Construction of an Explanatory Model for Quality of Life in Outpatients with Ulcerative Colitis

Ulcerative colitis is a diffuse nonspecific inflammation of unknown causes that forms erosions and ulcers in the large intestinal mucosa and submucosa. The number of patients with ulcerative colitis in Japan has increased 10-fold over the past 25 years [1]. Ongoing symptoms include abdominal pain and frequent defecation. There is no curative treatment for ulcerative colitis, and patients are treated with various drugs, such as 5-aminosalicylic acid, steroid preparations, immunomodulators, and biologics.

Previous studies have shown that the health-related quality of life (HRQoL) of patients with ulcerative colitis is considerably lower than that of the general population [2, 4]. In addition, factors related to the quality of life of these patients are age at onset [5], age [6], sex [4, 6, 8], disease activity [2, 3, 7, 9, 14], disease type [3, 13], disease duration [15], anxiety [4, 12, 15], depression [2, 4, 12, 14, 15], psychological stress [8, 12], occupation [5, 9, 16], workplace support [17], and social life [18, 19]. Fu et al. [20] examined the association between disease activity, social support, and HRQoL in patients with inflammatory bowel disease (IBD). They found that disease activity, anxiety, and depression decreased HRQoL, whereas social support increased HRQoL. In addition, anxiety and depression mediated the effects of disease activity and social support on HRQoL. However, findings in patients with IBD may not be applicable to patients with ulcerative colitis because of differences in pathophysiology, symptoms, and treatment. Therefore, identifying the factors related to HRQoL in patients with ulcerative colitis in Japan is necessary.

In recent years, various new medications have enabled remission and the maintenance of remission, and the treatment of ulcerative colitis has become centered on outpatient management. However, even during the remission period, daily medication and regular medical examinations are required. Moreover, patients are concerned about relapse because ulcerative colitis is intractable and cannot be cured completely [21]. Therefore, this study aimed to examine HRQoL and its related factors in outpatients with ulcerative colitis and to construct an explanatory model for HRQoL.

Figure 1 shows the predictive model for HRQoL in outpatients with ulcerative colitis used in this study. From previous studies, we hypothesized that HRQoL in patients with ulcerative colitis will be influenced by demographic, physical, psychological, and social factors.

Demographic factors include age, sex, and age at onset. Previous studies have shown that women and younger patients have lower HRQoL scores [4, 6, 7]. A considerable difference in HRQoL scores has also been suggested for the age at onset [5]. Physical factors include the disease activity, disease type, duration of disease, and presence or absence of treatment side effects. Disease activity is inversely associated with HRQoL [2, 3, 7, 9, 15, 20], whereby patients with total colitis have significantly lower HRQoL scores than patients with proctitis [3, 13]. In patients in remission, the disease duration is also inversely associated with HRQoL [13, 15]. In addition, even nonserious treatment side effects, such as headache, itching, and myalgia [10, 22], can reduce HRQoL. Psychological factors include anxiety and depression, which are negatively correlated with HRQoL [2, 4, 12, 14, 15, 20]. Therefore, because physical and social factors influence HRQoL, we considered that HRQoL is also mediated by psychological factors [20]. Social factors include the employment status, the cohabitation/marital status, workplace support, a medical beneficiary certificate for designated intractable diseases, and social support. Employment and support from others have previously been shown to be associated with HRQoL [5, 16, 17].

Quality of life is defined as an all-inclusive concept incorporating factors, such as physical, material, social, and emotional well-being that affect an individual’s life. HRQoL is a concept that includes only factors that are part of an individual’s health [23]. In this study, we defined HRQoL in patients with ulcerative colitis as an “individual’s subjective perception of the quality and satisfaction of life to live in their own way, even with ulcerative colitis, which is influenced by physical, psychological, social, and demographic factors.”

The study used a hypothesis-testing design on the basis of a cross-sectional survey. We recruited patients with ulcerative colitis aged >20 years who were outpatients at A Clinic in the Kansai region of Japan and consented to be included in the study between December 2019 and July 2020. A Clinic is a medical institution that specializes in the outpatient treatment of IBD, and approximately 600 patients with ulcerative colitis attend the clinic. Patients who were judged by a gastroenterologist as unfit for treatment at A Clinic and those who had insufficient understanding of Japanese were excluded from the study.

Prior to commencing the survey, announcements were posted on the homepage of A Clinic and in the outpatient waiting room disclosing the purpose and methods of the study. To recruit participants, the first author individually explained the purpose and methods of the study in writing and verbally to patients with ulcerative colitis who visited A Clinic during the survey period and met the inclusion criteria, using the waiting time for outpatient treatment.

The sample size was calculated using G*Power 3.1 (Franz Faul, University Kiel, Germany). The required sample was estimated as 146 to provide f² = 0.15 with an alpha = 0.05, a beta = 0.2, and 17 predictors. We used f² = 0.15 (effect size: medium) because previous studies did not report effect sizes. In a similar survey that we performed previously, approximately 30% of the participants returned invalid responses to the questionnaire. Therefore, we estimated that the required sample size would be 250.

The dependent variable in this study was HRQoL, which was measured using the 32-item Inflammatory Bowel Disease Questionnaire (IBDQ-32). This scale is most often used to measure and evaluate HRQoL in patients with IBD in Japan and other countries. We used the IBDQ-32 in this study because we wanted to compare our findings with those of previous studies. The IBDQ-32 was developed at McMaster University in 1989, and the reliability and validity of this scale have been verified [24]. The IBDQ-32 comprises 32 questions (10 on abdominal symptoms, 5 on systemic symptoms, 12 on emotions, and 5 on social activities). Items are scored on a 7-point Likert scale. The total score ranges from 32 to 224, with a score of ≥170 indicating clinical remission [17]. The reliability and validity of the Japanese version of the IBDQ-32 were verified by Hashimoto and colleagues in 2003 [25], and this version has since been used in numerous studies conducted in Japan. We obtained a license from McMaster University to use the scale and develop our model.

Age, sex, and age at onset were obtained from medical records.

The disease activity was evaluated by calculating the partial Mayo score. The disease type and duration of disease were obtained from medical records. The presence or absence of treatment side effects was determined by the responses to the questionnaire.

The perception of anxiety and depression was evaluated using the 14-item Hospital Anxiety and Depression Scale (HADS) [26]. This scale is the most commonly used scale to assess anxiety and depression in patients with IBD, and the reliability and validity of the Japanese version have been verified by Higashi and colleagues [27]. The HADS comprises 14 items, which are divided into subscales for HADS-Anxiety (HADS-A) and HADS-Depression (HADS-D). Each subscale comprises seven items that are scored on a 4-point Likert scale, ranging from 0 to 3 points. A total score of 0–7 points indicates no anxiety or depression, 8–10 points suspected anxiety or depression, and ≥11 points confirmed anxiety or depression.

Options for employment status were self-employed, professional/technical, office work, service industry, student, unemployed, homemaker, pensioner, and others. The presence or absence of a spouse/cohabitant and workplace support were determined by the responses to the questionnaire. Social support was evaluated using the five-item questionnaire created by Muraoka and colleagues [28], which included the availability of (1) a consultation partner when needed, (2) an advisor during difficult times, (3) someone to help with daily life (e.g., to carry out housework), (4) someone who would take the patient to the hospital when needed, and (5) someone to take care of the patient when ill. In addition, information was collected from electronic medical records regarding the availability of subsidies for a medical beneficiary certificate for designated intractable diseases.

We first evaluated the normality of each variable using the Shapiro-Wilk test. To construct a predictive model, we tested the correlation between the IBDQ-32 total score and the explanatory variables of the ordinal scale. Items with a correlation coefficient of <0.20 were excluded as explanatory variables. The relationship between the IBDQ-32 total score and the explanatory variables of the nominal scale was examined using the Mann-Whitney or Kruskal-Wallis test, and items with a significance level <5% were included as explanatory variables. A multiple regression analysis was performed on the remaining explanatory variables using the IBDQ-32 total score as the dependent variable. IBM SPSS version 28 (IBM Corp., Armonk, NY, USA) was used for the analysis. The HRQoL prediction model shown in Figure 1 was then used to perform path analysis with the variables selected from the results of the multiple regression analysis as the explanatory variables and the IBDQ-32 total score as the dependent variable. Path analysis was performed using Amos version 28 (IBM SPSS). The causal relationships among variables were determined by the path coefficient, and the indirect, direct, and total effects on HRQOL were calculated. The goodness-of-fit index (GFI), adjusted GFI, comparative fit index, root mean square error of approximation, and Akaike information criterion were used to evaluate the goodness-of-fit of the model.

The questionnaire was distributed to 287 patients, and 285 responded (response rate: 99.3%). Only responses corresponding to the IBDQ-32 total score aggregation method were regarded as valid. If one or more items of the HADS and social support questionnaires were missing responses, the respondent was excluded from the analysis. Valid responses were judged on the basis of these criteria, and 203 participants (valid response rate: 71.2%) were included in the analysis (Table 1).

The IBDQ-32 total score did not follow a normal distribution (p < 0.01). The median IBDQ-32 total score was 185, and the interquartile range (first to third quartile) was 166–202 (Table 2).

The explanatory variables that had a correlation of 0.40–0.70 with the IBDQ-32 total score were the partial Mayo score (r = −0.451), HADS-A score (r = −0.678), and HADS-D score (r = −0.528) (Table 3). The relationships between the IBDQ-32 total score and the explanatory variables of the nominal scale were examined using the Mann-Whitney or Kruskal-Wallis test (Table 4). The explanatory variables that showed a significance level <5% were the presence or absence of treatment side effects (p = 0.004), the availability of a consultation partner when needed (p = 0.010), the availability of an advisor during difficult times (p = 0.001), and the availability of someone who would take the patient to the hospital when needed (p = 0.041).

To identify the explanatory variables that independently affected HRQoL, we conducted a multiple regression analysis (forced input method) with the following seven variables: the partial Mayo score, presence or absence of treatment side effects, HADS-A score, HADS-D score, availability of a consultation partner when needed, availability of an advisor during difficult times, and availability of someone who would take the patient to the hospital when needed. With regard to the four nominal variables, “yes” and “no” responses were replaced with “1” and “0,” respectively.

In model 1, six variables, namely, the HADS-A score, partial Mayo score, presence or absence of treatment side effects, availability of a consultation partner when needed, availability of an advisor during difficult times, and availability of someone who would take the patient to the hospital when needed, were entered into the multiple regression analysis (adjusted R² = 0.594). Model 2 included the variables of the HADS-D score, partial Mayo score, presence or absence of treatment side effects, availability of a consultation partner when needed, availability of an advisor during difficult times, and availability of someone who would take the patient to the hospital when needed (adjusted R² = 0.384). Model 1 had a higher prediction accuracy than model 2. Therefore, to build a more accurate model, we used model 1 as the axis, changed the combination of the three social factors in models 3–8, and repeated the multiple regression analysis (Table 5). Model 7 showed the best fit and was thus selected (adjusted R² = 0.597).

We then constructed an HRQoL explanation model, which is shown in Figure 2, using a path analysis model with the IBDQ-32 total score as the dependent variable. In line with the predictive model, the partial Mayo score, availability of an advisor during difficult times, and presence or absence of treatment side effects were used as exogenous variables. Because the correlation between the partial Mayo score and the availability of an advisor during difficult times (r = 0.062), and that between the presence or absence of treatment side effects and the availability of an advisor during difficult times were not significant (r = −0.025), we did not set a covariation between these variables. Because we found a correlation between the partial Mayo score and the presence or absence of treatment side effects (r = −0.139), covariability was set. Using the IBDQ-32 total score and the HADS-A score as endogenous variables, we drew a path diagram showing the causal relationships among the variables.

Path analysis showed that the explanatory variable with the greatest direct effect on the IBDQ-32 total score was the HADS-A score (β = −0.586), followed by the partial Mayo score (β = −0.373), presence or absence of treatment side effects (β = 0.121), and availability of an advisor during difficult times (β = −0.101). The social factor indirectly affected the IBDQ-32 total score via the HADS-A score (β = 0.190×−0.586 = −0.111). The contribution rate to the IBDQ-32 total score in this model was R² = 0.600. This model confirmed the goodness-of-fit of the model with a GFI = 0.998, adjusted GFI = 0.989, comparative fit index = 1.000, root mean square error of approximation = 0.000, Akaike information criterion = 26.761, χ² = 0.761, and df = 2 (p = 0.684) (Fig. 2).

We selected four explanatory variables that affected the IBDQ-32 total score, namely, the HADS-A score, partial Mayo score, presence or absence of treatment side effects, and availability of an advisor during difficult times. Previous studies are in line with our findings, whereby the disease activity index [2, 3, 7, 9, 15, 20], social support [16, 17, 20], and HADS score [20] affected HRQoL. Additionally, we found that the presence or absence of treatment side effects affected HRQoL. The most frequently used treatment medication among the respondents was 5-aminosalicylic acid (98.5%), and it was the most common (36.2%) medication that caused their side effects. We considered that 84.2% of the respondents were in remission on the basis of the partial Mayo score, and their experiences of side effects by 5-aminosalicylic acid used to maintain remission could affect the HRQoL. The side effects associated with 5-aminosalicylic acid generally include digestive tract symptoms (e.g., flatulence, abdominal pain, nausea, and diarrhea), headache, rash, and worsening ulcerative colitis [29]. Experiencing such side effects may affect the HRQoL of patients with ulcerative colitis.

The HADS-A score had the largest direct effect on the IBDQ-32 total score. Patients with ulcerative colitis have uncertainty regarding the next relapse and anxiety regarding developing colorectal cancer, even if their condition is stable [30, 31]. Moreover, these patients have anxiety about unemployment because of their ongoing and unpredictable condition [32]. In addition, patients with ulcerative colitis are prone to mental stress from the burden of regular medical examinations, lifetime medication, and the need to make adjustments to their daily life [21]. We postulated that anxiety resulting from such mental stress affects HRQoL. The anxiety and concerns of patients vary between individuals and can change throughout life [33]. Therefore, nurses should consider patients’ concerns, even during the limited time of outpatient visits, and provide information tailored to each patient. Specifically, nurses should listen to the anxiety of individual patients. When a patient has anxiety about relapse, nurses should provide the patient with information on the specific signs of relapse, how to self-monitor them, and guidelines for judgment to see a doctor. Furthermore, physicians need to be aware of relevant resources and collaborate with professionals from other disciplines, such as psychological counselors, to help patients maintain their mental health.

We also found that the direct effect of the partial Mayo score and the presence or absence of treatment side effects on the IBDQ-32 total score was greater than way the indirect effect of the partial Mayo score and the presence or absence of treatment side effects on the IBDQ-32 total score via the HADS-A score. This finding indicates that controlling disease activity and monitoring treatment side effects are crucial for maintaining or improving the HRQoL of patients with ulcerative colitis. Nurses should carefully evaluate patients for changes in symptoms during outpatient visits and educate them regarding self-management of symptoms. Because small changes in a patient’s life may be signs of side effects, careful observation and assessments are vital.

In addition, nurses should obtain blood sampling data and endoscopy results, assess the therapeutic effect of treatments, and determine the presence or absence of side effects. Moreover, they should ensure that the patient has a clear understanding of their medications, provide an opportunity to have their medications explained by a physician if necessary, and use appropriate language to describe the effects and side effects. Watanabe et al. [34] reported that 66% of patients with ulcerative colitis in Japan wished that they had talked more about their fears of medical treatments, 53% worried that if they ask too many questions, they would be considered a difficult patient, and 51% said their physician rarely had time to address all of their concerns. Based on these results, physicians should not only explain about side effects, but also carefully check the patients’ understanding and provide opportunities for patients to express their fears about side effects without hesitation. Additionally, pharmacists should provide explanations and check patients’ understanding in accordance with individual patients’ medication based on physicians’ treatment policies.

We found that the availability of an advisor during difficult times indirectly affected the IBDQ-32 total score via the HADS-A score, and this indirect effect was at the same level as the direct effect. Fu and colleagues [20] did not find a direct effect of social support on HRQoL, but did observe an indirect effect via the HADS score. Social support may reduce anxiety in patients with ulcerative colitis, which results in an improvement of their HRQoL. Indeed, we found that the availability of an advisor during difficult times was related to HRQoL. Because many patients with ulcerative colitis can independently carry out basic activities of daily living, the availability of an individual who supports patients with housework or during illness was not a factor related to HRQoL. Tomita and colleagues [16] reported that such social support adds vitality to patients and reduces daily burdens. A previous study showed that many Japanese people consider that receiving support from others is a nuisance and feel apologetic about it [35]. Therefore, Japanese patients with ulcerative colitis may be reluctant to discuss their problems with others even if they are having trouble. The availability of an advisor during difficult times reduces isolation and anxiety and depressive symptoms, thereby improving HRQoL. Therefore, nurses should collect information on patients’ social support and arrange a support system if necessary.

In Europe and the USA, nurses can stabilize disease activity and improve HRQoL by responding to patients’ concerns via telephone or e-mail [36, 37]. Therefore, nurses should establish a support system using communication tools for patient consultations that are suitable for each patient, rather than simply waiting for outpatient visits.

The response options for treatment side effects and social support were only present or absent. However, a more detailed analysis of factors affecting HRQoL would be possible if responses were obtained at different levels or stages using the Likert scale. In addition, we did not examine the respondents’ perspectives and concerns regarding the efficacy or adverse events of the medication in this study. Fukuda et al. [38] showed that patients’ perceptions of efficacy and adverse events differed depending on the therapeutic agent. Future studies should examine patients’ perspectives and concerns of the medication and identify factors important to patients’ HRQoL.

Although the HADS-A score had the strongest influence on HRQoL, the contribution rate was low, which indicated that the anxiety of patients with ulcerative colitis was affected by factors other than the variables explored in this study. Future work should investigate factors related to anxiety in greater depth to improve the HRQoL of outpatients with ulcerative colitis.

We identified factors affecting the HRQoL of outpatients with ulcerative colitis and constructed an HRQoL explanatory model. The explanatory variables that were significantly associated with the IBDQ-32 total score, as an indicator of HRQoL, were the partial Mayo score of disease activity, presence or absence of treatment side effects, HADS-A score (a measure of anxiety), and availability of an advisor during difficult times. The HADS-A score exerted the strongest direct negative effect on the IBDQ-32 total score. The direct effect of the partial Mayo score and the presence or absence of treatment side effects on HRQoL were greater than the indirect effect of the partial Mayo score and the presence or absence of treatment side effects on HRQoL via the HADS-A score. In addition, the indirect effect of the availability of an advisor during difficult times on HRQoL via the HADS-A score was almost the same as the direct effect of the availability of an advisor during difficult times on HRQoL.

The authors are grateful to all participants for their cooperation. The authors also thank Sarina Iwabuchi, PhD, from Edanz (https://jp.edanz.com/ac), for editing a draft of this manuscript.

This study protocol was reviewed and approved by Aoyama Medicine Clinic Ethics Review Committee at Aoyama Medicine Clinic (approval number: 2019–5) and Mukogawa Women’s University Junior College Research Ethics Committee at Mukogawa Women’s University (approval number 19–74). The study purpose and methods were described in detail to the participants in writing and verbally. Informed consent to participate was not directly obtained but inferred by completion of the questionnaire. We added the study code number and a column at the beginning of the questionnaire to allow participants to state “I agree and cooperate” or “I do not agree or cooperate.” A check mark for “I agree and cooperate” was considered consent when the questionnaire was completed and deposited in the collection box.

The authors have no conflicts of interest to declare.

This work was supported by Mukogawa Women’s University.

M.T. and M.N. were responsible for the study design and conception of the article and performed data analysis. M.T. wrote the manuscript, and M.N. and N.A. revised the manuscript. M.T. and N.A. were responsible for recruitment and data collection. All authors read and approved the final version of the manuscript.

All data generated or analyzed during this study are included in this article. The data that support the findings of this study are available from the corresponding author upon reasonable request.