Developing the Patient Healthy Questionnaire-8 with a Greater Impact on Quality of Life of Patients with Functional Dyspepsia Compared with the Somatic Symptom Scale-8

Background: To develop the Patient Healthy Questionnaire-8 (PHQ-8), as a more reliable approach compared with the Somatic Symptom Scale-8 (SSS-8), evaluating somatization in patients with functional dyspepsia (FD), in which the effects of somatization assessed by these two approaches on quality of life (QoL) of FD patients were further compared. Methods: Herein, 612 FD patients completed a questionnaire involving 25 items. 8 out of 25 items were selected to develop the PHQ-8 by four methods of discrete degree, correlation coecient, factor analysis, and Cronbach’s α coecient. Reliability and validity of the PHQ-8 and the SSS-8 were compared by principal component and conrmatory factor analyses. The effects of somatization, depression, and anxiety on the Nepean Dyspepsia Index (NDI) for QoL were explored by Pearson’s correlation coecient and linear regression analysis. Results: The Cronbach’s α coecient for the PHQ-8 and the SSS-8 was 0.601 and 0.553, respectively, and the cumulative contribution rate of three extracted factors for the developed PHQ-8 and SSS-8 was 55.103% and 51.666%, respectively. Somatization evaluated by the PHQ-8 (r=0.309, P<0.001) and the SSS-8 (r=0.281, P<0.001) was found to be correlated with the NDI. The model used for the PHQ-8 showed that the values of chi-squared (χ 2 ), root mean square residual (RMR), root mean square error of approximation (RMSEA), goodness-of-t index (GFI), and adjusted GFI (AGFI) were 31.247, 0.01, 0.042, 0.984, and 0.967, respectively. Linear regression analysis unveiled that somatization (β=0.270, P<0.001), anxiety (β=0.163, P<0.001), and depression (β=0.136, P=0.003) assessed by the PHQ-8


FD patients with common somatization symptoms
In addition, FD patients often have dizziness, back pain, sleep disorders, fatigue [26], and other symptoms of digestive system that cannot be explained by biochemical and structural abnormalities. Clinically, these symptoms are called somatization symptoms [27,28]. Somatization is de ned as a chronic mental disorder characterized by the presence of one or more frequently changing somatic symptoms, involving multiple systems and organs of the body [29]. Those symptoms often induce patients' incorrect understanding or excessive attention, imposing a huge economic burden to the society [30]. Somatization can coexist with other medical disorders, such as anxiety and depression, and it often makes other diseases more complex and changeable [31]. Additionally, it affects the severity of dyspepsia and the quality of life (QoL) of FD patients [32,33]. Somatization plays a more signi cant role in dyspepsia symptom severity (DSS) compared with gastric sensitivity, anxiety, and depression in FD patients [21].
Somatization is an independent risk factor for impaired QoL of FD patients, and a 5-year follow-up study demonstrated that proximal gastric accommodation, gastric emptying, and H. pylori infection were not found as risk factors [33]. Therefore, assessment of somatization is highly essential for studying FD patients.

Limitations of questionnaires for somatization
The symptoms of somatization disorder were widely assessed by the Patient Health Questionnaire-15 (PHQ-15) developed by Kroenke et al. [34]. However, the PHQ-15 includes a number of items overlapped with gastrointestinal symptoms in FD patients. Therefore, symptoms of somatization disorder have been investigated by the Patient Healthy Questionnaire-12 (PHQ-12) with the removal of three gastrointestinal symptoms, while both PHQ-12 and PHQ-15 showed to have dysmenorrhea or other items related to menstrual discomfort, making the gender score different [34][35][36].
On the other hand, symptoms of somatization disorder were evaluated by the Somatic Symptom Scale-8 (SSS-8) with satisfactory reliability and validity developed in the eld of the Diagnostic and Statistical Manual of Mental Disorders (Fifth Edition) (DSM-5) [37][38][39]. Therefore, it was previously recommended to be a replacement for the PHQ-15 [40]. However, the SSS-8 is not a FD-speci c scale for evaluating symptoms of somatization disorder, in which the relevant items may overlap with gastrointestinal symptoms. Subsequently, we developed the Patient Healthy Questionnaire-8 (PHQ-8) for assessing FD patients with symptoms of somatization disorder by adapting the SSS-8, and con rmed its reliability and validity. It was unveiled that the developed PHQ-8 had a greater in uence on QoL of FD patients than the SSS-8.

Patients
From June 2017 to June 2018, patients who were suspected of having FD were admitted to the department of gastroenterology of our hospital. According to CONSORT guidelines, inclusion criteria were as follows: (1) patients who met Rome IV criteria [1], involving to have upper abdominal pain, a burning feeling in the upper stomach, postprandial fullness, early satiety of one or more symptoms, and no organic diseases' evidence related to the above-mentioned symptoms; existence of these symptoms for at least 6 months, as well as satisfying the Rome IV criteria for at least 3 months; (2) patients who aged ≥ 18 years old; and (3) patients whose level of education was primary school and above, and had a certain reading comprehension ability. Exclusion criteria were as follows: (1) patients with organic gastrointestinal diseases, including erosive esophagitis, Barrett's esophagus, etc., which were diagnosed by gastroscopy; (2) patients with systemic diseases, metabolic diseases, or malignant tumors diagnosed by B-scan ultrasound, blood routine and biochemical examinations; (3) patients with irritable bowel syndrome; (4) patients with gastroesophageal re ux disease with symptoms of acidic taste in the mouth, regurgitation, and heartburn; (5) having a history of abdominal surgery, such as cholecystectomy, intestinal resection, hysterectomy, or appendectomy; (6) patients who had recently taken anticholinergic drugs, antispasmodic and analgesic drugs, hormones, nonsteroidal and anti-in ammatory drugs, etc.; patients whose upper abdominal symptoms disappeared after eradicating H. pylori infection; (7) pregnant and lactating women; or (8) patients with psychosis and serious somatic diseases (e.g., anxiety and depression). The sample size was 395 patients in a developed IBS-speci c somatization scale [27]. A minimum sample size of 300 patients achieved, ranging from a 95% con dence interval to equally split chance (50/50) according to the previously presented formula, which was 350 cases suggested by Creswell [41]. We used a sample size of above 400 patients. In developing the PHQ-8, FD patients accompanied with anxiety and depression were excluded, in order to avoid the interaction effects of anxiety, depression, and somatization disorders. However, previous studies showed that anxiety, depression, and somatization had mutual effects [21,33]. Therefore, in the clinical application of the developed PHQ-8, FD patients with anxiety and depression could be included.  [35], in which each item was scored by three-level as 0 ("not bothered at all"), 1 ("slightly bothered"), or 2 ("remarkably bothered"). The items of the developed PHQ-8 were completed by discrete degree method, correlation coe cient method, factor analysis method, and Cronbach's alpha.
Items were selected by the four methods for the development of the PHQ-8 Discrete degree method: As low degree of dispersion of the selected item revealed poor ability of evaluation, thus, the items with high degree of dispersion were selected, and the degree of dispersion was measured by the standard deviation (SD) of scores of items, in which exclusion of the SD of the items was de ned as < 0.5 [45].
Correlation coe cient method: An item with a correlation coe cient > 0.3 was selected according to Cohen's criteria [39].
Factor analysis with the principal component method: The common factor was extracted according to the feature root > 1, and the variance was maximized by orthogonal rotation to select an item with a factor load > 0.4. The Kaiser-Meyer-Olkin (KMO) and Bartlett's spherical tests were performed. The value of KMO < 0.5 is unsuitable for factor analysis, and in Bartlett's spherical test, P<0.01 can negate the zero hypothesis that the correlation matrix is a unit matrix, that is, there is a signi cant correlation between the variables [46].
Cronbach's α coe cient: The items were screened for internal consistency, and the Cronbach's α coe cient of the initial total scale was calculated. If the Cronbach's α coe cient increased after the deletion of an item, it can be concluded that the existence of an item may reduce the internal consistency [47].

DSS
The DSS score was calculated as sum of all eight symptoms: postprandial fullness, early satiation, epigastric pain, burning, bloating, belching, nausea, and vomiting by a 3-point scale (absent, mild, moderate, severe) [33] . The relationship among the DSS, the PHQ-8, and the SSS-8 was evaluated, and the effects of somatization, anxiety, and depression disorders on DSS were analyzed.

Measurement of disease-speci c QoL
The Short-Form Nepean Dyspepsia Index (SF-NDI) was used to measure FD patients' QoL: inference with work/study, tension, inference with daily activities, disputation to daily eating/drinking, knowledge toward/control over disease symptoms over the past 2 weeks. Each item was assessed with a 5-point scale from 0 (not at all), 1 (slightly), 2 (moderately), 3 (remarkably) to 4 (extremely) [44]. The relationships among the SF-NDI, the PHQ-8, and the SSS-8 as well as the effects of somatization, depression and anxiety disorders on FD patients' QoL were assessed.
Statistical analysis SPSS 24.0 software (IBM, Armonk, NY, USA) was utilized to carry out statistical analysis. Quantitative data were expressed as mean ± SD, t-test was used for making comparison between two groups, and oneway analysis of variance (ANOVA) was used for comparing more than two groups. The qualitative data were presented by rate, and chi-square test was used for making comparison. Besides, AMOS 22.0 software was applied for con rmatory factor analysis; Pearson's correlation coe cient was used for correlation analysis; additionally, factors in uencing NDI, such as somatization, anxiety, and depression were analyzed by linear regression analysis; P<0.05 was considered statistically signi cant.  Table 1.

Results
Among the scores evaluated by these two scales, a signi cant difference was found in gender (P<0.05).
However, there were no signi cant differences between different ages and levels of educational (P<0.05).
Items used for developing the PHQ-8 were selected by the four methods Using discrete degree method, chest pain, shortness of breath, fainting, palpitation, sexual life pain or other discomfort, numbness/tingling, excessive sweat, heavy hands/foot, a bursts of cold/fever, memory loss/forgetfulness, urinary frequency, urinary pain /dysuria, blurred vision, neck and shoulder pain, and muscle pain were removed since the SD of the items was de ned < 0.5.
According to factor analysis method, the value of KMO value for the initial scale was 0.668, and the value of Bartlett's spherical test was 366.894, P < 0.01, which was appropriate for factor analysis, and indicated that each item had a factor loading of more than 0.4 in its dimension except for headaches and chest pain.
Cronbach's α coe cient method showed that the α coe cient increased after removing fainting and urinary pain /dysuria.
The screened items of the developed PHQ-8 are presented in Table 2, including back pain, pain in arms, legs, or joints, dizziness, fatigue, dry mouth, feeling tired or having low energy, insomnia or other sleep problems, and throat discomfort.
Reliability analysis of the developed PHQ-8 and the SSS-8

Intrinsic reliability analysis
Cronbach's α is a measure of the internal consistency or reliability. The Cronbach's α test is typically utilized to examine the consistency and stability of the questionnaires. Hence, the Cronbach's α was herein applied to ascertain whether the items were reliable in measuring the same dimension. It is generally believed that when the Cronbach's alpha coe cient is greater than 0.7, the reliability is satisfactory [48]. The Cronbach α coe cient of the developed PHQ-8 and the SSS-8 was 0.601 and 0.553, respectively. The correlation coe cient between each item and the total score was 0.426 0.652 and 0.359 0.573, respectively.

Criterion validity
Criterion validity unveiled that the correlation coe cient between the total score of the PHQ-8 and the SSS-8 and the PHQ-15 was (r=0.739, P=0.000) and (r=0.835, P=0.000), respectively. As shown in Table 3 01), respectively. The exploratory factor analysis was carried out on the scale, and the principal component method was used to maximize the orthogonal rotation through the covariance matrix and the variance, in which the common factor was extracted by using the Kaiser criterion (Eigenvalue > 1). Additionally, 3 common factors of the developed PHQ-8 were extracted, and the cumulative contribution rate was 55.103%. The range of factor loading was 0.482 0.802, which was higher than the minimum standard of structural validity test equal to 0.4 [46,48]; Besides, 3 common factors for the SSS-8 were extracted, and the cumulative contribution rate was 51.666%. The range of factor loading was 0.353 0.881. All coe cients related to factor loading for PHQ-8 and the SSS-8 are summarized in Tables 4 and  5.
Con rmatory factor analysis: Con rmatory factor analysis model often employs values of chi-squared values (c 2 ), root mean square residual (RMR), root mean square error of approximation (RMSEA), goodness-of-t index (GFI), adjusted GFI (AGFI), comparative tting index (CFI), Tucker-Lewis index (TLI), normed t index (NFI), and other indicators were utilized to evaluate the tting effect of the model. The smaller the value of c 2 , and GFI, AGFI, CFI, TLI, NFI > 0.9 indicated that the model tted well, the closer to the "l", the better; besides, RMR and RMSEA < 0.05 indicated that the model tted well, the closer to "0" t, the better. The developed PHQ-8 and the SSS-8 were separately assessed by exploratory factor analysis, and 3 common factors were extracted. Two scales con rmatory factor analysis had a good 3-factor model t, shown in Table 6.
Correlation analysis of anxiety, depression, somatization, and the DSS for the developed PHQ-8 and the SSS-8 Somatization was evaluated by the developed PHQ-8 and the SSS-8. Correlation analysis showed that anxiety, depression, and somatization were positively correlated with the DSS ( Table 7). The correlation coe cient between the DSS and somatization assessed by the SSS-8 was higher than that of the developed PHQ-8.
Linear regression analysis of the effects of anxiety, depression, somatization on the DSS Somatization, anxiety, and depression were taken as independent variables, and the DSS was taken as a dependent variable, and linear regression analysis was conducted by the backward elimination method. After adjusting for factors, such as gender, age, type of FD, level of education, and employment situation, it was uncovered that depression and somatization were found as factors in uencing the DSS (Table 8).
The adjusted R 2 for the PHQ-8 and the SSS-8 were 0.263 and 0.263, respectively (all P<0.001). It also was unveiled that the role of somatization might be more signi cant than depression. The standardized β for the SSS-8 seemed to be higher than that for the PHQ-8.

Correlation analysis of anxiety, depression, somatization and QoL
Correlation analysis showed that somatization, anxiety and depression were positively correlated with the NDI (Table 9). Compared with the SSS-8, the correlation coe cient between NDI and somatization for the developed PHQ-8 was higher.
Linear regression analysis of the effects of anxiety, depression, somatization on QoL The NDI for QoL was taken as a dependent variable, and somatization, anxiety and depression were as independent variables. The linear regression analysis was undertaken by backward elimination method. After adjusting for factors, such as gender, age, type of FD, level of education, and employment situation, somatization, anxiety and depression were noted as factors the in uencing QoL (Table 10). The adjusted R 2 for the PHQ-8 and the SSS-8 were 0.224 and 0.236, respectively (all P<0.001). Somatization appeared to be more important than anxiety and depression. The standardized β for the PHQ-8 was greater than that for the SSS-8.

Discussion
The FD is the result of the interaction of physiological function and psychosocial factors through braingut axis changes [2,5]. Somatization in psychosocial factors in uences health-related QoL after controlling symptoms of anxiety and depression disorders in FD patients [30,49]. Those factors may lead to the deterioration of symptoms by aggravating anxiety and depression, resulting in repeated medical treatment and economic burden [33]. Somatization symptoms are extensively assessed by the PHQ-15, possessing high reliability and validity not only in general population, but also in patient population [34,50].
However, assessment of somatization symptoms by the PHQ-15 accompanies with a number of limitations in the study of FD patients. First, the PHQ-15 is not a speci c scale for FD patients, which has three items overlapping with gastrointestinal symptoms. Secondly, there is a menstrual problem which is not applicable to menopausal women and may cause differences in gender. Thirdly, due to cultural differences, patients assessed by the PHQ-15 has a poor compliance with questions related to sexual issues. Fourthly, the relative incidence of syncopal items is low [35]. Finally, the majority of patients with somatic symptoms are often visited in a general outpatient clinic without non-psychiatric doctors, which may result in less satisfactory results. Hence, it is highly essential to use a more simple tool to assess FD patients' somatization symptoms and its severity, which is conducive to doctors to present patients more reliable therapeutic advices. Therefore, the SSS-8 has been used for fast assessment of somatization symptoms and its severity, avoiding differences in gender caused by menstrual abnormalities, sexual life problems caused by cultural differences, and low incidence of syncopal items [38], while it is a non-speci c scale for FD patients and overlaps with gastrointestinal symptoms in the clinical application. In addition, the SSS-8 did not include FD patients' typical symptoms, such as throat pain, which may lead to neglect those symptoms by clinicians.
Hence, due to limitations of the SSS-8, in the present study, the PHQ-8 was developed by screening 25 items related to somatization. Gierk et al. [40] used con rmatory factor analysis to analyze the validity of the SSS-8, and their results showed that the 3-factor model outperformed the 1-factor model. Similarly, we, in the current study, used a 3-factor model for con rmatory analysis of the PHQ-8 and the SSS-8. The 3-factor model tted for the SSS-8 showed to have a superior performance than that of the PHQ-8.
The developed PHQ-8, however, showed to have superiority compared with the SSS-8 for FD patients as follows: Firstly, the PHQ-8 is a speci c scale for FD patients; the PHQ-8 includes FD patients' typical symptoms of somatization, e.g. throat pain, which has no overlap with gastrointestinal symptoms.
Secondly, in the PHQ-8, each item is signi cantly correlated with the total score, re ecting higher reliability of the PHQ-8 than that of the SSS-8. Thirdly, the PHQ-8 may be superior than the SSS-8 in the internal consistency. Fourthly, the PHQ-8 has a higher cumulative contribution rate than the SSS-8 in the exploratory factor analysis. Finally, the PHQ-8 is markedly correlated with NDI compared with the SSS-8, when NDI is used as a classic QoL assessment scale in the criterion validity.
The developed PHQ-8 was further evaluated in order to compare with the SSS-8 for 612 FD patients including anxiety or depression, because previous studies showed that anxiety, depression, and somatization had mutual effects [21,33]. The results of the present study unveiled that the SSS-8 may play a more signi cant role in the DSS than the PHQ-8 because of the fact that the SSS-8 overlaps with gastrointestinal symptoms. However, somatization assessed by the PHQ-8 may have a greater in uence on QoL than that of the SSS-8 in FD patients.
There were some shortcomings in the present research. Firstly, the patients with FD came from tertiary hospitals. These patients may have severe symptoms, limiting the prevalence of other people with FD.
However, whether other FD patients from the community can achieve the same conclusion requires further studies. Secondly, this study mainly analyzed and evaluated speci c somatization symptoms of FD patients by the developed PHQ-8, which healthy people and patients with other diseases (e.g., irritable bowel syndrome IBS) were not involved. Thirdly, the PHQ-8 was not tested for retesting reliability for all the cases, thus, the reliability of the scales needs to be further con rmed. Fourthly, similar to the SSS-8, the PHQ-8 was used to assess the severity of somatoform disorders [34]. Finally, data achieved via questionnaires were often abnormally distributed, while those were analyzed similar to normally distributed data as reported in previous large-sample studies [33,44]. Accordingly, further researches with large sample size are required to improve the rationality of the developed PHQ-8.
In presence of the above-mentioned limitations, the results of the present study showed that the PHQ-8 developed for evaluating somatization and QoL outperformed the SSS-8 not only for FD patients without anxiety and depression, but also for FD patients with anxiety and depression. Somatization may be a more important factor in uencing DSS and NDI compared with anxiety and depression, as well as H. pylori infection, gastric sensitivity, gastric accommodation, and gastric emptying in FD patients [21,34]. Somatization was mainly treated by antidepressant drugs, which were often used to treat functional dyspepsia in patients without undergoing the effects of combined prokinetic and proton pump inhibitor therapy [34,51]. Therefore, evaluating somatization is essential in the study of FD patients. The PHQ-8 may be more helpful to quickly assess somatization in lieu of the SSS-8 in the future FD-based studies.

Conclusions
The PHQ-15 and the SSS-8 have overlap in gastrointestinal symptoms in FD patients, and the SSS-8 was recommended to be a replacement for the PHQ-15 in a short period of time. The developed PHQ-8 without having overlap in gastrointestinal symptoms showed to have satisfactory reliability and validity. Somatization assessed by the developed PHQ-8 may have a greater impact on QoL of FD patients compared with the SSS-8. Our ndings suggested that the developed PHQ-8 may be more effective for evaluating the in uences of somatization on QoL of FD patients rather than SSS-8, which may be helpful to further verify the e cacy of the developed PHQ-8 to promptly assess the effects of somatization on FD patients.

Declarations Ethics declarations
Ethics approval and consent to participate This study was approved by the Ethics Committee of the A liated Hospital of North Sichuan Medical College (Nanchong, China). All the patients signed the written informed consent form.

Consent for publication
Not applicable.

Availability of data and materials
The data that support the ndings of this study are available from the corresponding author upon a reasonable request.