Introduction

Somatic symptoms are the manifestations of psychological distress in the form of physical symptoms [1]. They have received much attention because of considerable impact on healthcare utilization, quality of life and behavioral problems [2, 3].

Historical background of studying somatic symptoms in Chinese

In the 1980s, somatization was found to be a recurring condition in psychiatric Chinese adults in whom few emotional symptoms were expressed and their psychological distress was attributed to physical problems [4]. In the 1990s, a study showed that some symptoms specific to the Chinese culture did not appear in the ICD-10, e.g., kidney weakness and feeling of pressure on the heart [5]. Chinese often use the heart to express emotions and feelings because the heart is the central and essential part of the body [6]. These findings demonstrated that ethnographic and cultural factors influence the expression of somatic symptoms [7]. In the 2010s, more researchers examined somatic symptoms in Chinese and reported that the prevalence of a moderate-to-high severity of somatic symptoms in general population aged 15–65 years was 14.6 % [8], which was comparable to the corresponding Western populations [9].

Stressors and somatic symptoms in Chinese

There had been increasing evidence supporting a link between stress and somatic symptoms [10]; however, little has been known about specific stressors that lead to somatic symptoms in Chinese. Older age, female gender, employment and marital status were known socioeconomic risk factors for somatic symptom in Chinese [8]. Yet, there have been other known chronic stressors such as childhood adversities, biological and psychological stressors. For example, although childhood adversities were found to affect stress responses in adulthood and caused toxic stress that leads to alternations in immune function and cardiovascular diseases [11, 12], whether and how such stressors would be associated with somatic symptoms had not been examined. Moreover, only few studies investigated the relationships between biological and psychological stressors and somatic symptoms in Chinese [13, 14]. In this connection, the aim of this study is to investigate the stressors that are risk factors for somatic symptoms in Chinese.

Methods

Participants and setting

A two-stage stratified sampling method was performed on a list of residential addresses obtained from the Hong Kong Census and Statistics Department in 2013. The addresses were stratified by geographical area and type of quarters, and households in a stratum were randomly selected by computer. A total of 202 participants were recruited. Eligible subjects who were Chinese and 15 years old or above were recruited. Those who were unable to read and communicate in Chinese were excluded.

Data collection procedure

Face-to-face interviews in lieu of self-completion in household visits were adopted to enhance the response rate and minimize missing data. The data were collected between August and September 2013. Before data collection, five interviewers were trained to ensure no misunderstanding of the questionnaire items. The eligible participant was first informed about the study background, purpose, and procedures. Written informed consent was obtained before study participation. This study was approved by the institutional review board of the HKU/HA HKWC.

Measurements

Somatic symptoms

The Patient Health Questionnaire-15 [15] has been widely used to measure somatic symptoms in the previous month. It comprises 15 somatic symptoms which are the most prevalent DSM-IV somatization disorder somatic symptoms. Each symptom was scored from 0 (“not bothered at all”) to 2 (“bothered a lot”). One item, menstrual pain, was only applicable to female participants; therefore, the total score was calculated as the average score of the response items multiplied by 15. Higher adjusted scores indicated a higher level of somatic symptoms experienced. The Chinese version used in this study has been validated in a general population aged 15–65 years in Hong Kong [8].

Potential stressors

  1. 1.

    Perceived general stressors The validated Chinese version of the 14-item Perceived Stress Scale [16] was used to assess perceived stress in life in the previous month. Higher scores indicate higher levels of perceived stress.

  2. 2.

    Psychological stressors The 14-item Hospital Anxiety and Depression Scale was used to assess common psychological stressors—anxiety and depression. The Chinese version has been validated with good psychometric properties [17]. Higher scores indicate higher levels of anxiety or depression.

  3. 3.

    Childhood adversity stressors Ten dichotomous items from the Adverse Childhood Experience Study were used to score adversities which occurred during childhood [18]. The items included were physical abuse, psychological abuse, sexual abuse, emotional neglect and physical neglect, mother treated violently, household substance abuse, household mental illness, parental separation or divorce and incarcerated household member. Higher scores indicate greater extent of exposure.

  4. 4.

    Biological stressors Chronic illness (e.g., diabetes, hypertension, cardiovascular diseases, stomach ulcer, and psychiatric illness) in the previous year and sleeping problems represented biological stressors. The Hong Kong Chinese version seven-item Insomnia Severity Index [19] was used to assess actual sleeping problems and the perceived consequences of insomnia. Higher scores indicate a higher severity of insomnia.

  5. 5.

    Socioeconomic stressors These items included age, gender, education level, employment, marital status, household income.

  6. 6.

    Stress buffering Consistent evidence shows that social support provides stress-buffering effects against life stressors [20]. To assess social support, the Chinese version of six-item ENRICHD Social Support Instrument was used [21]. Higher scores indicate higher social support.

Data analysis

Descriptive statistics were used to characterize the sample. To identify stressors associated with somatic symptoms, a structured multiphase regression was performed to allow appropriate adjustment of confounding variables [22, 23]. Firstly, potential stressors were classified into three groups. Group 1 included age, gender and childhood adversity stressors which were non-modifiable stressors, group 2 included socioeconomic stressors (marital status, employment status, education level and household income), and group 3 included perceived general stressors (perceived general stress), psychological stressors (anxiety and depression), biological stressors (insomnia and the presence of chronic illness) and lastly stress buffering (social support). The three groups had hypothesized causal links, i.e., group 1 variables may affect variables in groups 2 and 3 but not vice versa, and group 2 variables may affect those in group 3 but not vice versa. Secondly, structured multiphase regression began with phase 1 as a stepwise linear regression of somatic symptoms on group 1 variables. In phase 2, another stepwise regression of somatic symptoms was performed on group 2 variables after force entering group 1 variables that were significant in phase 1. In phase 3, stepwise regression was similarly performed on group 3 variables after force entering group 2 and group 3 variables that were significant in their respective phases. The independent effect of a variable was estimated as the regression coefficient of the variable in its phase of regression analysis. All statistical analyses were performed by using IBM SPSS Statistics software version 20.0 (IBM SPSS Inc, Chicago, IL, USA) with statistical tests two-sided and 5 % level of significance.

Results

We approached 130 households: of which, 13 households and five potential eligible participants refused to participate and 32 households were unreachable after three attempts. The response rate was 61.5 %. Table 1 presents the demographics of the sample and general population, and Table 2 presents the characteristics of the total sample. One-fifth of the participants (n = 40, 19.8 %) reported more than one type of somatic symptom in the previous month.

Table 1 Demographics of survey participants (n = 202) and the general population (n = 7,221,800) of Hong Kong in 2013 (n = 202)
Full size table
Table 2 Participant characteristics (n = 202)
Full size table

Table 3 gives the results from structured multiphase regression analysis. From phase 1, participants who were female (estimated coefficient = 1.47, 95 % CI .70–2.24, p < .001) or had adverse experiences in childhood (estimated coefficient = .41, 95 % CI .06–.76, p = .023) had significantly higher somatic symptoms. After adjusting for these variables in phase 2, only employment (estimated coefficient = .87, 95 % CI .003–1.74, p = .049) was significantly associated with higher somatic symptoms. Finally, in phase 3 when all the above-mentioned variables were adjusted for, the absence of chronic illness (estimated coefficient = −1.61, 95 % CI −2.55 to −.66, p = .001), a high level of anxiety (estimated coefficient = .26, 95 % CI .17–.36, p < .001) and severe insomnia (estimate coefficient = .11, 95 % CI .04–.18, p = .003) were significantly associated with a higher somatic symptoms.

Table 3 Structured multiphase regression of somatic symptoms in Chinese (n = 202)
Full size table

Discussion

This study provides additional supporting evidence for the link of stress and somatic symptoms. It also furthers our understanding of the stressors that predict somatic symptoms.

Among all the significant stressors, only insomnia and anxiety were modifiable risk factors for somatic symptoms. A large-scale cross-sectional study in Norway (n = 47,700) supported that insomnia is associated with pain-related somatic symptoms [24]. However, the association between pain and sleep may be considered bidirectional. Further longitudinal study is needed to investigate their causal relationships.

The influence of anxiety on somatic symptoms can be explained from biological perspectives. Anxiety has been characterized by sympathetic activation and parasympathetic deactivation, with an increased heart rate, decreased heart rate variability and increased blood pressure [25]. Although small changes in heart rate and blood pressure are unnoticed, they may trigger discomforts such as the feeling of pressure on the heart, which may be a unique Chinese somatic symptom [5]. Also, increased anxiety has been found to be associated with increased muscle tension [26], which could trigger somatic symptoms such as headaches and shoulder and limb pain. As supported by another study, patients with generalized anxiety disorder have been found to be more sensitive to bodily changes and therefore have more somatic complaints than people without these disorders [27].

In this study, chronic illness, another biological stressor, was not found as a risk factor for somatic symptoms, unlike a previous study conducted in Chinese elderly sample aged 65 or above [28]. In our study, chronic illness was referred to medically diagnosed diseases, such as diabetes, hypertension, cardiovascular diseases and psychiatric illness. These diseases were intentionally selected because they are non-somatic conditions. However, in the above-mentioned study, chronic illness included cancer and chronic musculoskeletal conditions such as arthritis, low back pain, gout and osteoporosis, which could cause joint or limb pain. Therefore, chronic illness in the previous study was expected to be highly associated with somatic symptoms.

This study reported that majority of Chinese (80.2 %) suffered from minimal somatic symptoms, which were less severe than those reported in a previous telephone survey in Hong Kong [8]. There could be a limited reach of individuals with the increasing trend of using mobile phones. Indeed, there has also been evidence of over-estimated worse self-reported health in telephone surveys [29].

The major limitation of this study is a cross-sectional design that did not provide for temporal changes in stressors and somatic symptoms. However, the structured multiphase regression model used in the data analysis maximized the prediction of associated factors on somatic symptoms. Another limitation was that the somatic symptoms were relied on self-reports on various measurements instead of medical diagnosis, which might induce recall bias. In addition, the response rate of the household survey was 61.5 %. Some eligible participants who had refused to participate in the study might have higher somatic symptoms. Therefore, selection bias could be possible. However, the demographics of our study participants were shown to be comparable to the general population in Hong Kong, which ensured the representativeness of the data. Furthermore, our sample included individuals as young as 15 years in whom the studied stressors, including education and marital status, would be mostly the same and their inclusion in the regression may attenuate their estimated effects. However, the effect of education was significant, while the effect of marital status remained insignificant even in individuals aged 18 years or above (results not shown). Hence, the attenuation effects deemed to be minimal.

To conclude, physical and psychological symptoms are intertwined. It was recommended that reducing anxiety and improving sleep quality would be helpful to reduce somatic symptoms.