Influence of Compliance to Diet and Self-Efficacy Expectation on Quality of Life in Patients with Celiac Disease in Spain

The purpose of this study is to understand the health-related quality of life (HRQoL) in patients with celiac disease (CD) and analyze its main determinants. A transversal descriptive study of 738 patients with celiac disease was carried out. A series of questionnaires were answered related to their HRQoL, adherence to a gluten-free diet (GFD), and self-efficacy beliefs among other relevant variables. Regression analyses were carried out in order to explore the predictive variables in adherence to the GFD and HRQoL. A total of 61.2% showed a good HRQoL, and the main predictors of HRQoL were specific self-efficacy, adherence to the diet, risk perception, time since diagnosis, and age. While 68.7% of participants showed good or excellent adherence to the GFD, and the main predictors of adherence were specific self-efficacy, perceived adoption of recommended behaviors, HRQoL and gender. The HRQoL of patients with CD, and adherence to the GFD in Spain, are good. It is the self-efficacy expectation, measured specifically and not generally, which is the best predictor of both adherence and HRQoL. It is necessary to develop programs to improve the HRQoL of patients with CD that focus on improving specific self-efficacy.


Introduction
In the last decades, health-related quality of life (HRQoL) has become an important issue in research. HRQol can be defined as "how well a person functions in their life and his or her perceived wellbeing in physical, mental, and social domains of health" [1]. Still, there is a growing interest in clarifying this term as well as in identifying the factors that affect HRQoL, in order to design more effective and efficient treatments for any chronic disease [2,3].
Celiac disease (CD) is a common systemic disease, mediated by immune function, which manifests as an enteropathy of the small intestine, from exposure to gluten [4] present in wheat, barley, rye,

Materials and Methods
The study population was composed of patients residing in Spain. The inclusion criteria were being 18 years of age or older, having a medical diagnosis of CD, and being prescribed a strict lifelong GFD. These criteria were determined by answering three self-reported questions in the affirmative.
Based on our previous research [27], patients included in the study completed a 190 item questionnaire regarding sociodemographic data (age, gender, place of residency, education, occupational status), clinical data (previous symptoms and their intensity, age at diagnosis, time with symptoms before diagnosis, time since diagnosis), and GFD related issues (associated feelings, experiences, and perceptions), and scales regarding HRQoL, adherence to the GFD, and expectations of self-efficacy. Participants also answered questions regarding the perceived adoption of recommended behaviors and their perception of the risks associated with not following those recommendations.
The patients were recruited online through the regional patients' associations in Spain and through associations' social networks during September-November 2019 and invited to complete the questionnaires online using Google Forms (Google, Mountain View, CA, USA). After receiving written information about the study through the regional associations, patients signed an informed consent document. The program SPSS v21 (IBM, Armonk, NY, USA) was used to perform the statistical analysis. Absolute frequencies were used in the descriptive analysis while relative frequencies were used for the qualitative analysis. Pearson's chi-squared was used for gender in the bivariate analyses, while for quantitative variables a t-student test was used. Age group comparisons were performed by one-way ANOVA and post hoc evaluations by Scheffe's test. Regression analyses were carried out to study the relationship between the dependent variable, HRQoL, and the independent variables of adherence to the GFD, and the other variables. In another regression analysis, adherence to the GFD was considered as a dependent variable and the rest of the factors studied as independent variables. The statistical significance was set at p < 0.05. This study was approved by the Ethics Committee of Clinical Research of Aragon (CEICA) with number CIPI16/0311.

Adherence to the GFD
Adherence to the GFD was evaluated with the celiac dietary adherence test (CDAT) questionnaire [35]. This questionnaire consists of 7 questions. It is easy to apply and has good psychometric properties. It has the advantage of being correlated with serological and histological variables as well as with interviews with nutritionists. The questionnaire was validated in Spain and again presented good psychometric properties [19]. Patients have to answer on a 5-point scale. The scores are additive , with higher scores indicating lower adherence. Scores below 13 indicate excellent or very good adherence, 13-17 moderate adherence, while scores above 17 indicate fair or poor adherence [35]. We added a separate question regarding adherence, where participants were asked to rate their adherence on a 0-10 scale, considering strict adherence to GFD scores ≥ 9.
Finally, according to the CDAT cut-off point, participants were divided into a moderate/low adherence group (group 1, with scores ≥ 13) and an excellent/good adherence group (group 2, with scores < 13). We compared differences between these two groups for the variables of the study. We also divided the participants into 3 age groups to compare differences.

General and Specific Self-Efficacy
General self-efficacy was evaluated with the Spanish adaptation of the general self-efficacy scale (GSES) [36]. This scale is comprised of 10 items (e.g., "I can solve difficult problems if I make enough effort") to which the patient answers on a 4-point Likert scale (1 = not at all true, 2 = hardly true, 3 = moderately true, 4 = exactly true). The scores are additive in a range of 10 to 40, with the highest scores indicating greater self-efficacy. Although the authors do not set a cut-off point, we will consider the 70% of the theoretical rank (score ≥ 31) as indicative of high general self-efficacy [37,38]. CD specific self-efficacy was evaluated using the celiac disease self-efficacy scale (Celiac-SE) [39] designed to measure the degree of perceived self-efficacy in adherence to the GFD in patients aged 12 years or older, in different situations such as shopping, eating at home with others, traveling, eating out, or eating at work or school. Patients answered a 25-question questionnaire between 0 (I could definitely not do this) and 10 (I am completely certain I could do this). The questionnaire provides an average score for the whole scale and for each of the areas. Scores at 7 and above indicate high self-efficacy. Cronbach's alpha for the total scale is 0.81.

HRQoL
The HRQoL of patients with CD was evaluated using the Spanish version of the CD Quality of life survey (CD-Qol) [32]. This questionnaire consists of 20 questions that are answered on a 5-point Likert scale. The questions are grouped into four dimensions: limitations, dysphoria, health concerns, and inadequate treatment. The scores are additive in a range of 10 to 100. Scores of 70 or higher indicate good HRQoL.
It is important to mention that we found a discrepancy in the wording of item 8, and we communicated this issue to the authors. Since all responses were considered to be additive in this questionnaire, the correct wording of this item must be negative ("I feel that diet is not sufficient treatment for my disease"). We believe that this was the original intention, but that the word "not" was accidentally omitted in the version of the questionnaire that was published in the appendix of Dorn et al.'s 2010 validation paper. This mistake can also be found in the Spanish adaptation. This item affects the subscale "inadequate treatment".
In order to compare HRQoL with studies of other diseases, participants responded to the Spanish version of the SF 12-item Short Form Health Survey (SF-12) [40]. It is an adaptation of the SF-36 which generates a physical component summary (PCS) and a mental one (MCS). Additionally, we evaluated, using two questionnaires, the perception of risks and the perceived adoption of the behaviors recommended when dealing with the GFD. These two questionnaires were developed from the content in the patients' association guide which was elaborated by experts [41], and previously used in our research [27,42]. The first questionnaire was the celiac disease risk assessment questionnaire (CDRAQ). This was an 18-item questionnaire, scored from 0 (not risky) to 10 (very risky), according to a patient´s perceived risk with respect to certain behaviors. The scores are additive, with higher scores indicating a stronger perception of risk. Examples of these items are: "I consider it to be a risk to consume processed products without a gluten free label" or "not eliminating bulk products from diet". In the present study this questionnaire showed a Cronbach's alpha of 0.91.
The second questionnaire was the celiac disease recommended behaviors questionnaire (CDRBQ). This second questionnaire consists of 18 questions where participants must assess if they follow the recommended behaviors, rated 0 (I do not follow this recommended behavior) to 10 (I follow this recommended behavior). An example of these items is "I wash my hands after touching something with gluten". Scores are additive and higher scores indicate stronger perception of the correct adoption of the recommended behaviors. In this study this questionnaire showed a Cronbach's alpha of 0.81.
Other variables studied were the time since diagnosis, the perceived consequences of abandoning the diet, the presence or not of digestive or non-digestive symptoms, and the presence of associated diseases.

Perceptions Associated with CD
We also included the celiac disease perceptions and emotions questionnaire (CDPEQ) to evaluate the frequency of some perceptions and emotions in these patients. This questionnaire was developed by this research team from the results of qualitative studies [29,31], and used previously in our research [42] to explore if those feelings and perceptions are common among Spanish patients with CD. This is a 17-item questionnaire that had to be answered on a 5-point Likert scale (never, sometimes, average, many times, always). The items asked included how often the patient feels that they have an unwanted role in social events, feel rejected or feel ignored by others, and questions about the frequency with which the patient experiences feelings of anxiety or sadness related to the illness.

Socio-Demographic Variables
Several sociodemographic questions related to age, place of residence, marital status, work, or educational level were incorporated.

Results
A total of 818 questionnaires were collected, of which 738 were from patients with CD, 45 reported non-celiac gluten sensitivity, 14 were from minors, 7 had consent problems, and 14 were incomplete.

Description of the Sample
Of the 738 patients with CD (85.3% women), 78.2% belonged to a patient association. The age range was between 18 years and 74 years (M = 39.41; SD = 11.57). Most participants had several years of experience with the disease (M = 9.80; SD = 10.12). Table 1 below displays the main characteristics of the sample.

Adherence to GFD
In all 96.88% of the participants reported a strict adherence to GFD when asked in the separate question added to the CDAT. The CDAT results showed good adherence (M = 11.49, SD = 2.97) in Table 2. In addition, 68.7% showed good or excellent adherence, with scores below 13 while 4.6% reported poor adherence with scores above 17. We found better adherence in those above 35 years of age (p < 0.001). Gender differences were found with women showing greater adherence (p = 0.012).

Self-Efficacy
Participants showed high general self-efficacy (M = 31.34; SD = 4.84) and we found differences between the high and moderate/low self-efficacy groups (p < 0.001). In relation to specific self-efficacy, this was high (M = 8.68; SD = 1.20) and we found differences between the groups of high and moderate/low adherence, with participants with better adherence to the GFD showing higher specific self-efficacy (p < 0.001).
In a more detailed analysis (Table 3), we found that the lowest self-efficacy occurs in the area of "traveling" (M = 7.95; SD = 1.83), while the highest self-efficacy appears in the area of "eating at home with others" (M = 9.37; SD = 1.47). Here, we have also found differences between the excellent/good and moderate/low adherence groups (p < 0.01) for each of the areas. No gender differences were found for specific self-efficacy. We found significant differences between groups (p < 0.001). The self-efficacy levels in the 18-35 group were significantly lower (p < 0.001) than for the 36-50 group. No differences were found between 36-50 and the >50 groups (p = 0.069). If we set a cut-off point of seven where lower scores indicate low/moderate self-efficacy, 9.7% have low/moderate specific self-efficacy for managing a GFD. In an area analysis, 25.9% show low/moderate specific self-efficacy for "traveling", 15.5% for "eating at work or school", 11.4% for "eating out", 8.4% for "shopping", and 7.3% for "eating at home with others".

HRQoL
The results in HRQoL with the SF12v2 questionnaire were moderately high (M = 36.59; SD = 4.47) with the physical component (M = 16.29; SD = 2.02) being slightly lower than the mental one (M = 20.30; SD = 3.06) ( Table 4). If we move it to a scale of 0 (worst HRQoL) to 100 (best HRQoL) we get a 62.65 (physical), 67.67 (mental) and 65.34 for the overall scale. We find differences in the adherence groups for MCS (p < 0.001) but not for the PCS. No gender differences were found in any of the dimensions. The results relating to HRQoL measured with CDQol were high (M = 72.73; SD = 16.83). If we express the four dimensions on a scale of 0-100, "health concerns" with a 66.92% (M = 16.73; SD = 5.16) had the lowest perceived HRQoL, followed by "limitations" with 70% (M = 31.50; SD = 9.10), "inadequate treatment" with 72.70 (M = 7.27; SD = 2.28), and "dysphoria" with 86.15% (M = 17.23; SD = 3.16). Additionally, 38.8% of patients obtained a score below 70, the cut-off point above which the score indicates good HRQoL.
Significant differences in HRQol in CDQoL were found between the high and moderate/low adherence groups (p < 0.001), with higher scores for those who had better adherence to the GFD. No gender differences were found. It is the over 50 age group which shows a better HRQoL, with significant differences with the other two groups (p < 0.001), while the 18-35 group does not reach the cut-off point of 70, indicative of high HRQoL.

Perception of Risk (CDRAQ), Perception of Adoption of Recommended Behaviors (CDRBQ), Age at Diagnosis, Time Since Diagnosis, Intensity of Symptoms, and Consequences of Abandoning
The excellent/good adherence group had a higher risk perception (p < 0.038) and reported a higher follow-up of recommended behaviors (p < 0.001) ( Table 5). Differences between high and low/moderate adherence groups were found (p < 0.047) for consequences of abandoning the GFD. Paradoxically, the low adherence group perceived the consequences of abandoning the diet to be more serious. Furthermore, differences in symptom intensity after transgression were found to be higher in the moderate/low adherence group (p = 0.006). Age at diagnosis (p = 0.299) and time since diagnosis (p = 0.056) do not seem to play a major role in adherence to the GFD.  Table 6 shows the results of the linear regression to study the impact of changes in the independent variables on adherence to the GFD. After studying the main variables, we only show those that have a very clear effect: specific self-efficacy, perceived adoption of patient association recommendations, risk perception, HRQoL (when measured with CDQoL), and gender, which are all significant. The model with these five variables explains 25.9% of the variance in GFD adherence. In relation to the semi-partial correlations, the specific self-efficacy represents 6.7%, while the HRQoL represents 5.1%. This would mean that the higher the specific self-efficacy, HRQoL, and the perception of adherence, the higher the adherence to a GFD, as the perception of risks does not appear to have an effect on the adherence to a GFD. Being a woman is also a predictor of adherence. Age, age at diagnosis, time since diagnosis, intensity of symptoms, risk perception, or belief in the serious consequences of abandonment were not predictive of adherence.

Relationship between the Variables, GFD and HRQoL
If we analyze HRQoL as a dependent variable (Table 7), this is explained by specific self-efficacy, adherence to the GFD, time since diagnosis, age and risk perception, with the perceived adoption of the recommended behaviors not being significant. In this model, these variables explain 30.6% of the HRQoL score. As far as semi-partial correlations are concerned, the specific self-efficacy variable explains 6.2% of the variability in HRQoL, 5.5% the time since diagnosis, while adherence to the GFD, explains 4% of the variability, while risk and age, although significant, have a negligible influence, 2.6% and 1.5%, respectively. No differences based on gender are found, neither for the questionnaire as a whole nor for each of the scales.

Perceptions and Emotions Associated with Having CD, According to CDPEQ
Lastly, participants were also asked about their perceptions and emotions associated with having CD ( Table 8). The most common perceptions were of a limited gluten-free offering (68.6%), the feeling of having to be constantly alert (60.7%), or an excessive protagonism in social events (55.9%). Also noteworthy is the perception that they have to work twice as hard because of their illness (shopping, cooking, etc.). It is also common to have the feeling that they are forgotten about at social events (31.2%). With regard to the most frequent negative emotions, anger (25.2%), envy (21.4), sadness (17.2%), anxiety (14.2%), or fear (13.9%) stand out, but it is worth noting the high frequency of positive feelings associated with the disease such as pride and self-confidence (50.5%), or joy (30.9%).

Discussion
This research shows that specific self-efficacy, adherence to the GFD as well as risk perception, time since diagnosis, and age play an important role in the HRQoL of patients with CD. Results show that it is self-efficacy, when specifically measured, that best predicts adherence to the GFD and the resulting HRQoL [27,39].

Adherence to GFD
Adherence to a GFD was high in this study, with 68.7% of participants showing good or excellent adherence according to CDAT criteria [35], but somewhat lower than in previous research [27] conducted in Spain. This high adherence may be due to the fact that a large part of the sample (78.2%) is associated with a patient association [16,23,43] and, therefore, has a high level of knowledge about the disease with almost 10 years of average experience of the GFD. In our study, this adherence to a GFD is determined by specific self-efficacy, HRQoL (measured with CDQoL) and perceived adoption of recommended behaviors, with women showing better adherence than men. Patients above 35 years old also show a better adherence than younger individuals.
We should note that it is self-efficacy measured in a specific way, rather than general self-efficacy, that plays a predictive role in adherence to the GFD [27,44]. Risk perception, although not a significant predictor, is included in the model because of the role it can play in the development of self-efficacy in adherence to risk, as people with high self-efficacy perceive risks as challenges, while people with low self-efficacy may perceive them as overwhelming threats [28].
No significant relationships were found with time-related variables such as age, age at diagnosis, or time since diagnosis [27], despite the predictive value that other studies place on age [24,45] or age at diagnosis [45].
Determining the degree of adherence to the GFD is complex, varies from study to study [16], and often involves subjective responses to self-reported questions (in our study, 96.88% reported strict adherence, well above the 68.7% indicated by CDAT). This subjective perception of adherence may be conditioned by their knowledge of the disease, its consequences, and an adequate or inadequate risk assessment. It is clear that there is a need to develop new tools to assess adherence as current histological and serological methodologies are expensive and invasive and not very sensitive to the detection of transgressions. Evaluations by expert dietitians [46,47] and the detection of immunogenic peptides [48,49] have been shown to be effective in the follow-up of patients with CD. Even so, there is a need for the development of new questionnaires for the assessment of adherence to GFD that take into account variables such as knowledge about the disease and the behaviors that determine adherence, the expectation of specific self-efficacy, and risk assessment.

Self-Efficacy
General self-efficacy levels were significantly higher compared to other studies [27,[50][51][52]. Patients with CD may develop these high levels because they have to manage the disease and its treatment for years, unlike the case of other diseases where that responsibility lies more with the doctor [53]. We also think that this responsibility in managing one's own disease and the GFD can contribute to this general feeling of self-efficacy. The same occurs with the level of specific self-efficacy which is high, and it is comparable to other studies recently carried out in Spain [27]. Considering regression analysis, it seems that specific rather than general self-efficacy plays a role in predicting adherence to a GFD, which is in line with previous research [27], and which emphasizes the need to measure control beliefs more specifically rather than in a general way. The results in our study support the idea, as in other studies [27,29,39], that the areas in which people with CD are less confident are "traveling," "shopping," and "eating out".
Neither in specific self-efficacy, nor in general, were differences found based on gender, time on a GFD, or year of diagnosis, which would emphasize the idea that it is more the type of situation and its management than other personal characteristics (such as experience with the disease) which explain the development of self-efficacy beliefs [28]. In a more detailed analysis by age, it is above 35 years old where we find a higher sense of specific self-efficacy. The fact that differences are found between groups, with different levels of adherence between the different areas evaluated by the Celiac-SE, reinforces the idea of the importance of the situation, and that we must have instruments that allow us to evaluate self-efficacy, and that it is useful to do so for each of the areas where difficulties may appear in the management of the GFD. As such, in a newly diagnosed patient, we can determine their levels of self-efficacy in "traveling" or "eating at work" which may well be different, and need different types of support to function in one area or other. This evaluation of specific self-efficacy with respect to each area of CD management will allow us to design tailored interventions in the clinical setting to improve adherence. Celiac-SE is a useful instrument for this purpose.

HRQoL
The results of this study show good HRQoL, although 38.8% of the participants showed a low HRQoL below the cut-off point of 70 that is established as indicative of a good HRQoL [32]. It is "health concerns" that have the lowest score as in previous studies [27,32]. These studies disagree with one recently published, with a very large sample (n = 1230), where the overall score was significantly lower (56.3 ± 18.27) but where, as in our study, "health concerns" [34] showed one of the lowest scores. According to our model, HRQoL is better when specific self-efficacy expectations are higher, the adherence is better, the risk perception is lower, and the age and time from diagnosis is higher. It is self-efficacy, again measured in a specific, rather than a general way, that best explains, of these factors, the HRQoL score [27]. Good adherence, among other factors due to the elimination of symptoms and health problems, is a good predictor of HRQoL [18,54]. A low perception of the risks associated with the disease, coupled with high self-efficacy [24,44], predict a better HRQoL. Finally, a broader experience with the disease and with the GFD can also lead to a better HRQoL [34]. Age is significant for the questionnaire as a whole and for each of the subscales, but with low correlations, while recent studies in Spain do not find this relationship except for the "health problems" scale [34]. In our study it is participants above 50 years old that show a better HRQoL.
In terms of the associated feelings and emotions, it seems that patients with CD have a feeling of having to be alert, constantly watching their diet, and that it is uncomfortable for them to take an unwanted prominence in celebrations involving food. These and other circumstances often generate feelings of anger or envy in one-fifth of them. Feelings of anxiety or fear, which are more directly mediated by their high expectations of self-efficacy, are more subdued, the latter not being as frequent. These feelings are in line with other research [29][30][31]34,42]. Remarkably, we did identify in this study feelings of pride or joy associated with the disease quite often which may indicate a correct acceptance of the diagnosis. This may be explained by the fact that the patients have a lot of experience with GFD and that most of them belong to patient associations. Finally, 43.2% of the participants reported associated diseases which may have affected their HRQoL.
This study has several limitations: first, participants were recruited mainly through patient associations so there may be an under-representation of patients with CD who are not associates. Second, although an interview with an expert dietician is usually required to become an associate, this is not always the case and participants were included by means of questions in which they stated that they had a firm diagnosis of CD and that they were, therefore, prescribed a lifelong GFD. On the other hand, although the sample is large, it was not possible to obtain a sufficiently representative sample of the Spanish population. Moreover, we must be cautious when interpreting gender differences as 85.3% of the respondents were women and this may affect the analysis. Finally, as it is a long survey, this could have influenced the number of questionnaires obtained. As strengths of this study we must mention that this research explores the HRQoL in patients with CD in Spain, their feelings and expectations with a substantial sample and highlights the variables to be included in programs for the management of the disease by patients with CD. This type of study is not frequent in Spain. Secondly, we found the discrepancy in item eight of the English version of the CD-Qol questionnaire and, at least, the Spanish version. The Italian version seems to be correct. We have communicated this issue to the authors, who have responded that they are going to adopt the necessary measures to solve this problem. This questionnaire has been widely used in research in CD [27,32,34,42,55,56].
In recent years, we have a clearer picture of what the characteristics are of patients with CD in Spain. The various studies indicate that they are mainly women, diagnosed in the first or fourth decade, belonging to patient associations, with high specific self-efficacy, good adherence to the GFD, and a good HRQoL. Even so, given that the only treatment to date for this disease is strict adherence to the GFD, specific programs must be developed in primary care and in patient associations themselves to care for these patients. In order to improve HRQoL in patients with CD, these programs should include, not only information on the disease, on the GFD and its risks but also an assessment of specific self-efficacy for each of the identified areas (shopping, traveling, eating out and at home with others and at work) and strategies to develop a strong sense of self-efficacy to manage the disease in these situations.

Conclusions
In Spain, people with CD have a good adherence to the GFD and a good HRQoL. The expectation of self-efficacy, measured in a specific rather than a general way, is the best predictor of both adherence and subsequent HRQoL. New GFD adherence programs for patients with CD must assess and promote the expectation of specific self-efficacy for each of the areas in which these patients have to achieve adherence. Funding: This research has been funded by project PI16/00570 (Instituto de Salud Carlos III) and by Fondo Europeo de Desarrollo Regional (FEDER) funds: "Una manera de hacer Europa".