Lymphedema Symptom Intensity and Distress Surveys—Truncal and Head and Neck,Version 2.0

Abstract

Background:

A comparison of symptom prevalence, intensity, and distress for participants with truncal lymphedema, head and neck lymphedema, and no lymphedema identified a need for a truncal-specific, lymphedema-related symptom assessment tool and a revision of the Lymphedema Symptom Intensity and Distress Survey—Head and Neck (LSIDS-H&N). The purpose of this study was to institute the development of the Lymphedema Symptom Intensity and Distress Survey—Truncal (LSIDS-T) and revise the LSIDS-H&N.

Methods and Results:

A comprehensive midline measure and subsequent series of analyses were used to develop the LSIDS-T and revise the initial LSIDS-H&N. Participants included 97 without lymphedema, 82 with truncal lymphedema, and 72 with head and neck lymphedema. Cluster analysis for the LSIDS-T resulted in five clusters with a total of 21 items. Cluster analysis for the LSIDS-H&N resulted in seven clusters with a total of 31 items. Key correlations in expected directions were found with the validated measures for both surveys, and correlations with the Marlowe Crown Social Desirability Scale did not indicate issues with social desirability of response.

Conclusion:

The 24-item LSIDS-T and the 31-item revised LSIDS-H&N v.2 are promising additions to the suite of other LSIDS measures for use in clinical environments.

Introduction

Lymphedema is traditionally viewed as a nonlife-threatening diagnosis with potentially serious complications if not well-managed^1,2; however, studies that focus on symptoms and functional deficits in patients with lymphedema illuminate a broad constellation of symptoms, some very serious and even life threatening.^3–6 Assessing lymphedema symptoms in a quantifiable manner is important not only to help illustrate the disease burden on patients, but to measure the impact of treatment on the patient's function and quality of life.

Although formal anatomical classification of lymphedema remains inconsistent, it is generally acknowledged that treatment of lymphedema is unique based on the anatomical region.^1,2 As an example, Viehoff et al. categorized lymphedema as upper limb, lower limb, and midline (including head and neck and thorax), but went on to further divide midline into head and neck and genital in their analysis.⁷ We have previously developed and completed testing of self-report symptom profile surveys for upper limb (Lymphedema Symptom Intensity and Distress Survey-Arm)⁸ and lower limb (Lymphedema Symptom Intensity and Distress Survey-L).⁹ We also conducted pilot testing of the Lymphedema Symptom Intensity and Distress Survey—Head and Neck (LSIDS-H&N).³ During the developmental work for the LSDIS-A, breast cancer survivors frequently reported swelling under their armpit and in their breast. The preliminary lower limb survey included a small section assessing truncal lymphedema prevalence and symptoms. That study identified a higher than expected prevalence of truncal lymphedema and suggested that truncal lymphedema symptom profiles could be distinct from those of lower limb lymphedema.

Based upon those findings, other work being conducted that clearly demonstrated some symptoms unique between head and neck and genital lymphedema, and recommendations that a midline measure be developed and/or existing tools modified,⁶ a previously published, cross-sectional study was undertaken. That study determined if truncal lymphedema symptom profiles differed between head and neck lymphedema regions and if a single midline measure would be clinically useful.¹⁰ In that study, a comparison of symptom prevalence, intensity, and distress for participants with truncal (swelling located between the shoulder and the groin) lymphedema, head and neck lymphedema, and no lymphedema established a need for a truncal-specific, lymphedema-related symptom assessment tool. It also supported revision of the pilot LSIDS-H&N. A single, midline (swelling located from the head to the groin) assessment tool was not supported. Thus, the purpose of this study was to institute the development of the Lymphedema Symptom Intensity and Distress Survey—Truncal (LSIDS-T) and revise the LSIDS-H&N.

Materials and Methods

Ethics

Institutional Review Board (IRB) approval was obtained from Vanderbilt University. The study was conducted in accordance with the ethical standards of the Vanderbilt IRB and with the Declaration of Helsinki of 1975, as revised in 2008. Informed consent was obtained from all participants before commencing study activities.

Sample

The sample comprised 97 healthy volunteers, 82 participants with truncal lymphedema, and 72 with head and neck lymphedema. Recruitment methods, details of data collection, and symptom descriptive summaries have been previously published.¹⁰

Data collection and measures

Participants completed on-line, self-report surveys through REDCap (Research Electronic Data Capture) as follows¹¹:

Demographics

A 10-item self-report form including questions about gender, race/ethnicity, marital and employment status, insurance, and income.

Health and lymphedema form

A seven-item self-report form including questions about medications, health conditions and surgeries, lymphedema diagnosis, type, cause, location, and treatment.

Preliminary LSIDS-T and revised LSIDS-H&N

A 48-item self-report tool includes symptoms related to both truncal lymphedema and head and neck lymphedema. This tool was developed based on the following steps: The 67 items comprising LSIDS-H&N, version 1.0 were developed and preliminarily tested in 30 patients with head and neck cancer (HNC).³ The survey used a 1 (slight) through 10 (severe) Likert rating scale for severity (intensity) and bother (distress).³ Subsequently, questions arose as to whether head-and-neck symptoms of lymphedema were distinct from more “midline” or “truncal” lymphedema symptoms and an International group of scholars, including author S.H.R., met during the International Lymphoedema Framework (ILF) 2014 Conference to discuss lymphedema classification.¹² To assist with addressing this question, a measure was developed that contained the most prevalent symptoms that had been endorsed by patients with head and neck lymphedema in the LSIDS-H&N pilot study, symptoms suggested by Viehoff et al. work in midline symptomatology,^6,12 and face-valid proposed truncal lymphedema symptoms. This resulted in a 48-item tool that incorporated the same response format as our previously developed suite of LSIDS measures.^8,9 Participants first indicated whether they were experiencing a given symptom. If present, they then indicated on separate five-point Likert response scales the intensity of the symptoms and the degree of symptom distress.

Functional Assessment Screening Questionnaire

A 15-item survey assessing functional impairment in adults with five Likert-type responses.^13,14 Scores generated in this study had an internal consistency of 0.92 (Cronbach's α).

Profile of Mood States—Short Form

A 37-item survey that captures mood in adults with five Likert-type responses.^15,16 The tool includes six subscales: tension, depression, anger, vigor, fatigue, and confusion. The Cronbach's alpha values for the subscale scores generated in this study ranged from 0.80 (Confusion) to 0.89 (Fatigue) and was 0.90 for the total score.

Lymphedema Quality of Life Inventory

A 45-item survey assessing quality of life in lymphedema patients using four Likert-type responses. Cronbach's alpha coefficients of 0.88, 0.92, and 0.88 for the three survey domains of physical, psychosocial, and practical have been reported.⁴ Scores include a mean score for each domain and scores for two single-item scores for qualitative aspects of lymphedema (typical in the last 4 weeks and overall quality-of-life rating). Internal consistency of the total scores in this study was 0.97; those for the three domains scores ranged from 0.90 to 0.95 (Cronbach's α).

Marlowe–Crowne Social Desirability Scale Short Form C

A 13-item survey assessing social desirability bias in adults with true/false questions found to correlate strongly with the 33-item full Marlowe–Crowne scale (r = 0.93).^17–19 The internal consistency of the scores generated in this study was 0.69 (Kuder–Richardson).

Statistical approaches

Data were analyzed using IBM SPSS Statistics 26 (Armonk, NY) and SAS 9.4 (Cary, NC). Frequency distributions were used to summarize the nominal and ordinal sample demographic characteristics. Because of skewed distributions, median and interquartile range (IQR) were used to summarize all continuous variable distributions including scores for the validated measures. Empirical generation of the clusters of symptom responses was conducted using SAS PROC VarClus. This approach is statistically derived from hierarchical clustering methods and loosens many of the assumptions inherent in traditional factor analysis (e.g., normality of item response distributions), thus making it most suitable for symptom clustering. Once symptom clusters were generated, the internal consistency of the respective cluster scores were evaluated using the Cronbach's alpha statistic. Initial assessments of the scores' validity were evaluated by correlating each of the scores with the scores from the validated measures using Spearman coefficients. For evidence of validity, the size of coefficients was required to be considerably higher than that required for minimal statistical significance, which would be a correlation as small as 0.33 (p < 0.05).

Results

Detailed sample demographics have been previously published.¹⁰ The sample of 82 participants with truncal lymphedema had a median age of 55.8 years (IQR = 50–63), was primarily white (90.1%), non-Hispanic (98.7%), mostly women (90.2%), and married or living with a partner (61.7%). The sample of 72 participants with head and neck lymphedema had a median age of 60.3 years (IQR = 50–66), was primarily white (87.3%), non-Hispanic (98.6%), approximately two-thirds men (66.7%), and married or living with a partner (74.6%; Table 1). Descriptive summaries of the measures used to assess the preliminary concurrent and construct validity of the LSIDS-T and LSIDS-H&N version 2.0 are given in Table 2. In general, scores on the validated measures were generally in the “positive” half of their possible ranges. Scores on Lymphedema Quality of Life Inventory (LyQLI) indicated in general very good quality of life (lower scores are better)^4,20 ; yet, as with the LSIDS-H&N distributions, there was sufficient variability (Table 2).

Table 1.

Demographic Characteristics of Participants with Truncal (N = 82) and Head and Neck (N = 72) Lymphedema

	Truncal (N = 82)	H&N (N = 72)
	Median [IQR]	Median [IQR]
Age	55.8 [50–63] (N = 77)	60.3 [50–66]
Years of education	16.0 [14–17]	14.0 [12–16]
	N (%)	N (%)
Gender
Female	74 (90.2)	24 (33.3)
Male	8 (9.8)	48 (66.7)
Race	N = 81	N = 71
Multiracial	3 (3.7)	2 (2.8)
Black or African American	4 (4.9)	5 (7.0)
White	73 (90.1)	62 (87.3)
Other^a	1 (1.2)	2 (2.8)
Ethnicity	N = 79	N = 69
Hispanic or Latino	1 (1.3)	1 (1.4)
Not Hispanic or Latino	78 (98.7)	68 (98.6)
Relationship status	N = 81	N = 70
Single/other	31 (38.3)	18 (25.4)
Married/living w/partner	50 (61.7)	53 (74.6)
Employment
Employed full- or part-time	39 (47.6)	34 (47.2)
Retired	18 (22.0)	19 (26.4)
On disability	17 (20.7)	12 (16.7)
Not employed or other^b	8 (9.8)	7 (9.7)
Residence
Urban	25 (30.5)	24 (33.3)
Rural	13 (15.9)	20 (27.8)
Suburb/other	44 (53.7)	28 (38.9)
Insurance
None	0 (0.0)	1 (1.4)
Government	26 (31.7)	18 (25.0)
Private	45 (54.9)	40 (55.6)
Multiple types/other	11 (13.4)	13 (18.1)
Annual household income	N = 79	N = 63
$30,000 or less	20 (25.3)	13 (20.6)
Over $30,000	59 (74.7)	50 (79.4)

Includes Asian and Columbian.

Includes homemaker, student, and self-employed.

H&N, head and neck; IQR, interquartile range.

Table 2.

Summaries of Scores from Measures Used to Validate the Lymphedema Symptom Intensity and Distress Survey—Truncal (N = 82) and Lymphedema Symptom Intensity Distress Survey—Head and Neck (N = 72)

	Truncal (N = 82)		H&N (N = 72)
	Median [IQR]	Min − max (N)	Median [IQR]	Min − max (N)
FASQ (0–60)^a	48.0 [40–54]	25–60 (82)	52.0 [44–57]	0–60 (72)
POMS-SF total (0–148)^b	41.5 [32–54]	24–106 (82)	35.0 [29–45]	24–92 (72)
POMS-SF subscales^c
Tension (0–24)	5.0 [2–8]	0–23 (82)	3.5 [1–7]	0–18 (72)
Depression (0–32)	5.5 [2–10]	0–29 (82)	3.0 [1–6]	0–19 (72)
Anger (0–28)	4.0 [1–7]	0–23 (82)	3.0 [0–5]	0–16 (72)
Vigor (0–24)	7.0 [4–11]	0–23 (82)	9.0 [6–14]	0–23 (72)
Fatigue (0–20)	8.0 [5–12]	0–20 (82)	5.0 [3–9]	0–19 (72)
Confusion (0–20)	3.5 [1–6]	0–15 (82)	2.0 [1–4]	0–17 (72)
LyQLI total (0–123)^d	1.80 [0.9–2.2]	0–3 (82)	0.40 [0.0–0.9]	0–2 (70)
LyQLI subscales^d
Physical (0–36)	1.71 [1.0–2.3]	0–3 (82)	0.50 [0.0–1.1]	0–3 (71)
Psychosocial (0–48)	1.50 [0.7–2.2]	0–3 (82)	0.25 [0.0–0.7]	0–2 (70)
Practical (0–39)	1.85 [1.1–2.4]	0–3 (81)	0.31 [0.0–0.7]	0–2 (70)
MCSDS-SFC (0–13)^e	8.5 [6–10]	1–13 (82)	9.0 [8–11]	0–13 (72)

Higher scores indicate higher function.

Higher scores indicate more negative mood state.

Each subscale score indicates higher levels of that state; note that vigor is reversed for POMS-SF total scoring because it is the only subscale for which higher scores indicate a more positive mood state.

Higher score indicates lower quality of life.

Higher scores indicate an increased tendency to socially desirable responses over accurate responses.

FASQ, Functional Assessment Screening Questionnaire; LyQLI, Lymphedema Quality of Life Inventory; MCSDS-SFC, Marlowe–Crowne Social Desirability Scale—Short Form C; POMS-SF, Profile of Mood States—Short Form.

Prevalence values of all the specific symptoms included in the initial pool of items for the two measures have been reported previously.¹⁰ As we observed in the development of our earlier LSIDS measures, ∼25% of the participants did not respond to one or more of the three items about sexuality.^8,9 Therefore, those items were grouped (clustered) together based on face validity only and not included in the statistical clustering analysis of either measure.

Preliminary statistical clustering of the 48 truncal and head–neck symptoms did not converge into a single “midline” symptom cluster solution for both groups of patients. Although the more systemic symptoms demonstrated convergence, the more specific activity, behavioral, and soft tissue symptoms did not. Therefore, a specific “truncal” measure was developed and an updated version, the “head-and-neck” measure was generated. Measurement characteristics of those measures are presented hereunder.

Lymphedema Symptom Intensity and Distress Survey—Truncal

Clustering of symptoms (content and structural validity)

Based on the earlier work on the prevalence of symptoms in the group with truncal lymphedema and the preliminary clustering work illustrating the need for separate measures, a set of 24 systemic and truncal-specific symptoms was included in a statistical clustering analysis. That analysis resulted in five clusters of symptoms labeled: Soft Tissue Sensation (seven items), Activity (six items), Resources (two items), Neurological Sensations (two items), and Biobehavioral (seven items). The items comprising each of those clusters are given in Table 3.

Table 3.

Lymphedema Symptom Intensity and Distress Survey—Truncal Clusters (N = 82)

Cluster label: Cronbach's α, no. of items	Item	Label	Median [IQR]	Min	Max
Soft tissue: 0.86, 7 items	1	Heaviness Trunk	3.3 [1.9–5.4]	0.0	10.0
	2	Tightness Trunk
	3	Pain Trunk
	4	Numbness Trunk
	5	Achiness Trunk
	6	Swelling Trunk
	7	Hardness Trunk
Activity: 0.87, 6 items	10	Difficulty Bending	4.0 [1.3–5.9]	0.0	9.8
	11	Difficulty Standing
	18	Fatigue
	20	Hobbies/Leisure Inhibited
	21	Decreased Social Activity
	22	Decreased Physical Activity
Resources: 0.90, 2 items	13	Lack Confidence Insurance	0.0 [0.0–3.3]	0.0	10.0
Resources: 0.90, 2 items	16	Frustration Insurance	0.0 [0.0–3.3]	0.0	10.0
Neuro: 0.83, 2 items	8	Tingling Trunk	0.0 [0.0–2.2]	0.0	10.0
Neuro: 0.83, 2 items	9	Pins/Needles Trunk
Biobehavioral: 0.92, 7 items	12	Lack Self-Confidence	4.0 [1.4–6.8]	0.0	10.0
	14	Appearance Concerns
	15	Less Sexually Attractive
	17	Lost Body Confidence
	19	Lack Sex Interest
	23	Decreased Sexual Activity
	24	Difficult Sexual Activity
Overall: 0.93, 24 items			3.2 [1.9–5.2]	0.0	9.0

Consistency of the symptom responses within each cluster were further investigated by generating Cronbach's alpha internal consistency values for scores that would result from those clusters. As given in Table 3 those values ranged from 0.83 to 0.90 confirming the cluster compositions. Scores for each cluster were generated by averaging the individual symptom scores comprising each respective cluster. For the clusters comprising more than two items, up to 40% of randomly missing item responses were allowed when calculating those average scores. An overall score for the LSIDS-T was computed by averaging all 24 items (allowing for up to five missing symptom responses). The Cronbach's alpha reliability coefficient of that overall score was 0.93. The possible range of values for each cluster and the overall score was 0–10. Although the median and IQR values for the scores tended to be in the lower half of the possible range, scores ≥8 were observed for each cluster.

Concurrent criterion and divergent validity

Correlations of the scores from the validated measures with each of the scores from the LSIDS-T are given in Table 4. To ease detection of patterns in those correlations, correlations of at least ±0.40 were highlighted.

Table 4.

Correlations of Lymphedema Symptom Intensity and Distress Survey—Truncal Cluster Scores with Validation Measures Scores (N = 82)

	LSIDS-T cluster or total score
	Soft tissue	Activity	Resources	Neurological	Biobehavioral	Total
FASQ total	−0.39	−0.82	0.03	−0.31	−0.54	−0.66
POMS-SF: Total	0.36	0.63	−0.19	0.27	0.54	0.55
POMS-SF: Tension	0.30	0.52	−0.18	0.34	0.44	0.47
POMS-SF: Depression	0.34	0.64	−0.19	0.22	0.52	0.53
POMS-SF: Anger	0.25	0.46	−0.23	0.19	0.43	0.41
POMS-SF: Vigor	−0.26	−0.58	0.03	−0.14	−0.42	−0.49
POMS-SF: Fatigue	0.26	0.75	−0.07	0.10	0.50	0.57
POMS-SF: Confusion	0.36	0.58	−0.07	0.32	0.49	0.54
LyQLI: Total	0.76	0.68	0.42	0.15	0.77	0.35
LyQLI: Physical	0.69	0.45	0.56	0.23	0.64	0.44
LyQLI: Psychosocial	0.70	0.75	0.29	0.08	0.71	0.27
LyQLI: Practical	0.68	0.59	0.35	0.17	0.71	0.30
Marlowe–Crowne	−0.12	−0.27	0.10	0.03	−0.13	−0.10
Marlowe–Crowne	≤−0.60	≤−0.40	≥ 0.40	≥0.60	−0.13	−0.10

We hypothesized that the LSIDS-T cluster scores would demonstrate a divergent pattern of validity coefficients (very low) with the scores from the MCSDS. Those observed correlations were all <±0.27.

It was hypothesized that all LSIDS-T clusters would be inversely correlated with the Functional Assessment Screening Questionnaire (FASQ) total score and that the strongest of those associations would be observed for the Activity cluster score. The strongest and most meaningful inverse correlation was observed with the Activity cluster score (r_s = −0.82). All remaining correlations with the FASQ total score were more than ±0.31 (which would represent ∼10% shared variance; Table 4).

Regarding the Profile of Mood States—Short Form (POMS-SF), we hypothesized that the Activity score would demonstrate the strongest associations with the POMS-SF Vigor and Fatigue subscales and that the Biobehavioral score, which in this case also included sexuality symptoms, would be most strongly associated with depression and fatigue scores. As given in Table 4, the strongest correlation of the Activity score was with the Fatigue subscale score (r_s = 0.75), the next strongest correlation was with the depression subscale (r_s = 0.64), not the vigor subscale. The LSIDS-T Biobehavioral cluster, as expected, was correlated with both the POMS-SF Depression (r_s = 0.52) and Fatigue (r_s = 0.50) subscales. It was expected that the strongest correlations for the LSIDS-T Soft Tissue would be with the POMS-SF total score. There was only a weak (r_s = 0.36) correlation with the total score. We thought the Resources scores would correlate most strongly with the Anger subscale, and the strongest correlation (compared the correlations with other subscales of the POMS-SF), although weak, was indeed with this subscale (r_s = 0.23). We did not anticipate a correlation between the neurological cluster scores and any POMS-SF subscale, given the physiological nature of the two symptoms. This assumption was met.

We hypothesized that the LSIDS-T Activity scores would correlate most strongly with the LyQLI Psychosocial subscale scores; Biobehavioral with the Psychosocial subscale score; Soft Tissue with the Physical subscale score, and Neurological with the Physical subscale scores. We did not anticipate that the Resources cluster would correlate strongly with any of the three subscales. As anticipated, the Activity Cluster correlated most strongly with the Psychosocial subscale (r_s = 0.75). The Biobehavioral cluster associated with both the Psychosocial (r_s = 0.71) and the Practical (r_s = 0.71) subscale scores. The Soft Tissue Cluster correlated closely with all three subscale scores, Psychosocial (r_s = 0.70), Physical (r_s = 0.69), and Practical (r_s = 0.68). Surprisingly, the Neurological cluster did not correlate with any relevant degree to any subscale (r_s = 0.08–0.23), whereas the Resources cluster correlated with the Physical subscale score (r_s = 0.56).

Lymphedema Symptom Intensity Distress Survey—Head and Neck

Clustering of symptoms (content and structural validity)

From our earlier work on prevalence of symptoms in the group of participants with lymphedema in the head and neck region, a set of 35 symptoms were included in a statistical clustering analysis. Results from that analysis indicated that four symptoms (taste change, skin change, difficulty talking, and sleep difficulty) had poor cluster loading in general, and/or poor consistency of response with the other symptoms in their respective cluster. The cluster analysis of the remaining 31 item scores resulted in seven clusters of symptoms labeled: Soft Tissue Sensation (nine items), Activity (six items), Oral Function (five items), Resources (two items), Neurological Sensations (two items), Biobehavioral (four items), and Sexuality (three items). The items comprising each of those clusters are given in Table 5.

Table 5.

Lymphedema Symptom Intensity Distress Survey—Head and Neck Version 2.0 Clusters (N = 72)

Cluster label: Cronbach's α, no. of items	Item	Label	Median [IQR]	Min	Max
Soft tissue: 0.91, 9 items	1	Heaviness H/N	2.0 [0.6–4.9]	0.0	8.7
	2	Tightness H/N
	3	Pain H/N
	4	Numbness H/N
	5	Achiness H/N
	6	Swelling H/N
	7	Hardness H/N
	10	Difficulty Moving H/N
	12	Difficulty Opening Mouth
Activity: 0.83, 6 items	11	Difficulty Bending	1.2 [0.0–2.8]	0.0	9.3
	14	Difficulty Standing
	25	Fatigue
	27	Hobbies/Leisure Inhibited
	28	Decreased Social Activity
	29	Decreased Physical Activity
Oral function: 0.83, 5 items	13	Difficulty Moving Tongue	2.1 [0.4–3.8]	0.0	9.6
	15	Difficulty Chewing
	16	Difficulty Swallowing Solids
	17	Difficulty Swallowing Soft/Liquids
	18	Stuck in Throat Sensation
Resources: 0.95, 2 items	20	Lack Confidence Insurance	0.0 [0.0–0.0]	0.0	10.0
Resources: 0.95, 2 items	23	Frustration Insurance	0.0 [0.0–0.0]	0.0	10.0
Neuro: 0.94, 2 items	8	Tingling H/N	0.0 [0.0–0.0]	0.0	8.0
Neuro: 0.94, 2 items	9	Pins/Needles H/N	0.0 [0.0–0.0]	0.0	8.0
Biobehavioral: 0.84, 4 items	19	Lack Self-Confidence	1.1 [0.0–2.5]	0.0	9.0
	21	Appearance Concerns
	22	Less Sexually Attractive
	24	Lost Body Confidence
Sexuality: 0.83, 3 items	26	Lack Sex Interest	1.3 [0.0–3.3]	0.0	10.0
	30	Decreased Sexual Activity
	31	Difficult Sexual Activity
Overall: 0.94, 31 items	31	Difficult Sexual Activity	1.7 [0.7–3.2]	0.1	8.1

Internal consistency values of scores that would be generated from the head and neck symptom clusters ranged from 0.83 to 0.95 confirming the cluster compositions. Scores for each cluster were generated by averaging the individual symptom scores comprising each respective cluster. For the clusters comprising more than two items, up to 40% of randomly missing item responses were allowed when calculating those average scores. An overall score for the LSIDS-H&N was computed by averaging all 31 items (allowing for up to five missing symptom responses). The Cronbach's alpha reliability coefficient of that overall score was 0.94. The possible range of values for each cluster and the overall score was 0–10. Although the median and IQR values for the scores tended to be in the lower half of the possible range, scores ≥8 were observed for each cluster.

Concurrent criterion and divergent validity

Correlations of the scores from the validated measures with each of the scores from the LSIDS-H&N are given in Table 6. As with the LSIDS-T, we hypothesized that the LSIDS-H&N cluster scores would demonstrate a divergent pattern of validity coefficients (very low) with the scores from the MCSDS. Those observed correlations were all less than ±0.26.

Table 6.

Correlations of Lymphedema Symptom Intensity Distress Survey—Head and Neck cluster scores with validation measures scores (N = 72).

	LSIDS-H&N cluster or total score
	Soft tissue	Activity	Oral function	Resources	Neurological	Biobehavioral	Sexuality	Total
FASQ total	−0.21	−0.44	−0.18	−0.12	−0.18	−0.19	−0.22	−0.31
POMS-SF: Total	0.42	0.48	0.31	−0.03	0.27	0.50	0.34	0.50
POMS-SF: Tension	0.27	0.42	0.20	−0.01	0.33	0.38	0.22	0.36
POMS-SF: Depression	0.50	0.55	0.41	0.02	0.20	0.50	0.34	0.60
POMS-SF: Anger	0.30	0.14	0.21	−0.09	0.22	0.34	0.28	0.28
POMS-SF: Vigor	−0.22	−0.58	−0.23	−0.24	−0.17	−0.28	−0.26	−0.42
POMS-SF: Fatigue	0.33	0.72	0.20	0.01	0.16	0.33	0.41	0.45
POMS-SF: Confusion	0.39	0.51	0.30	0.00	0.24	0.49	0.40	0.50
LyQLI: Total	0.67	0.57	0.34	0.04	0.23	0.51	0.36	0.66
LyQLI: Physical	0.72	0.52	0.40	−0.01	0.27	0.44	0.33	0.67
LyQLI: Psychosocial	0.60	0.57	0.28	0.07	0.19	0.51	0.32	0.60
LyQLI: Practical	0.62	0.61	0.28	0.13	0.24	0.39	0.42	0.62
Marlowe–Crowne	−0.10	−0.26	0.00	0.13	−0.24	−0.23	−0.02	−0.14
Marlowe–Crowne	≤−0.40	≥0.40	≥0.60	0.13	−0.24	−0.23	−0.02	−0.14

It was hypothesized that all LSIDS-H&N clusters would be inversely correlated with the FASQ total score, yet the strongest of those associations would be observed for the Activity, Soft Tissue, Biobehavioral, Sexuality, and Neurological sensation cluster scores. The strongest and most meaningful inverse correlation was observed with the Activity cluster score (r_s = −0.44). All remaining correlations with the FASQ total score were less than ±0.33 (which would represent ∼10% shared variance).

Regarding the POMS-SF, as in our earlier work, we hypothesized that the most informative validity information would come from correlation of the LSIDS-H&N cluster scores with specific subscales. We hypothesized that the Activity score would demonstrate the strongest associations with the POMS-SF Vigor and Fatigue subscales and that the Biobehavioral score would be most strongly associated with depression and confusion scores. As given in Table 6, the strongest correlations of the Activity score were with the Vigor and Fatigue subscale scores (r_s = 0.72 and −0.58, respectively); yet, similar correlations were also observed with the POMS-SF Depression and Confusion scores (r_s = 0.55 and 0.51, respectively). The LSIDS-H&N Biobehavioral cluster was equally correlated with both the POMS-SF Depression and Confusion subscale scores (r_s = 0.50 and 0.49, respectively). As expected, the strongest correlations for both the LSIDS-H&N Soft Tissue and Oral Function scores were with the POMS-SF Depression scores (r_s = 0.50 and 0.41, respectively); the Sexuality cluster score was most strongly and equally correlated with the POMS-SF Fatigue and Confusion scores (r_s = 0.41 and 0.40, respectively). Finally, considerably lower than expected correlations were observed for the LSIDS-H&N Resources and Neurological scores with the POMS-SF subscales. All correlations for the Resources scores were less than ±0.25 and the strongest correlation for the Neurological scores was with the POMS-SF Tension score (r_s = 0.33).

We hypothesized that the LSIDS-H&N Soft Tissue and Oral Function scores would correlate most strongly with the LyQLI Physical scores; Activity and Sexuality with the LyQLI Practical score; and Biobehavioral with the LyQLI Psychosocial subscale score. We expected that the LSIDS-H&N Resources and Neurological cluster scores would not correlate strongly with any of the LyQLI subscales. As given in Table 6, although the strongest correlation of the LSIDS-H&N Soft Tissue score was with the LyQLI Physical subscale (r_s = 0.72), correlations with the other subscales were almost equally strong (Psychosocial: r_s = 0.60, Practical: r_s = 0.62). The LSIDS-H&N Oral Function score was most strongly correlated with the LyQLI Physical score (r_s = 0.40). The strongest correlations for both the LSIDS-H&N Activity and Sexuality scores were with the LyQLI Practical subscale (r_s = 0.61 and 0.42, respectively); and the strongest Biobehavioral correlation was with the LyQLI Psychosocial (r_s = 0.61). Both the LSIDS-H&N Resources and Neurological demonstrated relatively low correlations with all the LyQLI Scores (Resources: r_s < ±0.13, Neurological: r_s < ±0.27).

Discussion

Findings in our previous study suggested the need for a truncal-specific lymphedema assessment measure, and a revision to the initial LSIDS-H&N. This was confirmed when preliminary statistical clustering of the 48 truncal and head–neck symptoms did not converge into a single “midline” symptom cluster solution for both groups of patients. Therefore, the LSIDS-T was developed and tested as part of our larger, on-going measurement development endeavors to broaden the battery of lymphedema symptom assessment tools available to clinicians and researchers.^8,9 The LSIDS-T represents the first known measure developed to specifically capture symptoms and symptom burden in patients with truncal lymphedema. The Cronbach's alpha reliability coefficient of that overall score was 0.93, and most of all hypothesized correlations were confirmed. The measure takes ∼5 minutes to complete. Therefore, clinicians and researchers can confidently use the LSIDS-T in their respective work environments,

The need to modify the LSIDS-H&N measure to assess head and neck lymphedema and its associated symptoms was especially high given the large percentage of HNC survivors who develop lymphedema.²¹ Over 100 patients with HNC-related lymphedema participated in the development of this measure across the initial study and this study. The Cronbach's alpha reliability coefficient of that overall score was 0.94. Although several hypothesized correlations were not confirmed and suggest the need for more work, key correlations in expected directions support the validity of this measure. For example, the LSIDS-H&N Soft Tissue and Oral Function scores clustered with the POMS-SF Depression scores as expected and a strong inverse correlation was observed with the Activity cluster score (r_s = −0.44) and the FASQ total score. Correlations with the Marlowe–Crowne Social Desirability Scale—Short Form C (MCSDS-SFC) did not indicate issues with social desirability of response. These findings support that the LSIDS-H&N can be used for the intended purpose, lymphedema symptom assessment. The 31-item measure takes ∼5–7 minutes to complete and can easily be completed by patients using pen and paper, or electronic formats such as the REDCap platform used in the study.¹¹ Thus, clinicians should be able to integrate the revised tool into clinical practice and use the information gleaned from the measure to plan patient care and to identify the need for referrals (e.g., social services, psychological counseling). Researchers can also use the LSIDS-H&N in this patient population.

Limitations of this study should be taken into consideration when using these measures. First, volumetric measurement of swelling in the truncal and head and neck areas is extremely difficult. We did not objectively measure swelling in the patients; however, many of the patients were recruited using a patient registry consisting of individuals with known lymphedema in the targeted regions. Thus, the relationship of severity of swelling with symptoms could not be determined. Second, the relative low level of severity and intensity of some of the clusters may have limited the ability to detect some correlations. Alternatively, the use of different measures may have yielded different results. Further testing of the tool using objective measures of swelling, when perfected, and different measures would be illuminating.

Conclusion

The 24-item LSIDS-T and the 31-item LSIDS-H&N v.2 demonstrated preliminary validity for use as tools to assess and quantify truncal and head and neck lymphedema-related symptom burden. Results suggest that the measures may be a promising addition to the suite of other LSIDS measures for use in clinical environments.

Footnotes

Acknowledgments

The authors thank Melissa Adair, RN for her work on this study. The authors also thank all participants in this study who gave their time to help others.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

Martha Rivers Ingram Chair of Nursing and CTSA award UL1TR000445 from the National Center for Advancing Translational Sciences.

References

Foldi

, Foldi

, Strobenreuther

, Kubik

. Foldi's Textbook of Lymphology for Physicians and Lymphedema Therapists. 3rd ed. Munich, Germany: Elsevier; 2012.

Lee

, Bergan

, Rockson

. Lymphedema: A Concise Compendium of Theory and Practice. London, United Kingdom: Springer; 2011.

Deng

, Ridner

, Murphy

, Dietrich

. Preliminary development of a lymphedema symptom assessment scale for patients with head and neck cancer. Support Care Cancer, 2012; 20:1911–1918.

Klernas

, Johnsson

, Horstmann

, Kristjanson

, Johansson

. Lymphedema Quality of Life Inventory (LyQLI)-Development and investigation of validity and reliability. Qual Life Res, 2015; 24:427–439.

Ridner

, McMahon

, Dietrich

, Hoy

. Home-based lymphedema treatment in patients with cancer-related lymphedema or noncancer-related lymphedema. Oncol Nurs Forum, 2008; 35:671–680.

Viehoff

, Gielink

, Damstra

, Heerkens

, van Ravensberg

, Neumann

. Functioning in lymphedema from the patients' perspective using the International Classification of Functioning, Disability and health (ICF) as a reference. Acta Oncol, 2015; 54:411–421.

Viehoff

, Hidding

, Heerkens

, van Ravensberg

, Neumann

. Coding of meaningful concepts in lymphedema-specific questionnaires with the ICF. Disabil Rehabil, 2013; 35:2105–2112.

Ridner

, Dietrich

. Development and validation of the Lymphedema Symptom and Intensity Survey-Arm. Support Care Cancer, 2015; 23:3103–3112.

Ridner

, Doersam

, Stolldorf

, Dietrich

. Development and validation of the lymphedema symptom intensity and distress survey-lower limb. Lymphat Res Biol, 2018; 16:538–546.

10.

Doersam

, Dietrich

, Adair

, Rhoten

, Deng

, Ridner

. A Comparison of symptoms among patients with head and neck or truncal lymphedema and normal controls. Lymphat Res Biol, 2019; 17:661–670.

11.

Harris

, Taylor

, Thielke

, Payne

, Gonzalez

, Conde

. Research electronic data capture (REDCap)—A metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform, 2009; 42:377–381.

12.

Viehoff

, Heerkens

, Van Ravensberg

, et al. Development of consensus International Classification of Functioning, Disability and Health (ICF) core sets for lymphedema. Lymphology, 2015; 48:38–50.

13.

Millard

. The Functional Assessment Screening Questionnaire: Application for evaluating pain-related disability. Arch Phys Med Rehabil, 1989; 70:303–307.

14.

Seltzer

, Granger

, Wineberg

. Functional assessment: Bridge between family and rehabilitation medicine within an ambulatory practice. Arch Phys Med Rehabil, 1982; 63:453–457.

15.

Curran

, Andrykowski

, Studts

. Short form of the Profile of Mood States (POMS-SF): Psychometric information. Psychol Assess, 1995; 7:80–83.

16.

Shacham

. A shortened version of the Profile of Mood States. J Pers Assess, 1983; 47:305–306.

17.

Crowne

, Marlowe

. A new scale of social desirability independent of psychopathology. J Consult Psychol, 1960; 24:349–354.

18.

Goldfried

. A cross-validation of the Marlowe-Crowne Social Desirability Scale Items. J Soc Psychol, 1964; 64:137–145.

19.

Robinette

. The relationship between the Marlowe-Crowne Form C and the validity scales of the MMPI. J Clin Psychol, 1991; 47:396–399.

20.

Klernas

, Johnsson

, Boyages

, Brorson

, Munnoch

, Johansson

. Test of responsiveness and sensitivity of the questionnaire “Lymphedema Quality of Life Inventory”. Lymphat Res Biol, 2018; 16:300–308.

21.

Ridner

, Dietrich

, Niermann

, Cmelak

, Mannion

, Murphy

. A Prospective study of the lymphedema and fibrosis continuum in patients with head and neck cancer. Lymphat Res Biol, 2016; 14:198–205.