1. Introduction
Depression is a mental illness that affects approximately 23.1% of Korean males and 27.4% of females [
1]. Depression has been claimed as one of the risk factors for lower quality of life [
2], physical diseases [
3], and suicide [
4]. Unlike Japan, the suicide rates have continued to rise in Korea and is now the fourth leading cause of death [
5]. Considering depression has such a large impact on society (e.g., healthcare costs, quality of life, and life expectancy) and is highly prevalent across all ages, it is important to understand the behavioral factors that may be linked to it.
Sleep, diet, and exercise are three factors that play a significant mediating role in the development, progression, and treatment of depression [
6]. While a large number of studies supported the benefit of exercise on reducing depression, studies on sleep and diet were less explored. A prospective study of Japanese adolescents revealed that sleep disturbance was significantly associated with poor mental health [
7]. Chang et al. [
8] found associations between poor sleep quality and depression in the elderly. A community health survey in Korea revealed that short sleep duration was associated with the increased prevalence of depression [
9].
A recent meta-analysis study indicated that high-fish consumption could reduce the risk of depression [
10]. A study conducted among Japanese employees showed that fish consumption was associated with resilience to depression [
11]. Tanskanen et al. [
12] found that the likelihood of having depressive symptoms was significantly higher among infrequent fish consumers than among frequent consumers. A longitudinal study of young adults aged seven to 15 years old from Australia suggested that fish consumption was associated with depression in women, but not in men [
13].
Although previous studies have associated sleep and fish consumption with depression, several important issues need to be highlighted. First, previous studies on sleep and depression have mainly been conducted in Western countries; only a few studies have been conducted on the Korean population [
8,
14]. Due to differences in lifestyle and culture, findings obtained from Western countries might not be applicable to the Korean population. Second, while sleep disturbance such as insomnia is often the focus of sleep and depression research, aspects of sleep such as sleep quality and circadian chronotype are increasingly recognized for their substantial contribution to well-being. For example, the Netherlands Study of Depression and Anxiety revealed that late chronotype (evening-type of sleep) was associated with depressive disorder [
15]. Studies conducted among workers [
16] and freshmen [
17] in Japan showed similar associations between late bedtime and depressive symptoms. Third, sex differences exist in sleep quality, duration, latency, and architecture in the general population [
18]; however, the impact of sex differences in relation to sleep, fish consumption, and depression in the Korean population is not well-understood.
In the current study, we attempted to overcome some of these limitations. Using a large, population-based national sample of Korean adults, we aimed to examine the associations between depressive symptoms, fish consumption, and various sleep behaviors—such as bedtime, sleep onset latency, sleep duration, and sleep quality—in men and women. We hypothesized that depressive symptoms would be higher in respondents who had less fish consumption, late bedtime, late wake-up time, prolonged sleep latency, short sleep duration, and poor sleep quality. To our knowledge, our study was the first to specifically examine sex differences of fish consumption and various sleep-behaviors in relation to depressive symptoms among the general population in Korea.
2. Materials and Methods
2.1. Study Population
This study was performed over the course of one month from 23–31 January 2017 using a questionnaire developed by the authors. The online survey was conducted in Korea using a stratified random sample of 600 participants between the ages of 20 and 69, whose gender and age were proportional to an estimation of Korea’s general population.
The participant had to fill out the online informed consent, which explained that the answers would be analyzed only for our research purposes and that there was no way to link the data to their identity when they registered as potential respondents for the survey company. All participants were assigned a random number in the survey by the server, and their names remained anonymous.
2.2. Assessment of Depressive Symptoms
To assess the depressive symptoms, we used the Korean version of the Center for Epidemiological Depression Scale (CES-D), which had been previously validated in the Korean population [
19]. The CES-D consists of 20 items with a total CES-D score ranged between 0–60. A score of 16 or higher indicated that the person is at risk of clinical depression [
20].
2.3. Assessment of Sleep Behavior
Bedtime, wake-up time, sleep onset latency, sleep duration, and sleep quality were assessed by the Korean version of the Pittsburgh Sleep Quality Index (PSQI-K) [
21]. The PSQI is a 19-item instrument that assesses sleep disturbances in seven areas during the prior month: subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of sleep medications, and daytime dysfunctioning [
22]. The global PSQI scores range from 0–21 and scores of five or higher indicate sleep disturbances [
22]. However, in this study, we employed an 8.5-point threshold where people with 8.5 or more points were categorized as the poor sleep quality group, while people below 8.5 points were in the good sleep quality group. In the Korean version of the PSQI, a cutoff score of 8.5 represents a sensitivity of 0.943 and a specificity of 0.844, which is higher than the score of five in the original paper [
21].
For this study, bedtime, sleep duration, and sleep onset latency were categorized into tertiles with references of early bedtime (≤23:00) for bedtime, (<6 h) for sleep duration, and 10 min or less for sleep onset latency.
2.4. Assessment of Fish Consumption
Using a questionnaire on health and the environment developed by the authors, participants were asked to estimate their fish consumption intake by choosing out of six response options ranging from never and less than once a year to almost every day. We categorized the fish consumption intake into three categories: frequently (a few times a week to almost every day), occasionally (a few times a month), and rarely (a few times a year or less). Participants were also asked whether they consumed smoked salmon which were categorized into three categories: frequently (a few times a week to almost every day), occasionally (a few times a month), and rarely (a few times a year or less).
2.5. Assessment of Other Variables
The questionnaire on health and the environment developed by the authors, comprised of questions regarding sociodemographic characteristics, lifestyle habits, and health conditions. Sociodemographic characteristics included in the analysis were age (20–29, 30–39, 40–49, 50–59, 60–69 years old), sex, living condition (living alone, living with family or friends), perceived health (very healthy, healthy, unhealthy), average monthly family income (10 million Korean won (KRW) or less, 20 million KRW or less, 30 million KRW or less, 40 million KRW or less, 50 million KRW or less, 50 million KRW or more), educational status (high school or lower, undergraduate level, graduate level or higher), job position (public official, company employee, business owner, self-employed, temporary employee, part-timer, student, housewife, househusband, unemployed, others), and “yes” or “no” responses to questions regarding religion, smoking habit, drinking habit, and exercise habit.
2.6. Statistical Analysis
Chi-square test was used to examine the differences in the proportion of sex, depressive symptoms cut-offs (absent and present), and sociodemographic characteristics.
Multiple logistic regression analyses stratified by sex were used to analyze whether sleep behavior and fish consumption predict depressive symptoms. Three different models were tested to examine the association between sleep behavior and depression. First, we adjusted for age (Model 1). Previous research suggested that dietary factors and exercise may play a role in depression [
6]. Therefore, in Model 2, fish consumption and exercise were mutually adjusted. In Model 3, we additionally adjusted for sociodemographic factors (education level, occupation, household income, living condition, and religion) and health behavior (perceived health status, drinking habit, and smoking habit). Likewise, three different models were tested to examine the association between fish consumption and depression. First, we adjusted for age (Model 1). In Model 2, sleep quality, exercise and smoked salmon were mutually adjusted. In Model 3, we additionally adjusted for socio-demographic factors (education level, occupation, household income, living condition, and religion) and health behavior (perceived health status, drinking habit, and smoking habit). A recent study indicated that omega-3 fatty acids have been shown to be associated with depression [
23]. Therefore, we adjusted smoked salmon consumption in Model 2 and Model 3. Statistical analysis software SAS Version 9.2 (SAS Institute, Inc., Cary, NC, USA) was used for data analysis.
4. Discussion
The aim of our study was to detect gender-specific differences in the relationship between sleep behavior, fish consumption, and depressive symptoms. This study produced three main findings. First, sleep-onset latency and sleep quality were associated with increased odds of depressive symptoms, independent of all confounders, regardless of sex differences. Second, bedtime and sleep duration were significantly associated with an increased prevalence of depressive symptoms only in women. Third, higher fish consumption was significantly associated with decreased prevalence of depressive symptoms in men, but not in women.
Previous studies have reported a significant association between evening chronotype and depressive symptoms [
15,
24]. A study of almost 16,000 adolescents found an association between earlier parental set bedtime and depression [
25]. Epidemiological studies conducted in Japan reported that late bedtime predicted depression in workers [
16] and university students [
17]. Consistent with those studies, our study found that those who had late bedtime habits were also at a higher risk of having depressive symptoms. However, our study found a significant association between late bedtime and depressive symptoms in women, but not in men. This could be due to the fact that men and women are different biologically and physiologically, and may underlie the differential risk for chronotype on depression [
26]. These results highlight the importance for public health practitioners to promote early sleep timing for women in particular to prevent the development of depression.
Sleep onset latency is one symptom of depression [
27]. However, this association is thought to be bidirectional [
28]. A previous epidemiology study among Japanese university students also identified that sleep-onset latency was associated with an increased prevalence of depressive symptoms [
17]. These previous findings were consistent with our results which proved the importance of reducing sleep onset latency to lower the risk of depressive symptoms. Further research on factors related to sleep onset latency is warranted.
A recent meta-analysis on sleep duration and depression indicated that short and long sleep duration were significantly associated with increased risk of depression in adults [
29]. Previous epidemiology studies among Japanese workers and university students also identified that short sleep duration was associated with increased prevalence of depressive symptoms [
16,
17]. Consistent with these findings, we found an increased prevalence of depressive symptoms among women with a short sleep duration (<6 h), but the association was not significant in men. The mechanism was unclear; however, a previous study found that poor sleep was strongly associated with high levels of psychological distress and greater feelings of hostility, depression, and anger [
30]. Dysregulation of the serotonergic system is a potential mechanism that underlies the observed gender-specific relationship between sleep symptoms and depression [
31].
The association between sleep quality and depression has been described in previous research [
32,
33]. Previous epidemiological studies revealed that sleep quality was associated with increased risk of depressive symptoms [
17,
32]. In a clinical study, sleep quality revealed itself as a significant determinant of onset of major depression [
33]. In line with these findings, our study found that sleep quality predicted depressive symptoms in men and women, which demonstrated the importance of targeting sleep quality in men and women to prevent the development of depressive symptoms.
A longitudinal study of Australian young adults aged 7–15 years old suggested that fish consumption was associated with depression in women, but not in men [
13]. In contrast, our study found that higher fish consumption was significantly associated with a decreased odds ratio of depressive symptoms in men. Similar to our finding, several cross-sectional studies and a prospective study have also observed a significant association between fish consumption and depression in men. For example, two cross-sectional studies conducted in Finland showed that higher fish consumption was associated with lower depression in men, but not in women [
34]. In a prospective study using a large national US sample, men who consumed more fish were less likely to have depressive symptoms after a 10-year follow-up [
35]. The reasons for these sex differences are not clear; however, age and hormonal differences may reflect the results in men and women. This result may be useful as a reference to reduce the risk of depression.
The strengths of this study were the use of consecutive sampling with proportional sample size. However, the limitations of this study need to be addressed. First, we could not rule out recall bias in self-reported sleep behavior and depressive symptoms. Second, its cross-sectional design did not allow us to infer causality or specify the direction of the effect. Third, although we used reliable and valid measures of depressive symptoms, these measures did not constitute a clinical diagnosis of depressive disorders. Fourth, an internet survey can be subject to significant biases resulting from under-coverage and nonresponse. However, we have reduced the bias by using random sample strategy and we only included the respondents who completed the survey.