Predictors of Self-Reported Depression in Korean Children 9 to 12 Years of Age

Early-onset depression is a relatively common, enduring and recurrent condition. It has been shown that childhood depression may be a risk factor in the development of other psychological problems, as well as adulthood depression.1, 2 According to Harrington et al. although most instances of adult depression are not preceded by childhood depression, depressive symptoms in children increase the risk for affective disorders in adulthood.3 In addition, it was found that most depressed children who were not treated at the proper time were likely to become depressed adults.4 These findings suggest that early-onset depression is a potentially dangerous disease.

Despite the debilitating effects of childhood depression on development and mental health, little research has been conducted regarding the predictors and correlates of early-onset depression. A number of assumptions have been made regarding this occurrence. One is that the high prevalence of comorbidity impedes professionals when diagnosing the mood symptoms of children.5, 6 A systematic review using meta-analysis reported that anxiety disorder, conduct disorder, and oppositional defiant disorder were often accompanied with childhood depression.7

Another problem is that parents are not the most reliable reporters of depressive symptoms. Even if they tend to be reliable reporters of externalizing behaviors and tend to view depression as a deviant behavior problem,8 they generally report fewer symptoms than do their children.9 It has become increasingly evident that children eight years old and older are reliable reporters of their internal mood states.10 Therefore, it has been proposed that clinicians should pay attention to reports by both parents and children.

Studies using a children's self-reporting scale such as the Children's Depression Inventory (CDI) have been conducted mainly in North America and in Europe.8, 11-13 Children's self-reported depression scores differ in different countries; previous studies have shown that significantly higher depression scores are correlated with the variables of not living with both natural parents, a lower socioeconomic status, or having an unemployed parent.10, 14, 15 However, many investigators have turned their attention to cultural background issues such as the style of emotional expression, the type of rearing, and educational practices.16-18 The clinical presentation of childhood depression and the risk factors for depression identified in Western cultures cannot be assumed to be equivalent to those observed in other cultures.19, 20 The point at issue is particularly relevant when self-reported measurement is utilized, because the cultural context presumably influences the task-taking attitudes of individuals.

The early onset of depression is chronic and likely to persist into adulthood; therefore, identifying the risk factors of early-onset depression in Korean children is crucial for reducing adult depression. The objectives of this study are summarized below.

Table 1 shows the sample characteristics and mean and standard deviation of CDI scores according to the age and gender variables. Of the 1,279 children, 632 (49.5%) were boys and 647 (50.5%) were girls, with an average age of 10.21 and 10.33 years, respectively. The highest CDI mean scores were found in 9-year-old children, whereas the lowest scores were found in 11-year-old children.

Table 1
Prevalence of Gender, Age, and CDI Scores

• Click for larger image

Approximately 14% of the children were identified as belonging to the high-depression group. There was no significant difference between boys and girls (χ² = 0.23, p > 0.05), but a significant age effect on the prevalence of the high-depression group was found (χ² = 10.65, p < 0.05). Additionally, 81% of 9-year-old subjects showed a low level of depression, and the remaining 19% of subjects in this age group were classified into the high-depression group. In contrast, 88.5% of the 11-year-olds were classified into the low-depression group (Table 2).

Table 2
Prevalence of Clinical Level of CDI Scores by Gender and Age

• Click for larger image

No significant family income effect on the prevalence of the high-depression group was found (χ² = 2.03, p > 0.05), but the education levels of the father (χ² = 11.58, p < 0.01) and the mother (χ² = 8.29, p < 0.05) were strongly correlated with the prevalence of depression. Lower education levels of the father and mother were found at rates of 12 (32.4%) and 13 (27.7%) in the high-CDI group. These frequencies were higher than the expected frequencies of 5.2 and 6.6 for the father's education level and the mother's education level, respectively (Table 3).

Table 3
Background Characteristics of the Children

• Click for larger image

In Table 3, parent-reported factors of children's depression are investigated, and the results are shown in adjusted odd ratios at 95% confidence intervals. All significant odds ratios are reported. In a univariate logistic regression analysis, all parent-reported factors from the K-CBCL subscales were found to be related significantly to selfreported depression in children (Table 4).

Table 4
Factors Associated with Children Depression: Results of Univariate Logistic Regression Analysis for K-CBCL

• Click for larger image

According to a multiple logistic regression analysis, factors associated with childhood depression included age (OR = 0.76 - 0.78, p < 0.01), a high education level of the father (OR = 3.01 - 3.44, p < 0.05), total behavioral problems (OR = 1.03, p < 0.01), total competence scale (OR = 0.76, p < 0.01), problem internalization (OR = 1.08, p < 0.01), social scale (OR = 0.78, p < 0.01), and school scale (OR = 0.59, p < 0.01). The remaining factors were not significant (Table 5).

Table 5
Factors Associated with Children Depression: Results of Multiple Logistic Regression Analysis

• Click for larger image

The aim of this study was to investigate descriptive terms of self-reported depression in Korean children as well as to explore which sociodemographic factors, family factors, and parental-reported behavioral problems were strongly associated with levels of depression. In this study, the self-reported depression mean score as measured by Kovacs Children's Depression Inventory was 12.33 (standard deviation 6.79). A cut-off point of 19 identified approximately 14% of the children as depressed. Higher CDI scores have been found more frequently in some European and Asian countries29, 30 compared with Western countries.10, 12, 22 Charman and Pervova (1996)30 reported in samples from eastern Europe and Russia that in terms of the mean scores (mean 14.0, SD 5.6), 22.7% of participants in their study were above the cut-off point. In a study to assess the level of depression among different ethnic groups, it was found that Mexican-American adolescents appeared to be at a higher risk for depression compared with Anglo adolescents. It was concluded that depressive symptoms were positively related to deprived social economic status.31 Accordingly, it can be assumed that high levels of depression in children are associated with impoverished circumstances. However, Japanese scholars have reported that Japanese adolescents responded to negatively-worded items; this finding may indicate a cultural difference.18 They reasoned that Eastern people restrained the expression of positive affect. Moreover, a recent Korean study that verified a tripartite model of anxiety and depression in children and adolescents demonstrated that three factors (a low positive affect, a negative affect, and physiological hyperarousal) showed a higher correlation in the Korean sample than in a Western sample.17 They proposed that because Korean culture was based on Confucianism, which emphasizes courtesy and good manners, people repressed emotional expression. Considering that the socioeconomic level in Japan and Korea is relatively high, it is reasonable to assume that expressions of negative emotions, such as depression, are affected by social factors.

There was no marked differences in the CDI scores by gender in this study, which is consistent with studies that found that the ratio of childhood depression for boys and girls is 1:1; gender differences begin to appear in older adolescence.32, 33

In the analysis of the results of this study, the level of depression differed depending on the age of the subjects, as well as on the level of education of the father and the mother. The findings regarding age and gender in depressed children and adolescents are inconsistent, however. Many authors have reported that the prevalence of depressive symptoms increases over time from childhood to adolescence,34, 35 while other studies have found that prepubescent children and older adolescents have higher depression scores.8, 36 Our results, which showed that younger children were found to have higher CDI scores, is in line with findings that suggest that the proportion of subjects that scored above cut-off score in CDI was significantly higher in elementary school students than Junior high school.8, 12 Several explanations are possible for this result: First, the data in the present study suggest that a subject's attitude toward depressive symptoms has an impact on the results. Poli et al.8 explained that some young adolescents believe that depressive symptoms are a sign of weakness. In contrast, younger or older students can more easily accept their depressive feelings and express them more overtly. Similarly, younger subjects in the present study may be less reluctant to express depressive symptoms, because they are not highly self-conscious at their age. Second, it can be inferred that the course of psychopathology has an influence on this age effect. In a community-based research study, it was found that anxiety symptoms were more frequently reported in childhood and that the symptoms of depression were more frequently reported in adolescence.36 It is possible, however, that 9-old-year children may confuse anxiety with depression. Third, according to Twenge and Nolen-Hoeksema,32 depression scores for girls are stable from ages 8 to 11 and then increase between the ages of 12 to 16. Thus, the limited age range of the sample in the present study may explain why its results differ from those of some studies.

Univariate logistic regression analysis shows that the high-CDI group and the low-CDI group differ in all behavioral domains identified in the parent-reported K-CBCL. It is generally accepted that parents tend to report their children's depression as oppositional behavior, attention problems, crying, irritability, and somatic symptoms.37 Consistent with other studies, the present study found that parents are more sensitive reporters of externalizing behaviors and have a tendency to interpret some depressive symptoms as other behavioral problems.

Multivariable logistic regression analysis revealed that age and the education level of the father were independently associated with the risk of childhood depression. These findings support previous findings which suggested that numerous socio-demographic factors are implicated in the early development of childhood depression.38, 39 In this study, information regarding the level of parental education and income was collected, and it was found that only a higher education level of the father decreases a child's odds of having a high CDI score. In contrast, low socioeconomic status has repeatedly been associated with behavioral problems in children,40, 41 and evidence that suggests that parental education level is related to problematic behavior has not been consistent.40, 42 According to Kalff et al.14 parental education level and occupation are potentially important, because they have a considerable influence not only on the experience of economic hardship but also on the knowledge, experience, and rearing patterns of the parents. Therefore, it was postulated that a higher education level of the father was related to a decreased incidence of mental illness in children.

As shown in the logistic regression analysis, total behavioral problems, problem internalization, and the total competence scale including the social scale and the school scale measured by the K-CBCL were independent predictors for child depression, while the parent-reported anxiety/depression scale was not significant. Although the parental report did not discriminate depressive/anxious symptoms on the K-CBCL, parents were able to identify the total internalization of problems, which is predictive of high CDI scores, showing that perhaps parents can also be useful reporters, since problem internalization has been shown to be correlated with depression.43 Therefore, it appears that parents are aware of some internal states; parents have some perception of depressive symptoms, which makes the consideration of both parental and child reports important in clinical diagnosis.

Although the sample was considerably large and well-established instruments were used, this study's findings are limited. One limitation is the absence of information on psychiatric diagnoses. Given that this investigation utilized a large sample, person-to-person interviews using a structured diagnosing interview form were not feasible. Diagnostic formulation is necessary to obtain a more accurate concept of the prevalence of depression in young Korean children.

Other limitations include the restriction of the study to one geographical region in Korea and the study of children from 9 to 12 years of age only, thus introducing a generalization problem. The absence of teacher reports may be another limitation. Parents and children may differ in their opinions of a child's symptoms, especially when the children are in higher grades; therefore, teacher reports may contain important information. In addition, this study did not include the employment status of both parents; this constitutes another limitation. Although the number of dual-income couples was low when the K-CBCL was validated in Korea, the proportion of those working together for a living has increased markedly since that time.

In conclusion, these findings imply that self-reported depression scores measured by Kovac's Children's Depression Inventory in Korean children are slightly higher when compared to scores from some Western countries. Additionally, CDI scores are associated with the age of the child and the education level of the parent, with multiple behavior problems being reported by their parents as predictors of depression. Self-reported depression by children was correlated with parental reports of emotional problems, social and school adaptation not anxious/depressed scale, suggesting that parents are less perceptive in discriminating depression from other various emotional problems.