Factor analysis of subjective descriptors of chronic pruritus and association with quality of life: a cross-sectional survey

Chronic pruritus is common with substantial quality-of-life (QOL) impact. Existing surveys capture multiple itch characteristics, but limited data provide empirical evidence on how aspects of itch characteristics inform itch-related QOL. To address this gap, we aimed to identify latent factors (hidden combinations of itch characteristics) from subjective itch descriptors that predict impact on QOL.

In a secondary analysis of a cross-sectional survey of 405 US veterans reporting chronic pruritus lasting >6 weeks¹, we identified latent factors within 45 subjective itch descriptors as measured by the Questionnaire for the Assessment of Pruritus (QAP)². Latent factors were empirically derived from observed itch descriptors using exploratory factor analysis with varimax rotation. Latent factors were correlated with itch-specific QOL impact as measured by ItchyQoL³ total score and domain scores using multivariable linear regression, adjusting for significant sociodemographic differences. P<0.05 in 2-sided tests was considered significant.

Participants were mostly male (92.6%) with mean age of 60.8 (standard deviation [SD], 13.1) and mean total ItchyQoL score of 53.4 (SD, 18.5). Four latent factors were identified with eigenvalue >1. The top 5 contributing descriptors (with corresponding factor loading) to each latent factor are shown in Table 1. Latent factors 1 and 2 were loaded with descriptors that aligned with severity (e.g., “annoying,” “terrible”) and respectively explained 69% and 9% of variance (Table 1). Latent factors 3 and 4 were loaded with descriptors that aligned with sensation (e.g.,“burning,” “pricking”) and explained 8% and 5% of variance. All 4 factors were significantly associated with ItchyQoL total score and subscores (Table 2). Secondary analyses using only complete patient questionnaires (N=322) and using promax rotation exploratory factor analysis yielded similar factor loading patterns (data not shown).

TABLE 1.

Rotated Factor Matrix Outcomes

	Factor 1		Factor 2		Factor 3		Factor 4
Top 5 Contributors	Annoying	(84)	Terrible	(75)	Burning	(68)	Pricking	(63)
(factor loading x 100%)	Bothering	(81)	Torturing	(70)	Hurting	(63)	Pinprick- Like	(58)
	Bothersome	(80)	Dreadful	(68)	Hot	(62)	Tingling	(55)
	Unpleasant	(76)	Oppressive	(62)	Painful	(61)	Sharp	(53)
	Itching	(67)	Awful	(62)	Warm	(58)	Feels Ant Like	(43)
Eigenvalues*	17.38		2.26		1.27		1.20
Proportions	0.69		0.09		0.08		0.05
Cumulative Proportion	0.69		0.78		0.83		0.88

^*The initial number of factors equaled the number of descriptors (45), however, only the first few factors were retained based on an eigenvalue > 1. Thus, four factors were retained for the analysis and explained a cumulative proportion of variance of 88% which is sufficient to explain the data. The top 5 contributing descriptors from each factor are presented here.

^**Promax rotation of the data was also performed as a sensitivity analysis, which yielded essentially identical latent factors, thus the data is not shown here.

TABLE 2.

Linear regression model predicting ItchyQoL scores using Factors 1–4 and top contributors to Factors 1–4.

	Total ItchyQoL		Itchy QoL: Symptom		ItchyQoL: Function		ItchyQoL: Emotion
	Beta	P	Beta	P	Beta	P	Beta	P
F1	5.05	<.001	1.15	<.001	1.45	<.001	2.55	<.001
F2	9.97	<.001	1.53	<.001	2.97	<.001	5.35	<.001
F3	7.48	<.001	2.01	<.001	2.29	<.001	2.87	<.001
F4	2.14	0.003	0.39	0.091	0.57	0.038	1.15	0.003
	Adjusted R2	0.58	Adjusted R2	0.42	Adjusted R2	0.45	Adjusted R2	0.53

Our study supports current conceptual models of chronic pruritus^2,4,5 and expands upon these models to provide a framework for interpretation and use of data derived from clinical surveys that capture subjective itch descriptions. Our empirical data aligned with proposed groups of itch descriptors on “affective dimension” (“severity” in latent factors 1–2) and “sensation” (latent factors 3–4)². We also showed that these severity and sensation descriptor groupings were highly predictive of ItchyQoL. Additionally, our identified factors highlight a more parsimonious group of descriptors that may further inform theoretical frameworks^2,4,5 by which subjective itch characteristics impact QOL. Extensive assessment of itch characteristics, especially in relation to itch-related QOL, can be cumbersome and inefficient in the clinical setting. These latent factors of subjective descriptors may inform future research on parsimonious and effective methods to assess the impact of itch in clinical settings. Our survey participants were limited to older male veterans with modest response rate, which may limit generalizability. While the QAP included 45 descriptors, they were not all-encompassing in the subjective assessment of itch. Confirmatory factor analysis will be required to verify their importance in future research and clinical settings. In sum, we identified latent factors which represent distinct patterns of itch characteristics and demonstrated how each pattern associated with QOL, thus expanding upon current theoretical frameworks of the assessment of pruritus.