was validated using the six established measures that involve a wide range of relevant biopsychosocial outcomes, which allowed to assess and empirically support the convergent and divergent validity of each subscale of the HN-LEF SI. There are a number of limitations in the study. First, the convenience sample may result in selection bias. Those who agreed to participate may not represent the targeted population as a whole; however, this represents the accessible population. Second, the study followed participants up to 12 months postcancer treatment. Further testing of the HN-LEF SI is needed among HNC survivors who complete their cancer treatment over 12 months.

Clinical implications

HN-LEF can lead to symptoms such as discomfort, difficulty swallowing, and stiffness in the jaw and neck with loss of range of motion. In addition, emotional distress owing to altered physical appearance is common. Previously there were no standard tools to capture symptoms related to LEF in HNC patients. To fill in the gap, we developed a patient-reported outcome measure (HN-LEF SI) and reported the content validity and reliability of the HN-LEF SI previously. In this article, we further examined the construct validity of the HN-LEF SI. The results demonstrate expected correlations supporting construct validity.

Data from our work demonstrated that the HN-LEF SI is a valid and reliable tool that can be easily used in clinical settings to assess symptom burden related to LEF among HNC patients. Clinicians can use this measure to identify patients' symptoms related to LEF; refer appropriate patients for timely lymphedema treatment; and monitor the symptomology results of therapy. Researchers can use this tool for future investigations of interventional studies in evaluating symptomology changes among HNC patients with LEF.

Conclusions

We correlated patient responses to the HN-LEF SI with six validated measures. Evidence from this work supports the construct validity of the HN-LEF SI. Further longitudinal validation is ongoing.

Authors' Contributions

J.D., B.M., and M.D. contributed to the study conception and design. Material preparation, data collection, and analysis were performed by J.D., M.D., and B.M. The first draft of the article was written by J.D., B.M., and M.D. and all authors commented on previous versions of the article. All authors read and approved the final version of the article.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

Research reported in this publication was supported by the National Institute of Dental & Craniofacial Research of the National Institutes of Health under award number R01DE024982. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.