Principal component analysis of the ICIQ-FLUTS / ICIQ-MLUTS questionnaires, gender differences, and LUTS-related quality of life

Shenhar C1, Mohammad A2, Dobberfuhl A2, van Uem S2

Research Type

Clinical

Abstract Category

Quality of Life / Patient and Caregiver Experiences

Abstract 271
Best Conservative Management
Scientific Podium Session 32
Friday 29th September 2023
11:30 - 11:45
Room 104CD
Questionnaire Quality of Life (QoL) Female Male Incontinence
1. Stanford University, California , USA, Rabin Medical Center, Israel, and Cleveland Clinic, Ohio, USA, 2. Stanford University, California, USA
Presenter
Links

Abstract

Hypothesis / aims of study
Various patient reported outcome measures exist to study lower urinary tract symptoms (LUTS) and LUTS-related quality of life (QoL). The International Consultation on Incontinence Questionnaire - female lower urinary tract symptoms (ICIQ-FLUTS) and male lower urinary tract symptoms (ICIQ-MLUTS) are validated, highly-recommended questionnaires with a broad scope, but are relatively underused in clinical practice and the interpretation of their scores continues to be investigated. The International Prostate Symptom Score (IPSS) is a long-established LUTS questionnaire with a narrower scope.
Our aim was to study the effects of questionnaire choice and gender on the capturing of LUTS and LUTS-related QoL, using a principal component analysis.
Study design, materials and methods
We prospectively enrolled consenting adults presenting with lower urinary tract symptoms (LUTS) to a functional urology clinic. Patients underwent clinical evaluation and answered validated questionnaires about their LUTS and quality of life (QoL). Questionnaires included the International Consultation on Incontinence Questionnaire (ICIQ) female and male LUTS (ICIQ-FLUTS and ICIQ-MLUTS) as well as the International Prostate Symptom Score (IPSS).

We used Statistical Analysis System software (SAS, Cary, NC) to analyze data, using logistic regression to model effects of gender and age on LUTS. Tabulated data were presented as mean ± standard deviation (SD) and number (percentage) of observations. Baseline characteristics were compared using Student’s t test and Chi-square test. A p-value of p<0.05 was considered statistically significant. Spearman’s rho was used to test for non-parametric correlations between corresponding questionnaire items.
We employed an unsupervised principal component analysis (PCA) to reduce data dimensionality and identify LUTS domain groups in male and female IPSS, ICIQ-FLUTS, and ICIQ-MLUTS questionnaires. Briefly, questionnaire answers can be modeled as a multidimensional cloud of data, where every question is represented by an axis in a single dimension, and every patient is represented by a data point. PCA detects new axes traversing this cloud, representing variance over several dimensions. In an unsupervised process, new axes that represent more variance than that of a former axis (or a single question), are selected as principal components, thus reducing the number of axes and dimensions while keeping most of the variance. The data is then projected on the new axes, or “rotated”, to determine the contribution of each questionnaire item to the principal components and infer the clinical domain this component represents.
We used the principal components summarizing domains in each questionnaire as variables in a logistic regression to determine their effect on LUTS-related QoL in females and in males.
Results
A total of 304 Patients, 178 females and 126 males, were enrolled and presented for 540 clinic visits. Mean age at enrollment was 58±17 years. Response rates were 93% (500/540) for the IPSS, 91% (490/540) for the ICIQ symptom scores, and 64% (343/540) for the ICIQ bother scores.

Regression model for effects of gender on LUTS
A logistic regression model by gender, adjusted for age (Figure 1), demonstrated a greater number of women with urgency using both instruments (IPSS OR 2.27, 95% CI 1.55-3.33; ICIQ OR 2.25, 95% CI 1.42-3.56). Women reported more voiding symptoms than men, including straining (IPSS OR 1.70, 95% CI 1.10-2.62; ICIQ OR 1.79, 95% CI 1.04-3.06). Incontinence, too, was reported more by women then by men (ICIQ stress urinary incontinence (UI) OR 6.14, 95% CI 2.55-14.77; ICIQ unaware UI OR 2.18, 95% CI 1.03-4.62; ICIQ sleep UI OR 1.99, 95% CI 1.02-3.90; ICIQ urge UI OR 1.94, 95% CI 1.16-3.26).

Principal Component Analysis
For female patients, unsupervised PCA identified 2 principal components of IPSS, explaining 53% and 16% of the overall variance, and 3 principal components of the ICIQ-FLUTS questionnaire, accounting for 35%, 23%, and 11% of data variance.
After varimax rotation, factor 1 in the IPSS received the most contribution from questions regarding incomplete emptying, intermittent stream, weak stream, and straining to urinate, and determined to be a voiding-phase symptom component. Factor 2 in IPSS was contributed by daytime frequency, urgency, and nocturia, and deemed to represent storage-phase symptoms.
The 3 ICIQ-FLUTS principal components were concluded to represent: i. incontinence (contributed by urge incontinence, stress incontinence, frequency of incontinence, unaware incontinence, and sleep incontinence); ii. voiding symptoms and pain (hesitancy, straining, intermittency, and pain related to the bladder); and iii. storage phase symptoms (nocturia, urgency, and daytime frequency).
In a similar manner, unsupervised PCA detected 2 principal components for male IPSS and 3 principal components for ICIQ-MLUTS. IPSS components corresponded to voiding and storage symptom domains, while ICIQ-MLUTS components represented: i. incontinence (urge, stress, unaware, sleep, and post-void incontinence; and a lower yet worth-mentioning contribution from urgency); ii. voiding (hesitancy, straining, weak stream, intermittency, incomplete emptying); and iii. storage-phase symptoms (urgency, daytime frequency, and nocturia).

Principal Component Logistic regression
In females, maximum likelihood estimates for the effect of LUTS domains on QoL were significant for all components of the ICIQ-FLUTS (incontinence p<0.0001, voiding p<0.0001, storage p=0.008) and for the storage component of the IPSS (p=0.004). In males, maximum likelihood estimates were significant for the incontinence (p=0.024) and storage (p=0.027) components of the ICIQ MLUTS, as well as the voiding (p=0.018) and storage (p=0.006) components of the IPSS. 
Given the significant effect of urinary incontinence on QoL, we plotted the predicted probabilities of each level of LUTS-related QoL using the incontinence principal component scores, while holding the other principal components constant. This identified the inflection point of QoL reporting probabilities. Females with positive incontinence scores had higher probabilities of reporting IPSS QoL of 5 (unhappy) or 6 (terrible), while in males, positive incontinence scores were related to a higher probability of IPSS QoL 4, 5 or 6 (mostly dissatisfied, unhappy, or terrible) (Figure 2).
Interpretation of results
In this study we analyzed gender differences in LUTS. We found that male and female LUTS questionnaires, with 7 to 13 items, may be condensed into 2 to 3 principal components, each with its unique effect on LUTS-related quality of life. We show how ICIQ & IPSS questionnaire capture LUTS in a gender-specific manner. Voiding symptoms, in particular, were a better predictor of female QoL when captured by ICIQ, and a better predictor of male QoL when captured by the IPSS. Incontinence, a separate predictor of poor LUTS related QoL in both males and females, is not represented in the IPSS at all.
An unsupervised PCA detected components in line with our current understanding of LUTS domains, although interestingly, incontinence was a focal component on its own and not fully represented by storage-phase symptoms.

Interestingly, the male principal components of the ICIQ-MLUTS closely followed the female pattern of voiding, storage, and incontinence, despite the different organization of the ICIQ-MLUTS subscales and items - namely voiding, incontinence, and stand-alone items for daytime frequency and nocturia.

Our study emphasizes the strengths and weaknesses of the ICIQ-FLUTS/MLUTS and IPSS questionnaires. While the ICIQ questionnaires captured more dimensions of bothersome LUTS, namely incontinence, they lack an overall LUTS-related QoL score.
Concluding message
Voiding, storage and incontinence LUTS are captured differently in the ICIQ-FLUTS/MLUTS and the IPSS questionnaires, in a manner that varies by gender. Urinary incontinence, absent from IPSS, is a major driver of overall LUTS-related QoL.
In translating our findings to clinical practice, clinicians who only use the IPSS questionnaire, should practice with a heightened index of suspicion for incontinence as a driver of bothersome QoL.
Figure 1 Figure 1. LUTS domain odds ratios (OR) by gender.
Figure 2 Figure 2: Principal component effect model of the ICIQ incontinence domain on IPSS QoL in females and males, holding other LUTS components constant
References
  1. Uren AD, Cotterill N, Pardoe M, Abrams P. The International Consultation on Incontinence Questionnaires (ICIQ): An update on status and direction. Neurourology and Urodynamics. 2020;39(6):1889–96.
  2. Barry MJ, Fowler FJ, O’Leary MP, Bruskewitz RC, Holtgrewe HL, Mebust WK, et al. The American Urological Association Symptom Index for Benign Prostatic Hyperplasia. Journal of Urology. 1992 Nov;148:1549–57.
  3. Donovan JL, Peters TJ, Abrams P, Brookes ST, De la rosette JJMCH, Schäfer W. Scoring the Short Form ICSmaleSF Questionnaire. The Journal of Urology. 2000 Dec 1;164(6):1948–55.
Disclosures
Funding NIH 1L30DK115056-01, Stanford WSDM Seed Grant Clinical Trial No Subjects Human Ethics Committee Stanford University School of Medicine IRB Helsinki Yes Informed Consent Yes
Citation

Continence 7S1 (2023) 100988
DOI: 10.1016/j.cont.2023.100988

13/11/2024 22:56:32