Abridging the 22‐Item Sino‐Nasal Outcome Test (SNOT‐22) in People With Cystic Fibrosis: Limiting Survey Burden

1. Introduction

Summary

• Item response theory identified the most informative items of the SNOT‐22 survey in PwCF.

• 11‐ and 6‐item sinonasal symptom surveys for PwCF were developed with initial high reliability and validity via IRT.

• Abbreviated versions aim to reduce survey fatigue in PwCF while accurately assessing CRS severity.

Chronic rhinosinusitis (CRS) affects quality‐of‐life (QoL) in people with cystic fibrosis (PwCF), despite the reduction in symptoms and symptom severity associated with cystic fibrosis transmembrane conductance regulator (CFTR) modulators [1]. To precisely assess sinus symptoms and treatment needs in the post‐modulator period, patient‐related QOL instruments should be revisited. The 22‐item SinoNasal Outcome Test (SNOT‐22) measures CRS burden and QoL. Abbreviated surveys can lessen respondent fatigue. Item response theory (IRT) can help refine surveys by identifying the most informative survey items while maintaining reliability [2, 3, 4]. IRT assesses item discrimination (ɑ), the ability to differentiate trait levels, and difficulty (β), which defines thresholds for response categories [2, 3, 4].

Liu et al. refined the SNOT‐22 using IRT with strong reliability and validity [5]. Given PwCF's unique CRS burden, this study used IRT to create abbreviated SNOT‐22 versions by selecting the most informative items. A secondary aim explored SNOT‐22 differences by modulator therapy history.

2. Methods

This cross‐sectional study, approved by local Institutional Review Boards, analyzed 185 adults (age 18 and over) with cystic fibrosis (CF) and CRS from 12 academic centers between 2018 and 2023, primarily from an ongoing trial (NCT04469439). Baseline SNOT‐22 scores were analyzed, with full information maximum likelihood used to handle missingness in the graded response IRT models assessing item discrimination and difficulty. Demographic and clinical data were collected.

Shortened surveys were developed by first retaining items contributing above‐average information within subdomains (nasal, otologic/facial pain, sleep, emotional), whereupon we applied a stricter rule ensuring at least 30% of the test information was retained from each subdomain to produce a further shortened survey, as implemented by Feng et al. [6]. Internal consistency (Cronbach's α > 0.7) and convergent validity (Pearson's r >0.8) confirmed reliability and efficacy of shortened surveys. Statistical analyses were conducted using R with the “mirt” and “psych” packages [7, 8].

3. Results

The study included 185 PwCF with CRS, predominantly male (60%), with history of modulator therapy use (77%), and with F508del variants (59%) (Table 1).

TABLE 1.

Cohort characteristics.

Cohort characteristics (n = 185)
Sex	Male	n = 111, 60.0%
	Female	n = 70, 37.8%
	Missing	n = 4, 2.2%
Smoking	Current/former	n = 6, 3.2%
	Never	n = 172, 93.0%
	Missing	n = 7, 3.8%
Alcohol	Current/former	n = 78, 42.2%
	Never	n = 78, 42.2%
	Missing	n = 29, 15.7%
History of modulator therapy	Yes	n = 142, 76.8%
	No	n = 37, 20.0%
	Missing	n = 6, 3.2%
History of endoscopic sinus surgery	Yes	n = 31, 16.8%
	No	n = 68, 36.8%
	Missing	n = 86, 46.5%
F508del variant	Yes	n = 109, 58.9%
	No	n = 63, 34.1%
	Missing	n = 13, 7.0%
History of cystic fibrosis related diabetes (CFRD)	Yes	n = 68, 36.8%
	No	n = 107, 57.8%
	Missing	n = 10, 5.4%
Prior lung transplant	Yes	n = 20, 10.8%
	No	n = 152, 82.2%
	Missing	n = 13, 7.0%
Pancreatic insufficiency	Yes	n = 144, 77.8%
	No	n = 28, 15.1%
	Missing	n = 13, 7.0%

Note: All participants completed SNOT‐22 surveys, with seven individuals providing incomplete responses to six survey items.

The mean total SNOT‐22 score was 39.3 [±19.8]. Subdomain mean scores were 13.0 [±8.3] for nasal, 4.8 [±3.8] for otologic/facial pain, 13.0 [±9.7] for sleep, and 1.5 [±2.0] for emotional. Responses ranged from 0 (“No problem”) to 5 (“Severe problem”) across most subdomains, except for the emotional subdomain, which lacked the most extreme responses.

Initial IRT analysis for all PwCF retained above‐average information items across subdomains, preserving 55%–63% of subdomain information and yielding an 11‐item survey. A stricter 30% threshold retained six items, covering 34%–54% of information in each subdomain (Figure 1). Despite shortening, difficulty thresholds remained consistent, capturing the full range of symptom severity.

FIGURE 1.

(A and B) Item‐based information of SNOT‐22 items categorized into subdomains with percentage of total information. (A) The horizontal axis represents the percentage of total information in a subdomain. The vertical orange line represents the cut‐off percentage for “above‐average” information, calculated by 100%/number of items in a subdomain. Items with “above‐average” information were retained in the abbreviated surveys. The darker bar in each graph indicates information of all PwCF, while the lighter bar indicates that of PwCF with history of modulator therapy use. (B) The horizontal axis represents the percentage of total information in a subdomain. The six‐ to seven‐item surveys retained the fewest items that encompassed at least 30% of the test information of that subdomain. The darker bar in each graph indicates information of all PwCF, while the lighter bar indicates that of PwCF with history of modulator therapy use.

Of 142 PwCF with history of modulator therapy, IRT analysis produced nine‐item and seven‐item surveys. Compared to the total PwCF cohort, the nine‐item survey excluded “reduced concentration” and “reduced productivity,” while the seven‐item survey retained “dizziness.” All other items were consistent between groups (Figure 1). Notably, PwCF with history of modulator therapy had lower average scores for “thick nasal discharge” (2.0) and “post nasal discharge” (1.7) than those not on therapy (2.4 and 2.2, respectively). Item contributions are detailed in Table S1.

Correlation analysis showed strong agreement between shortened and original SNOT‐22 scales, with Pearson's coefficients of r = 0.97 (11‐item) and r = 0.94 (6‐item) for the total PwCF cohort, and r = 0.96 (9‐item) and r = 0.95 (7‐item) for the modulator therapy subgroup. The original SNOT‐22 had a Cronbach's α of 0.93, while shortened versions maintained strong internal consistency (0.90–0.84), ensuring reliable CRS‐specific QoL assessment in PwCF.

4. Discussion

The SNOT‐22 is a validated CRS assessment tool for PwCF [5]. Using IRT in a multi‐institutional cohort, we retained high‐information items and created reliable abbreviated versions of the questionnaire specifically relevant for PwCF.

PwCF retained different SNOT‐22 items using IRT than other populations [5], indicating distinct symptom distributions and QoL impacts. The 11‐item version kept “dizziness,” “facial pain/pressure,” and “embarrassed,” while omitting “ear pain/pressure,” “nasal blockage,” and “sad.” The six‐item version showed greater divergence, retaining “waking up tired” and “fatigue.” These differences underscore the need for further study on CRS symptom distribution across populations.

When IRT analysis was applied only to PwCF on modulator therapy, it generated largely similar abbreviated surveys to across the entire sample. PwCF on modulators, however, omitted “reduced concentration” and “reduced productivity” from the 11‐item SNOT‐22 creating a nine‐item version. Using the 30% threshold, they retained the six‐item SNOT‐22 with the addition of “dizziness.” These variations may reflect differences in symptom perception post‐therapy and warrant further study.

Respondent fatigue, a decline in data quality due to survey length and complexity, may be mitigated by abbreviated surveys [9]. PwCF experience significant fatigue, with severe cases affecting up to 26% [10]. While respondent fatigue in PwCF remains unstudied, abbreviated surveys could enhance participation and data quality. Our findings support shortening the SNOT‐22 to reduce fatigue while retaining key information, emphasizing the need for tailored CRS patient reported outcome measures and validation for PwCF.

Limitations of this study include modest sample size, limited sociodemographic data, skewed beta parameters, and SNOT‐22 midpoint clustering, which may underrepresent CRS severity. Key symptoms such as nasal blockage and dysosmia were excluded in IRT‐based shortening, raising concerns about capturing CRS impact. Additional psychometric validation of the shortened scales is required. Despite this, the study supports IRT‐based SNOT‐22 shortening for PwCF while preserving key information.

Conflicts of Interest

Daniel M. Beswick: In the last 36 months. Daniel M. Beswick has received grant support from NIH/NHLBI, CF Foundation, International Society of Inflammation and Allergy of the Nose and the American Rhinologic Society CORE/Sue Ann and John L. Weinberg Foundation; honoraria from sources including from National Jewish Health; consulting fees from Amgen, on medicolegal cases and from Garner Health (equity).

Jeremiah A. Alt: Consultant for OptiNose and Medtronic. Speaker panel GSK. GlycoMira board and equity holder. Kristine A. Smith: In the last 24 months, served as consultant for SanofiGenzyme.

Zachary M. Soler: Consultant for OptiNose, Regeneron, and Lyra; Medical Directory for Healthy Humming. Rodney J. Schlosser: Consultant for OptiNose, Medtronic, Stryker, Cyrano; Medical Directory for Healthy Humming. Jennifer L. Taylor‐Cousar: In the last 36 months, JLT‐C has received grants from the CF Foundation related to this work as well as for work unrelated to the manuscript. Unrelated to this work, she has received grants for her institution from Vertex Pharmaceuticals Incorporated, Eloxx, and 4DMT; received fees from Vertex Pharmaceuticals Incorporated related to consultation on clinical research design, participation on advisory boards, and speaking engagements; and served on advisory boards and/or provided clinical trial design consultation for Insmed, 4DMT, and AbbVie. She serves on a DMC for AbbVie. She serves as the adult patient care representative to the CFF Board of Trustees, and on the CF Foundation's Clinical Research Executive Committee, Clinical Research Advisory Board, Racial Justice Working Group and as immediate past chair of the CF TDN's Sexual Health, Reproduction and Gender Research Working Group, on the scientific advisory board for Emily's Entourage, and on the ATS Respiratory Health Awards, Scientific Grant Review and Clinical Problems Assembly Programming Committees. All other authors declare no conflict of interest.

Supporting information

Supporting Information