The cross-sectional study involved four defined steps schematized in Figure 1 as follows: (1) translation and retranslation of the French version of CDQ into Spanish; (2) cultural adaptation and semantic evaluation; (3) validation of CDQ through the internal consistency determination and reproducibility of the QoL; and (4) application of the questionnaire to Spanish adult and adolescent patients with CD and statistical analysis. A study that included patients with CD was conducted between October 2020 and September 2021 by the Gastroenterology Unit in two hospitals in southern Spain (Sevilla and Algeciras).

The inclusion criteria were as follows: age > 15 years, biopsy-proven CD diagnosis, and self-reported adherence to a GFD. Demographic characteristics were collected: sex, time on the GFD (< 2, 2–5, and > 2 years), level of education, and professional status (Table 1).

The exclusion criteria included the inability to complete all the items questioned and refusing to sign the informed consent form. A total of 153 participants with CD (95 females and 58 males, age range 15–78 years) were randomly enrolled. Ethical approval was obtained from the Regional Ethics Review Board (0504-N-19). All the participants received a debriefing in advance and written consent was obtained from all the adult patients and parents or legal guardians in the case of adolescents.

The CDQ was translated from French to Spanish by an external independent translator unrelated to medical staff. After translation, two health professionals specializing in CD met with the translator and integrated views and discrepancies. Subsequently, a retrotranslation of the Spanish questionnaire into French was carried out to confirm its accuracy.

The widely used Delphi method was used for cultural adaptation and semantic evaluation. To assess whether the translated questionnaire was interpreted in our geographical, cultural, and social contexts, 20 CD volunteers were recruited in the Digestive Specialties Institute of Algeciras (Spain). The sample size calculation for the pilot study was performed after review of the populations in the previous studies and the minimum proportion in the reference groups were assumed to be 0.04. Accepting an expected difference of 0.06, with an alpha risk of 5%, a beta risk of 20% (80% of statistical power), and a loss rate of 10%, the sample size needed was 20 subjects. The calculation was made using the tool GRANMO v7.12 April 2012 (Institut Municipal d’Investigació Mèdica, Barcelona, Spain). Although these initial questionnaires were not included in the analysis, these patients were also asked to fill out a new questionnaire afterward in the validation phase. The Spanish-translated CDQ version was administered to answer the following questions on each CDQ item: (1) Is the question well understood? (2) Do the questions have a single meaning? and (3) Would you change any words or group of words to provide a better understanding of the question? in which the possible answer was “yes” or “no, providing a free text to explain if the answer was “no.” Finally, a subheading with suggestions is available. Subsequently, the two doctors met again together with the translator to provide the final translated CDQ.

To validate Spanish version of the CDQ, test–retest reliability, internal consistency, and ceiling/floor effects were calculated. The Spearman’s rank correlation coefficient was used to analyze test–retest reliability and the Cronbach’s coefficient alpha was used as an adequate measure of internal consistency. Alpha values between 0.70 and 0.95 were considered acceptable (26). The identification of floor and ceiling effects indicates the ability of the questionnaire to distinguish between respondents at the extreme ends of the scale. Floor and ceiling effects were assessed by dimension and considered as significant (≥ 15%), moderate (10 to < 15%), minor (5 to < 10%), and negligible (< 5%) (27) in relation to the participants scored the highest or lowest possible score.

The Spanish version of the CDQ was held as shown in Figure 1. Although the original version was the German, we chose the French version due to its excellent validation value, as well as its proximity to the Spanish since Mediterranean culture and lifestyle are the opposite of the German’ culture and lifestyle. After various stages, the translator and health professionals provided the final translated CDQ (Supplementary Material).

The Spanish CDQ version comprising 28 questions was classified into four subscales as follows: gastrointestinal area, emotional, social, and worries. Each of the 28 questions presented a Likert-type answer model with seven possible answers, where score 1 considers “the worst possible state,” while score 7 is “the best possible state.” The maximum overall score was established at 196 points, with a maximum score on each subscale of 49 points (16). The CDQ was sent through Google Forms application.

The EQ–5D survey comprised five questions on mobility, self-care, ability to perform daily activities, presence of pain or discomfort, and presence of anxiety and depression symptoms with three possible answers for each item (1, no problems; 2, moderate problems; and 3, severe problems). A summary index with a maximum score of 1 can be derived from these five dimensions by conversion to a table of scores. In addition, a visual analog scale (VAS) indicated the general health status, with 100 points indicating the best health status. By scoring on the VAS, patients can be classified as having good QoL (≥ 70), regular (between 50 and 69), and poor (< 50). The EQ–5D survey was sent through Google Forms application.

All the results are expressed as the mean SD. Cronbach’s alpha and Spearman’s correlation coefficients were used to assess the internal consistency and test–retest reliability, respectively. Pearson’s chi-square test was used to assess the agreement between the CDQ score and the EQ–5D survey. For intraindividual comparisons, differences of ≥ 12 within the total score and ≥ 3 within each subscore can be regarded as a minimum important clinical difference. For comparison of groups in all statistical tests, a p-value of less than 0.05 was considered statistically significant. Statistical analyses were performed using IBM SPSS 25.0 version (IBM Corporation, Armonk, NY, United States).

Of the 153 participants with CD who were initially enrolled in the study and signed informed consent, 61 were excluded due to not meeting the eligibility criteria or partially completing both questionnaires CDQ and EQ–5D (46 lack one question, nine participants lack two questions, and six incomplete questionnaires concurred with patients with three or more unfilled questions). None of the questionnaires was rejected due to multiple responses. The profiles of the excluded respondents more frequently were females, aged 26–59 years, and on GFD for < 2 years.

We included a total of 92 participants (55 [59.8%] females and 37 [40.2%] males), with an average age of 29.8 ± 14.6 years (range, 15–78 years) in the analysis. As shown in Table 1, homogeneous groups for the study CD cohorts were attempted in relation to the distribution of the volunteers by sex, time of CD diagnosis, and time on GFD.

The Spanish version of the CDQ followed the flowchart held in Figure 1. The descriptive statistical analysis was presented using subscales and the sum and mean of the scores were examined. The 20 patients did not mention any possible change to the initial Spanish translation. Test–retest reliability for each subscale of the questionnaire for the first 20 patients with CD who responded to the questionnaire was calculated and obtained a score of 0.99 on each of the subscales (Table 2). According to the international agreement, the reliability value is indicated as excellent (1–0.8), good (> 0.6), moderate (> 0.4), and poor (< 0.4) (28), therefore, our results indicate excellent reliability. The internal consistency of all the subscales was verified by the Cronbach’s alpha, yielding a score of 0.83–0.92 (Table 2), indicating a good internal consistency as internal consistency is indicated as 1 < α ≥ 0.9, excellent; 0.7 ≤ α < 0.9, good; 0.6 ≤ α < 0.7, acceptable; 0.5 ≤ α < 0.6, poor; and α < 0.5, unacceptable (29). A comparison of the internal consistency of this study was attempted with all translated and validated CDQs according to the domains. The Cronbach’s alpha was the highest in the Spanish CDQ in the worries domain and remained among the best in the social and gastrointestinal domains, except for the emotion. Similarly, the ceiling and floor effects of the subscales were established, ranging from 0.51 to 3.11% and 0.13 to 1.06%, respectively, and were therefore considered negligible (Table 2).

The comparison between the EQ–5D visual scale and the Spanish CDQ was also performed. Mobility, personal self-care, ability to accomplish daily activities, pain, and anxiety/depression symptoms scores were worse in the coeliac patients with the worst scores on the CDQ, and these results were consistent across all the four subscales (Table 3). Furthermore, higher consistency was added by the Pearson’s coefficient correlation between the total score of the study questionnaire and the total score of the validated Spanish version of the EQ–5D QoL questionnaire (r = −0.592; p < 0.01) and the score obtained on the visual scale of this questionnaire (r = 0.514; p < 0.01) (Figures 2A,B).

A total of 153 participants with CD received the specific CDQ and general EQ–5D through Google Forms application and signed an informed consent form. The common factor of all the patients with CD included in the study was the maintenance of a GFD. However, previous studies have observed differences in the adherence to GFD in different age groups; therefore, in this study, the respondents were divided into age groups: adolescent (15–17 years), young (18–25 years), adults (26–59 years), and older (> 60 years). Similarly, the time from diagnosis (< 2 and > 2 years) and the level of education (primary, secondary, and university) have been found to be associated with compliance and, therefore, with the influence of subsequent complications. All these data were compared based on the CDQ score in each subscale. Similarly, data on mobility, daily activities, personal care, presence of pain or discomfort, presence of symptoms of anxiety and/or depression, smoking, level of education, and current work activity were compared. Finally, the score data on the EQ–5D visual QoL scale were examined by comparison with the scores on the different subscales of the Spanish CDQ (Table 3).

Males showed higher scores than females in all domains; however, they were only statistically significant in the emotional and social domains (p = 0.003 and p = 0.036, respectively). The results related to age are noteworthy, as on the four subscales, a lower score was obtained in the older group than in the teens (p < 0.01). Despite newly diagnosed coeliacs displaying a short period of GFD, a lower score was obtained on all the scales, and no statistically significant results were obtained except for emotional (p = 0.012), with low scores obtained in patients who had been on GFD for < 2 years compared with those of 2–5 and > 5 years on GFD.

The effect of employment status was analyzed in four domains, with the highest scores in students and the worst scores in retired people (p < 0.01); however, similar scores were obtained for the different items among employed and unemployed participants. We analyzed participants according to their level of education, as the level of education could influence education on the disease, available gluten-free products, or proper food labeling, and therefore, could improve compliance and QoL. Statistically significant results were obtained only in the social domain (p = 0.034), with the highest score for those with secondary education and the lowest for those with primary education.

Patients with CD who are active smokers showed the worst results on all the subscales compared with those without a history of smoking. However, the gap between these subgroups of patients’ analysis was less demonstrative in the gastrointestinal symptom subscale than in the other domains (33.40 ± 7.71 vs. 30.20 ± 7.53; p = 0.078).

In addition, the interesting results obtained in all domains in the teenager group led to a more in-depth study to examine their HRQoL, and even the onset, sex, and time on GFD affected the health self-valuation. Females had worse scores on all the subscales, reaching statistically significant differences in the emotional and social subscales. Patients with scores: < 50 on the EQ–5D visual scale had a significantly lower score on the gastrointestinal symptom’s subscale than those with scores of > 50 (41.33 ± 2.51 vs. 34.00 ± 7.07; p < 0.01). However, there were no significant differences in scores on the different subscales depending on the time they had been on the GFD (Table 4).

Of the validated CDQs, the acceptance of the Spanish CDQ was like that of the French and Italian CDQs, although the number of patients included in this study was lower, with the Turkish and Iranian questionnaires having the lowest number of patients included and the German and French questionnaires having the highest number. However, this study was the most homogeneous in terms of sex and remarkable for its teenager inclusion.