The Expression of Depression in Asian Americans and European Americans

Transcription

1 Journal of Abnormal Psychology 2014 Psychological Association 2014, Vol. 123, No. 4, X/14/$ The Expression of Depression in Asian s and European s Jean M. Kim and Steven Regeser López University of Southern California Past studies of the expression of depression in people of Asian descent have not considered whether observed ethnic differences in somatization or psychologization are a function of differences in the expression of the disorder or of group differences in the degree of depressive symptomatology. In the present study, we carried out 2 and Item Response Theory (IRT) analyses to examine ethnic differences in symptoms of Major Depressive Disorder in a nationally representative community sample of noninstitutionalized Asian s (n 310) and European s (n 1,763). IRT analyses were included because they can help discern whether there are differences in the expression of depressive symptoms, regardless of ethnic differences in the degree of depressive symptomatology. In general, although we found that Asian s have lower rates of depression than European s, when examining specific symptoms, there were more similarities (i.e., symptoms with no ethnic differences) than differences. An examination of the differences using both 2 and IRT analyses revealed that when there were differences, Asian s were less likely to endorse specific somatic and psychological symptoms than European s, even when matched in degree of depressive symptomatology. Together, these community-based findings indicate that depression among Asian s is more similar than different to that of European s. When differences do occur, they are not an artifact of the degree of depressive symptomatology but instead a true difference in the expression of the disorder, specifically a lesser likelihood of expressing specific somatic and psychological symptoms in Asian s compared with European s. Keywords: somatization, psychological symptoms, depression, Asian s, Item Response Theory It is a commonly held notion by anthropologists and psychologists that mood disorders are embedded in Western culture and may not apply as well to people of non-western cultural backgrounds (e.g., Kleinman, 1988; Ryder et al., 2008). The Diagnostic and Statistical Manual of Mental Disorders-Fifth Edition (DSM-5; Psychiatric Association, 2013) states that culture provides interpretive frameworks that shape the experience and expression of the symptoms, signs, and behaviors that are criteria for diagnosis (p. 14). In support of this, past research has demonstrated that participants of Asian background are more likely than participants of European background to express depression in somatic terms (Yen, Robins, & Lin, 2000; Yeung & Chang, 2002). Examples of somatic symptoms include fatigue, gastrointestinal problems, headache, and pain. There are many descriptions and theoretical discussions of somatization in people of Asian origin (Kleinman, 1982; Parker, Gladstone, & Chee, 2001; Ryder et al., 2008). A recent literature review, however, reveals only limited evidence that people of Asian background are more likely than others This article was published Online First October 13, Jean M. Kim and Steven Regeser López, Department of Psychology, University of Southern California. We thank Richard John for his consultation with the statistical analyses. Correspondence concerning this article should be addressed to Jean M. Kim, Department of Psychology, University of Southern California, 3620 McClintock Avenue, SGM 501, Los Angeles, CA jeanmkim@usc.edu to report somatic symptoms (Uebelacker, Strong, Weinstock, & Miller, 2009). On the one hand, some clinical studies suggest that, for example, Chinese patients are more likely to endorse somatic symptoms than European patients (e.g., Huang, Chung, Kroenke, Delucchi, & Spitzer, 2006), and that depressed Malaysian Chinese outpatients are more likely than depressed Australian White outpatients to endorse a somatic symptom as the primary complaint (Parker, Cheah, & Roy, 2001). However, the findings across studies, especially those using community samples, have not been robust or consistent (Uebelacker et al., 2009). International community studies also do not find increased somatization among Asian participants. Weiss, Tram, Weisz, Rescorla, and Achenbach (2009) compared symptoms of depression in Thai and children and adolescents from a community sample and found that the Thai and groups endorsed similar levels of somatic (and psychological) symptoms (effect size of the mean contrasts 0.00, CI [.05.05]). When Kadir and Bifulco (2010) examined a community sample of Malaysian women, they found that both somatic and psychological symptoms of depression were expressed by these participants. Although this study was qualitative in nature, it is consistent with other community-based studies (e.g., Cheng, 1989; Cheung, 1982) that suggest that the high prevalence of somatization in people of Asian background is not likely observed in community samples. Similarly, Ryder and his research team also suggest that there may be little difference between Asian origin and European origin adults in the presentation of somatic symptoms. They argue that the difference is in people of Asian background endorsing fewer 754

2 ETHNICITY AND DEPRESSIVE SYMPTOMATOLOGY 755 psychological symptoms than people of European background. In their study of outpatients, Ryder and colleagues (2008) found that Chinese participants endorsed a significantly higher level of somatic symptoms than European Canadian participants on two of the three depression measures. The Chinese participants also endorsed a significantly lower level of psychological symptoms than the European Canadian participants on all three depression measures. The effect sizes for psychological symptoms were larger and more consistently significant than the effect sizes for somatic symptoms. Therefore, the authors argued that the truly distinctive cross-cultural feature in the expression of depression for people of Asian background may be their reporting of fewer psychological symptoms than people of European background (Ryder et al., 2008). Psychologization can be defined as the tendency to express distress in affective or cognitive terms (Kirmayer, 2001). Examples of psychological symptoms include feelings of worthlessness, irritability, tearfulness, and depressed mood. Thus, based on Ryder and colleagues findings, it is important that the study of depression examines both somatic and psychological symptoms to assess whether people of Asian origin tend to somatize, people of European origin tend to psychologize, or some combination of the two. The Diagnosticity of Somatic and Psychological Symptoms One problem with past studies that examine ethnic group differences in symptom frequencies is that it is unclear whether a difference in frequency reflects an ethnic difference in the expression of depression or simply a difference in the degree of depressive symptomatology. For example, in the Ryder et al. (2008) study, it may be that less psychologization among those of Asian background may actually reflect that they have less depressive symptomatology overall and not simply lower levels of psychological symptoms. Thus, it is important that efforts be taken to control for the degree of depressive symptomatology when carrying out these analyses. Item Response Theory (IRT) provides a statistical approach to examine whether ethnic group differences in depressive symptoms reflect differences in the expression of depression or group differences in the level of the latent construct. IRT provides mathematical expressions of the relationship between participants responses on an item (in this case, symptoms) and the underlying latent construct (in this case, depressive symptomatology). It accounts for the potentially confounding effect of the degree of depressive symptomatology by assessing whether the association between the item and the latent construct differs depending on race or ethnicity (e.g., Asian or European origin) when both racial or ethnic groups are matched in the degree of depressive symptomatology (Uebelacker et al., 2009). Although in Classical Test Theory, the trait is based on the total number of items endorsed, in IRT, the latent trait ( ) is estimated based on the participants responses and the properties of the items (Yang & Kao, 2014). Theta has a mean of 0 and a SD of 1, with an arbitrary range for the latent construct that is measured. Those with a more negative value of theta are thought to have less of the latent construct of depressive symptomatology, and those with a more positive value of theta have more of the construct of depressive symptomatology (Yang & Kao, 2014). Another advantage of the IRT approach is that it may be a more precise manner of testing whether there are ethnic differences in symptom expression than the traditional frequency approach. For example, with the traditional approach, there may be significant differences between two groups in the frequency of psychological symptoms endorsed, suggesting that the two groups express depression differently; yet the association between a given symptom and the construct of depressive symptomatology may be no different for the two groups. Conversely, there may be no difference in the frequency of psychological symptoms reported for the two groups, suggesting no difference in the expression of the disorder; however, the relationship between a psychological symptom and the construct of depressive symptomatology may be much stronger for one group than another. In IRT, a mathematical function specifies an item characteristic curve (ICC) that represents the probability of a response on an item varying with the level of the underlying latent construct, in this case depressive symptomatology. The relative position and slope are two important characteristics of this curve. First, the relative position of the ICC indicates the trait strength of the underlying construct (i.e., severity or difficulty parameter). A curve that is shifted more toward the right indicates that the item is more difficult for that group (compared with the group with a curve that is shifted more toward the left); this group requires more of the latent construct (i.e., depressive symptomatology) than the other group for the same probability of item endorsement. In other words, given an equivalent degree of depressive symptomatology, the group with the curve that is shifted more to the right is less likely to endorse the item. Second, the slope indicates the discriminability of the item (i.e., discrimination parameter). An item with a steeper ICC slope discriminates more effectively between different levels of the underlying construct. Differential item functioning (DIF) is a statistical approach that is used to test the null hypothesis that these item parameters do not differ between two groups. The use of IRT methodology for analyzing depressive symptoms and testing the somatic hypothesis is an emerging area of research (Uebelacker et al., 2009). Using IRT and a community sample, Uebelacker and colleagues (2009) tested whether Asian s would be more likely to somatize than European s, given similar degrees of depressive symptomatology. This study utilized the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) dataset, a large, nationally representative epidemiological sample of adults. Uebelacker and colleagues examined symptoms of depression, including analyses comparing Asian s and non-latino White s. For the severity parameter, given similar degrees of depressive symptomatology, Asian participants were more likely to endorse only suicidal ideation than non-latino White participants. For the discrimination parameter, only one symptom met criteria for significant DIF; difficulty in concentrating was less discriminating for Asian s than for European s. Overall, the results failed to find support for the notion that Asian s express depression differently than European s, even when controlling for degree of depressive symptomatology as carried out by IRT. The study was limited, however, in that their community sample included a relatively small sample of Asian s (n 291), compared with the other racial or ethnic groups included in this study (ns ranged from 468 to 10,958). The study also excluded those who were not fluent in English; thus, the sample likely reflected a more acculturated

3 756 KIM AND LÓPEZ Asian sample, thereby reducing the likelihood of finding ethnic differences. Finally, the analyses focused only on seven MDD symptoms in a 2-week episode of depressed mood or anhedonia. Only one other study to date has used IRT to examine DSM symptoms of depression between groups (Simon & Von Korff, 2006). However, these researchers included depressed primary care participants with and without a comorbid medical condition and did not examine race or ethnicity differences. There is also a large body of work using DIF analyses to examine measurement bias (e.g., Camilli & Shepard, 1994; Embretson & Reise, 2000; Holland & Wainer, 1993), including studies not using IRT but other statistical methods. For depression specifically, work using multiple regression (Birnholz & Young, 2012) and nonparametric kernel-smoothing techniques (Santor, Ramsay, & Zuroff, 1994) has examined depressive symptom severity scores and item bias between groups (i.e., female sexuality groups and males and females, respectively); however, these studies did not test for race or ethnicity differences. Another study (Dere et al., 2013) did examine race or ethnicity differences in depressive symptoms. With a clinical sample, Dere et al. (2013) used the standardized mean difference technique to assess for DIF among Han-Chinese and European Canadian participants. They found no DIF for typical somatic symptoms but did find DIF for atypical somatic symptoms and for psychological symptoms. Specifically, the Chinese reported higher levels of suppressed emotions and depressed mood, and European Canadians reported higher levels of atypical somatic symptoms and hopelessness, relative to their overall symptom reporting. Overview and Hypotheses In the present study, we drew on the Collaborative Psychiatric Epidemiology Surveys (CPES), a national psychiatric epidemiology database that includes both Asian s and European s, to examine the expression of depressive symptoms among Asian s and European s. The advantage of this database is that it is comprised of nationally representative samples of people residing in the community, whereas many past studies have only used clinical samples. Another advantage is that the CPES applies a broad definition of Asian reflecting many countries of origin within Asia. In addition, unlike the Uebelacker et al. (2009) study, non-english speaking Asian respondents were included, suggesting more variance with regard to acculturation. We applied two statistical approaches. The first was to identify the specific symptoms for which the two groups differed, using 2 analyses. The second approach was to examine whether these differences held, even accounting for degree of depressive symptomatology, using IRT. Although we believe that the IRT approach is the more precise way to test our hypotheses, by including the traditional approach, we were in a position to assess how our findings map on to past research and how they compare with the IRT approach. Accordingly, it allowed us to explore whether the statistical methods lead to similar or different results. Our first objective was to test whether Asian s and European s differ with regard to both somatic and psychological symptoms. Although Ryder et al. (2008) found smaller but significant differences in somatic symptoms between Asian and European Canadian outpatients on two out of three measures, other studies using community samples (e.g., Cheng, 1989; Cheung, 1982; Weiss et al., 2009) did not find a higher prevalence of somatization in people of Asian background. We predicted that Asian s and European s in this sample would not significantly differ in their levels of somatic symptom endorsement. For psychological symptoms, consistent with previous findings by Ryder et al. (2008), we expected Asian s in our sample would endorse lower levels of psychological symptoms than European s. Next, we examined the relationship between the symptom and the latent construct using IRT. For somatic symptoms, we expected that the severity parameters would not be significantly different between the two racial or ethnic groups. In other words, Asian s and European s would have similar probabilities of endorsing somatic symptoms, given the same degree of depressive symptomatology, and thus, similar severity parameters. On the other hand, for psychological symptoms, we expected that the severity parameter would be greater for Asian s than for European s. In other words, Asian s would have a lower probability of endorsing psychological symptoms than European s with the same degree of depressive symptomatology. For the discrimination parameter, we explored whether somatization or psychologization is differentially related to the construct of depressive symptomatology for Asian s or European s. Method Participants The participants were part of the CPES, specifically the National Latino and Asian Study (NLAAS) and the National Comorbidity Survey Replication (NCS-R). These studies together create one combined, nationally representative dataset with enough power to examine cultural and ethnic correlates of mental illness. The Asian data were selected from the NLAAS, which included participants 18 years and older in the contiguous United States and Hawaii. The NLAAS Asian sample (N 2,095) included: Chinese (n 600), Filipino (n 508), Vietnamese (n 520), and other Asian (n 467) participants. The category other Asian included Bangladeshi, Burmese, Cambodian, Hmong, Indian, Indonesian, Japanese, Korean, Laotian, Malaysian, Mongolian, Myanmai, Pakistani, Singaporean, Sri Lankan, Taiwanese, and Thai participants. Among the Asian s, 454 were born in the United States, 1,639 were born outside of the United States, and two did not report their place of birth. Interviews were completed in English, Mandarin, Cantonese, Tagalog, and Vietnamese. The mean age was 41.0 (SD 14.7). Forty-seven percent of the Asian s were male, and 53% were female. The response rate for Asian s was 69.3%. As this was a first step to examining ethnicity and culture using IRT and a nationally representative sample, Asian s were studied as an entire group, as were European s. To only focus on one of the specific ethnic groups (e.g., Chinese s) would restrict the focus to that specific group instead of a national sample of Asian origin adults. Moreover, the sample size of a specific group would be limited. In addition, using the entire sample is consistent with other NLAAS studies that examine Asian s as one group (e.g., Gee, Ro, Gavin, & Takeuchi, 2008;

4 ETHNICITY AND DEPRESSIVE SYMPTOMATOLOGY 757 Leu et al., 2008; Marques et al., 2011; Takeuchi, Alegría, Jackson, & Williams, 2007). The term Asian is used here because the participants reported national origin in a country located in the continent of Asia and they reported having residence in the United States. The Asian s who received the Depression Module (N 310) included: Chinese (n 100), Filipino (n 61), Vietnamese (n 50), and other Asian (n 99) participants from the countries of origin noted earlier. Among these Asian s, 106 were born in the United States, 202 were born outside of the United States, and two did not report their place of birth. The mean age was 38.7 (SD 14.1), and 39% were male. The European data were selected from the NCS-R, which included participants 18 years and older in the contiguous United States. The non-hispanic White sample included 4,180 people with a mean age of 46.5 years (SD 17.8). Forty-six percent of the European s were male, and 54% were female. Their response rate was 70.9%. The European s who received the Depression Module included 1,763 participants. Among these European s, 1,658 were born in the United States, 38 were born outside of the United States, and 67 did not report their place of birth. The mean age was 44.2 years (SD 15.0), and 36% were male. Measures Composite International Diagnostic Interview. The Composite International Diagnostic Interview (CIDI, World Health Organization) is a structured diagnostic interview that generates International Classification of Diseases (ICD-10) and Diagnostic and Statistical Manual of Mental Disorders-Fourth Edition (DSM IV) diagnoses. It was designed to be used across cultures. Trained, nonclinical interviewers administered the CIDI in person. This study primarily focused on the following sections of the CIDI. Screening section. In this section, participants were asked three questions specific to depression. These items consisted of questions asking about times when most of the day, one felt sad, empty, or depressed, very discouraged about how things were going, or when one lost interest in most things [he or she] usually [enjoys]. If the participant endorsed at least one of the screening items, they went on to answer the questions in the depression module. Depression module. This module consisted of symptoms that mapped on to the Diagnostic and Statistical Manual of Mental Disorders-Fourth Edition-Text Revision (DSM IV TR) criteria for Major Depressive Disorder. It included nine somatic symptoms (e.g., Did you have a much larger appetite than usual nearly every day? ) and 23 psychological symptoms (e.g., Did you feel hopeless about the future nearly every day? ). The depressive symptoms were separated as somatic or psychological based on consensus from past studies (e.g., Kadir & Bifulco, 2010; Kleinman, 1982; Ryder et al., 2008; Uebelacker et al., 2009; Weiss et al., 2009; Yen et al., 2000) and the face validity of the items. Results Overview We first examined the entire sample to see whether there were ethnic group differences in the depression screening items. We then carried out two distinct sets of analyses with those who screened positive for possible depression. These analyses were carried out with this subsample because all of the depressive symptoms were assessed with this group. To examine the rate of endorsement of specific somatic and psychological symptoms by Asian and European s, we conducted 2 analyses to examine specifically where (i.e., in which symptoms) the racial or ethnic group differences may lie. Because of the potentially confounding effect of the degree of depressive symptomatology in evaluating group differences in the expression of depression, our next set of analyses used IRT to examine any potential DIF in somatic and psychological symptoms. Ethnicity and Degree of Depression Across all assessments of depression, Asian s reported significantly less depression. First, when the overall sample is considered, a smaller percentage of Asian s than European s reported any of the depression screening items. For example, 31.9% of Asian s endorsed the item sad/empty/depressed, whereas half of European s (49.9%) endorsed this item (p.001, Cramer s V 0.16; see Table 1). Second, when considering the screened-in sample (i.e., those who received the full depression module), a smaller percentage of Asian s than European s reported two of the three depression screening items. For example, 84.2% of Asian s endorsed the item discouraged about life, whereas 92.0% of European Table 1 Rates of Depression Screening Item Endorsement by Race or Ethnicity (%) for the Overall Sample and for Those Who Screened In to Receive the Depression Module Overall sample Screened in sample Depression screening item Asian (n 2284) European (n 6696) Cramer s V Asian (n 310) European (n 1763) Cramer s V Sad/empty/depressed Discouraged about life Lost interest in enjoyable things Note. For Asian s, the number of refused responses ranged from 1 2, and the number of don t know responses ranged from 0 1. For European s, there were no refused responses, and the number of don t know responses ranged from 1 5. p.05. p.001.

5 758 KIM AND LÓPEZ s endorsed this item (p.001, Cramer s V 0.10; see Table 1). There was no significant difference between racial or ethnic groups for the item sad/empty/depressed in the screened-in sample. Last, Asian s had lower rates of Major Depressive Disorder than European s (Pearson 2, ps.001); this was the case for lifetime (Asian s: 9.2%; European s: 18.0%; Cramer s V 0.11) and 12-month Major Depressive Disorder (Asian s: 4.5%; European s: 7.2%; Cramer s V 0.05). Ethnicity and Symptom Endorsement Asian s and European s did not differ in the median somatic symptoms endorsed (Asian s: 4.00 symptoms, European s: 4.00 symptoms; , p.30). However, Asian s were significantly lower in the median psychological symptoms endorsed, when compared with European s (Asian s: 13.00; European s: 14.00; , p.02). In terms of the specific symptoms, 2 analyses revealed that Asian s were significantly lower than European s in their endorsement rate for 3 of the 9 somatic symptoms (33%) and 9 of the 23 psychological symptoms (39%). An example of a somatic symptom that was endorsed less by Asian s is a larger appetite; 8.2% of the Asian s reported having a larger appetite, whereas 14.3% of the European s reported having a larger appetite (p.004, Cramer s V 0.06; see Table 2). An example of a psychological symptom endorsed less by Asian s is guilt; only 41.9% of the Asian s reported guilt, whereas 50.9% of the European s reported this symptom (p.004, Cramer s V 0.06; see Table 3). Item Response Theory For the IRT analyses, Mplus Version 6.12 was used, applying a two-parameter model (2PL), which involves estimating a severity and discrimination parameter for each symptom. The 2PL model was preferred over the one-parameter (1PL) model, using Akaike information criterion (AIC) and sample-size adjusted Bayesian information criterion (BIC). Assumptions of unidimensionality of the trait (.99), local independence of items, and ability to model the response for an item via an item response function were met. Analysis was run containing all of the depressive symptoms (i.e., both somatic and psychological symptoms) in one IRT analysis, but the symptoms are presented in separate somatic and psychological symptom tables for clarity and consistency with our discussion of the results. The item parameters were compared across groups according to Linacre and Wright (1986). Results from these analyses are shown in Tables 4 and 5, which list the severity and discrimination parameters for each somatic and psychological depressive symptom across the racial or ethnic group comparisons. For somatic symptoms, only two out of nine somatic symptoms (22.2%) exceeded criteria for statistical significance in DIF for the severity parameter (see Table 4). For example, Asian s were less likely to endorse the somatic symptom fatigue/loss of energy than European s, given similar degrees of depressive symptomatology. This item s ICCs are plotted in Figure 1a. The severity parameter is typically examined from the horizontal axis, where the probability of endorsement of the item is 0.5 or 50%. Figure 1a shows that the curve for Asian s is shifted more toward the right. This indicates that this item is more difficult for Asian s, or requires more of the latent construct (i.e., depressive symptomatology) for the same probability of endorsement as European s. Stated another way, given equivalent degrees of depressive symptomatology, Asian s tended to be less likely to endorse the item fatigue/loss of energy than European s. In addition, one somatic symptom was more discriminating for Asian s than European s psychomotor agitation. The item s ICCs are plotted in Figure 1b. Again, the discrimination parameter refers to the degree to which the item discriminates between participants along the continuum of depressive symptomatology. The steeper slope for Asian s than European s in Figure 1b indicates that the item psychomotor agitation is more discriminating for Asian s than European s. Table 2 Rates of Depressive Somatic Symptom Endorsement for Those Who Screened In to the Depression Module by Race or Ethnicity Depressive somatic symptom of CIDI Asian European % endorsement n % endorsement n Cramer s V Lost weight Gained weight , Smaller appetite , Larger appetite , Insomnia , Hypersomnia , Psychomotor agitation , Psychomotor retardation , Fatigue or loss of energy , Note. The number of respondents per item varies given the computer algorithm of the Composite International Diagnostic Interview (CIDI) and its skip function. The maximum subsamples by ethnicity are Asian n 310 and European n 1,763. p.05. p.01. p.001.

6 ETHNICITY AND DEPRESSIVE SYMPTOMATOLOGY 759 Table 3 Rates of Depressive Psychological Symptom Endorsement for Those Who Screened In to the Depression Module by Race or Ethnicity Depressive psychological symptom of CIDI Asian European % endorsement n % endorsement n Cramer s V Felt depressed , Nothing could cheer , Discouraged , Hopelessness , Loss of interest , Nothing was fun , Worthlessness , Loss of confidence , Not as good as others , Guilt , Trouble concentrating , Indecisiveness , Thoughts come slowly , Thought about death , Better if dead , Thought about suicide , Made suicide plan , Made suicide attempt , Irritability , Could not cope with responsibility , Wanted to be alone , Less talkative , Tearfulness , Note. The number of respondents per item varies given the computer algorithm of the Composite International Diagnostic Interview (CIDI) and its skip function. The maximum subsamples by ethnicity are Asian n 310 and European n 1,763. p.05. p.01. p.001. For the psychological symptoms, an examination of DIF showed that 11 of these items differed across racial or ethnic groups in terms of severity or difficulty, and one item (thoughts about death) differed in terms of discrimination. The 11 of 23 psychological symptoms (47.8%) that exceeded criteria for statistical significance in DIF for the severity parameter are shown in Table 5. As an example, the ICCs for the item guilt are plotted in Figure 2a. This shows that given equivalent degrees of depressive symptomatology, Asian s are less likely to endorse feelings of guilt than European respondents. In addition, the discriminability of the item thought about death is plotted in Figure 2b, showing a steeper slope (i.e., better discriminability) for Asian s than European s. Discussion We found that Asian s within a national sample of U.S. residents are less likely to present with depressive symptoms and disorders than European s. Despite the clear difference in prevalence rates, there were more ethnic similarities in the report of symptoms than differences. There were no differences in nearly two-thirds (62.5%) of the depressive symptoms 6 of the 9 Table 4 Differential Item Functioning of DSM-IV Somatic Symptoms of Major Depressive Disorder for Asian and European s Severity or difficulty parameter Discrimination parameter Somatic symptoms Asian European p Asian European p Lost weight 1.10 (0.78) 4.01 (1.12) (0.12) 0.22 (0.06) 0.44 Gained weight 0.90 (1.60) (179.82) (0.33) 0.01 (0.09) 0.22 Smaller appetite 2.20 (1.01) 1.93 (0.41) (0.09) 0.18 (0.04) 0.38 Larger appetite 1.02 (0.42) 0.41 (0.12) (0.23) 0.52 (0.07) 0.37 Insomnia 2.72 (1.08) 2.74 (0.50) (0.10) 0.22 (0.04) 0.36 Hypersomnia 0.29 (0.43) 0.61 (0.09) (0.21) 0.76 (0.10) 0.96 Psychomotor agitation 0.69 (0.23) 1.28 (0.27) (0.29) 0.38 (0.06) 0.01 Psychomotor retardation 0.05 (0.11) 0.07 (0.06) (0.13) 0.63 (0.05) 0.10 Fatigue or loss of energy 1.22 (0.19) 1.70 (0.13) (0.19) 0.66 (0.06) 0.05

7 760 KIM AND LÓPEZ Table 5 Differential Item Functioning of DSM-IV Psychological Symptoms of Major Depressive Disorder for Asian and European s Psychological symptoms Severity or difficulty parameter Asian European p Discrimination parameter Asian European p Felt depressed 2.03 (0.32) 2.16 (0.14) (0.16) 0.95 (0.09) 0.84 Nothing could cheer 0.25 (0.11) 0.48 (0.05) (0.17) 0.84 (0.06) 0.15 Discouraged 1.21 (0.17) 1.56 (0.08) (0.16) 1.02 (0.08) 0.69 Hopelessness 0.16 (0.11) 0.52 (0.05) (0.19) 0.97 (0.07) 0.35 Loss of interest 0.70 (0.12) 0.91 (0.06) (0.15) 1.03 (0.07) 0.64 Nothing was fun 0.44 (0.11) 0.68 (0.05) (0.15) 0.89 (0.06) 0.25 Worthlessness 0.07 (0.10) 0.23 (0.06) (0.31) 1.06 (0.10) 0.09 Loss of confidence 0.81 (0.14) 1.13 (0.07) (0.16) 0.86 (0.07) 0.32 Not as good as others 0.34 (0.12) 0.35 (0.05) (0.14) 0.82 (0.06) 0.40 Guilt 0.43 (0.14) 0.02 (0.06) (0.12) 0.59 (0.04) 0.29 Trouble concentrating 0.95 (0.18) 1.17 (0.08) (0.14) 0.81 (0.06) 0.65 Indecisiveness 0.16 (0.14) 0.49 (0.05) (0.11) 0.80 (0.06) 0.95 Thoughts come slowly 0.23 (0.12) 0.19 (0.05) (0.14) 0.82 (0.06) 0.53 Thought about death 0.03 (0.11) 0.35 (0.06) (0.14) 0.56 (0.04) 0.02 Better if dead 0.35 (0.10) 0.29 (0.05) (0.17) 0.89 (0.06) 0.19 Thought about suicide 0.94 (0.15) 0.78 (0.06) (0.15) 0.76 (0.05) 0.19 Made suicide plan 1.27 (0.35) 1.63 (0.17) (0.21) 0.50 (0.08) 0.54 Made suicide attempt 2.14 (0.71) 2.88 (0.54) (0.20) 0.31 (0.08) 0.36 Irritability 0.46 (0.17) 0.46 (0.08) (0.11) 0.44 (0.04) 0.13 Could not cope with responsibility 0.20 (0.09) 0.20 (0.04) (0.19) 1.10 (0.07) 0.20 Wanted to be alone 0.91 (0.21) 1.32 (0.10) (0.12) 0.63 (0.05) 0.62 Less talkative 1.54 (0.32) 1.61 (0.12) (0.13) 0.64 (0.06) 0.61 Tearfulness 0.85 (0.35) 1.47 (0.20) (0.09) 0.30 (0.04) 0.50 somatic symptoms and 14 of the 23 psychological symptoms. When there were ethnic differences, however, Asian s were less likely than European s to report both somatic and psychological symptoms. Our hypothesis that Asian s would endorse similar levels of somatic symptoms compared with European s was not fully supported. For the majority of the somatic symptoms, similar to other studies using community samples, there were no differences between the two groups in level of endorsement. However, for one-third of the somatic symptoms, a lower percentage of Asian s than European s endorsed these symptoms, which is opposite to what Ryder et al. (2008) found using their clinical Chinese and European Canadian sample and contrary to what would be expected according to the somatization hypothesis. However, our finding is consistent with Dere et al. (2013), who also found larger appetite and hypersomnia to be less common in Han Chinese than European Canadian outpatients in their study. For psychological symptoms, the frequency analysis supports our hypothesis and Ryder s prior clinical findings that a lower percentage of Asian s endorse psychological symptoms than European s. To rule out the possibility of an artifact of less depressive symptomatology among Asian s compared with European s, we carried out IRT analyses. As mentioned, an advantage of the IRT approach over the typical frequency approach is that it accounts for the potentially confounding effect of degree of depressive symptomatology, and it may be a more precise manner of testing whether there are ethnic differences in symptom expression. With the IRT analyses, we found very similar results to the frequency analyses. There were more ethnic similarities than differences for both somatic symptoms (7 of 9 severity parameters; 8 of 9 discrimination parameters) and psychological symptoms (12 of 23 severity parameters; 22 of 23 discrimination parameters). However, when there were ethnic differences, relative to European s, Asian s were less likely to endorse specific somatic and psychological symptoms, given similar degrees of depressive symptomatology. The IRT analyses suggested some true differences in the expression of depression, but largely in the severity parameter for these symptoms. Asian s were less likely to endorse two somatic symptoms and 11 psychological symptoms. This may reflect differences in the relevance of these items across ethnic groups, particularly with psychological symptoms. Overall, it can be seen that the ethnic differences are largely the same using both the frequency and IRT approaches. The 2 and IRT analyses yielded significant differences across ethnic groups for both somatic and psychological symptoms, such that when there were differences, Asian s were less likely to endorse both somatic and psychological symptoms than European s. The IRT findings indicate that Asian s lower rates of some somatic and psychological symptoms are not an artifact of different degrees of symptomatology among Asian s compared with European s. Together, these findings challenge the view that persons of Asian origin somatize their depression. New neural evidence supports our findings that when there are differences between Asian s and European s, Asian s are less likely to endorse specific somatic and psychological symptoms. Immordino-Yang, Yang, and Damasio (2014) provide physiological and imaging data that suggest that Chinese and East Asian participants use somatosensory

8 ETHNICITY AND DEPRESSIVE SYMPTOMATOLOGY 761 Figure 1. (a) Illustrative item characteristic curves for the fatigue/loss of energy somatic depressive symptom item derived from Asian and European s. (b) Illustrative item characteristic curves for the psychomotor agitation somatic depressive symptom item derived from Asian and European s. information significantly less than non-asian s in the expression of emotion. This is consistent with Chinese Confucian principles, which emphasize settling the body to be better able to tune in to the social context (Markus & Kitayama, 1991). In contrast, it may be a more mainstream strategy to use bodily information to aid in the assessment of one s emotions. It is possible then that Asian s are less likely to endorse some somatic and psychological symptoms, because for Asian s they are not as interrelated and used in conjunction to assess emotional state. In contrast, Immordino-Yang, Yang, and Damasio (2014) suggest that European s may use somatic words to describe psychological states. Thus, it is not surprising that in our sample, Asian s are less likely to endorse somatic symptoms of Major Depression than European s, opposite of the somatization hypothesis. The study by Immordino-Yang et al. (2014), however, does not explain the difference seen between clinical and community samples. Past studies using clinical samples (e.g., Huang et al., 2006; Parker et al., 2001; Ryder et al., 2008) have found greater somatization in Asian origin than European origin patients. In contrast, our community study, along with Uebelacker et al. (2009) and Weiss et al. (2009), found no difference in somatization between Asian origin and European origin groups. An interesting area for future research is to better understand why we tend to see a pattern of somatization of depression in Asian origin clinical samples but not community samples. One possibility is that in Chinese societies there is less tolerance for disclosing one s illness outside the family (e.g., Lin, Tardiff, Donetz, & Goresky, 1978). To reduce the burden of stigma, Asian origin persons who seek professional help may tend to present their distress in somatic symptoms rather than in psychological symptoms, as somatic symptoms are less stigmatized (Goldberg & Bridges, 1988). Certainly, there is a body of evidence that suggests that mental illness is stigmatized, at least in Chinese societies (e.g., Chan & Parker, 2004; Chung & Wong, 2004). In contrast, in community samples, there is more heterogeneity those who do not have psychiatric illness, participants who have kept illness to themselves or within the family, and some who have sought help outside the family. Therefore, compared with a clinical sample where 100% of the participants have revealed a potentially stigmatized illness identity outside the family, there may be less of a need to emphasize somatic symptoms of depression. As stated in Ryder et al. (2008), [s]omatization allows psychologically distressed individuals to inhabit the sick role in their societies without bearing the burden of stigma (Goldberg & Bridges, 1988) (p. 302). Thus, it is possible that the pattern of greater somatization in Asian origin participants repre- Figure 2. (a) Illustrative item characteristic curves for the guilt psychological depressive symptom item derived from Asian and European s. (b) Illustrative item characteristic curves for the thought about death psychological depressive symptom item derived from Asian and European s.

9 762 KIM AND LÓPEZ sents a help-seeking bias in clinical samples. The source of the sample is a potentially important moderator for future studies to examine. Our findings add to the literature in important ways. First, although past literature has emphasized potential differences in the expression of depression, these results reveal greater similarities between community samples of Asian and European s in the expression of both somatic and psychological symptoms of depression. Second, the IRT analyses suggest that there truly are some differences in the expression of depression, but largely in the severity parameter for somatic and psychological symptoms. Compared with European s with a similar degree of depressive symptomatology, Asian s in a community sample are less likely to report some somatic and psychological symptoms, and they require more depressive symptomatology to report these symptoms. These findings differ from the IRT analysis of Uebelacker and colleagues (2009) who found only one ethnic difference for the severity parameter and one ethnic difference for the discrimination parameter. We place more confidence in the findings of the current study given the large sample size and given that the sample likely represents more sociocultural variability in the expression of depression, because those who did not speak English fluently were included in the present study but not in the Uebelacker et al. (2009) study. Another contribution of this study is that prior research indicating less psychologization in Asian s compared with European s was based largely on clinical populations. The current study provides complementary evidence using a nationally representative, community sample of noninstitutionalized populations. This is one of a few studies using a nationally representative community sample of Asian and European s that examines both somatic and psychological symptoms. Moreover, with the exception of Uebelacker et al. (2009) and Dere et al. (2013), the prior clinical and community studies (e.g., Weiss et al., 2009) only carried out frequency (i.e., classical test theory) analyses and did not include DIF analyses, which take into account degree of depressive symptomatology. The findings also have potential clinical implications. Keeping in mind this pattern of less somatization and psychologization in Asian s may assist clinicians in the detection and assessment of depression. Although the DSM is heavily weighted toward psychological symptoms (i.e., more than 50% of the DSM criteria are psychologically minded), it may be important to consider that it may take an Asian who has more depressive symptomatology to endorse some specific psychological symptoms than a European client. These findings also challenge the notion that Asian s, at least in a community sample, somatize their depression. Being cognizant of these patterns has the potential to assist clinicians in better detecting depression in Asian s who are experiencing the disorder. Limitations and Future Directions Because of the nature of the survey, this dataset did not allow us to examine depressive symptoms in the entire sample, as only those who passed the screening items received the depression module. Thus, these results may only apply to those who are already more elevated in depressive symptoms and not a true community sample. It will be important to replicate these findings in future studies, using both an entire community and clinical sample. Second, there may be important subgroup differences, within the broad ethnic categories. For example, some research suggests that Korean female participants report more somatic symptoms than Japanese female participants, but that somatic symptoms account for less variance in Beck Depression Inventory scores for the Korean participants than for the Japanese participants (Arnault & Kim, 2008). Our intent was not to apply a broad brush in examining Asian s and European s as two groups, but the current project was an initial study on ethnic differences, and thus, we did not examine subgroup differences within Asian s and within European s. There might be variability within these two groups, because of nationality, acculturation, language, country of birth, gender, age, and so forth. These would be interesting and important areas of future research, especially as the neural data begin to show some of these differences. Another limitation is that we only used depressive symptoms defined by the DSM IV. This may be a narrow lens by which to identify depression in our two ethnic groups, and thus, it may contribute to the detection of few group differences. It may be that group differences in the expression of depression would be more easily identified if culture-specific manifestations of depression were included in the assessment. Conclusions Although the somatization hypothesis has been popular in past theory and research of symptom expression in racial and ethnic minority groups, more recent studies, particularly those using community samples, do not support this hypothesis. For the majority of both somatic and psychological symptoms of depression in Asian and European s, there were no differences in level of endorsement. Where there were differences, we found less somatization in Asian s than in European s, contrary to the somatization hypothesis. In addition, similar to Ryder et al. (2008), when there were differences, we found less psychologization in Asian s than in European s. These results were supported by IRT analyses as well, which suggest that the observed ethnic differences are not an artifact of less depressive symptomatology among Asian s compared with European s. Thus, the presumed cultural differences in the expression of depression, especially the notion that Asian s tend to present depression in somatic terms, receive little support in a nationally representative community sample of Asian s and European s. References Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Arlington, VA: Psychiatric Publishing. Arnault, D. S., & Kim, O. (2008). Is there an Asian idiom of distress? Somatic symptoms in female Japanese and Korean students. Archives of Psychiatric Nursing, 22, doi: /j.apnu Birnholz, J. L., & Young, M. A. (2012). Differential item functioning for lesbians, bisexual, and heterosexual women in the Center for Epidemiological Studies Depression Scale. Assessment, 19, doi: /

10 ETHNICITY AND DEPRESSIVE SYMPTOMATOLOGY 763 Camilli, G., & Shepard, L. A. (1994). Methods for identifying biased test items. Thousand Oaks, CA: Sage. Chan, B., & Parker, G. (2004). Some recommendations to address depression in Chinese people in Australasia. Australian and New Zealand Journal of Psychiatry, 38, doi: /j x Cheng, T. A. (1989). Symptomatology of minor psychiatric morbidity: A cross-cultural comparison. Psychological Medicine, 19, doi: /S Cheung, F. M. (1982). Psychological symptoms among Chinese in urban Hong Kong. Social Science & Medicine, 16, doi: / (82) Chung, K. F., & Wong, M. C. (2004). Experience of stigma among Chinese mental health patients in Hong Kong. Psychiatric Bulletin, 28, doi: /pb Dere, J., Sun, J., Zhao, Y., Persson, T. J., Zhu, X., Yao, S., Bagby, R. M., & Ryder, A. G. (2013). Beyond somatization and psychologization : Symptom-level variation in depressed Han Chinese and Euro-Canadian outpatients. Frontiers in Psychology, 4, Retrieved from doi: / fpsyg Embretson, S. E., & Reise, S. P. (2000). Item Response Theory for psychologists. Hillsdale, NJ: Erlbaum, Inc. Gee, G. C., Ro, A., Gavin, A., & Takeuchi, D. (2008). Disentangling the effects of racial and weight discrimination on body mass index and obesity among Asian s. Journal of Public Health, 98, doi: /ajph Goldberg, D. P., & Bridges, K. (1988). Somatic presentations of psychiatric illness in primary care setting. Journal of Psychosomatic Research, 32, Retrieved from doi: / (88) Holland, P. W., & Wainer, H. (Eds.). (1993). Differential item functioning. Hillsdale, NJ: Erlbaum, Inc. Huang, F. Y., Chung, H., Kroenke, K., Delucchi, K. L., & Spitzer, R. L. (2006). Using the Patient Health Questionnaire-9 to measure depression among racially and ethnically diverse primary care patients. Journal of General Medicine, 21, doi: /j x Immordino-Yang, M. H., Yang, X., & Damasio, H. (2014). Correlations between social-emotional feelings and anterior insular activity are independent from visceral states but influenced by culture. Frontiers in Human Neuroscience, 8. doi: /fnhum Kadir, N. B. A., & Bifulco, A. (2010). Malaysian moslem mothers experience of depression and service use. Culture, Medicine, and Psychiatry, 34, doi: /s x Kirmayer, L. J. (2001). Cultural variations in the clinical presentation of depression and anxiety: Implications for diagnosis and treatment. Journal of Clinical Psychiatry, 62 (Suppl. 13), Retrieved from Kleinman, A. (1982). Neurasthenia and depression: A study of somatization and culture in China. Culture, Medicine, and Psychiatry, 6, doi: /bf Kleinman, A. (1988). Rethinking psychiatry: From cultural category to personal experience. New York, NY: The Free Press. Leu, J., Yen, I., Gansky, S., Walton, E., Adler, N., & Takeuchi, D. (2008). The association between subjective social status and mental health among Asian immigrants: Investigating the influence of age at immigration. Social Science & Medicine, 66, doi: /j.socscimed Lin, T.-Y., Tardiff, K., Donetz, G., & Goresky, W. (1978). Ethnicity and patterns of help-seeking. Culture, Medicine, and Psychiatry, 2, Retrieved from doi: /bf Linacre, J. M., & Wright, B. D. (1986). Item bias: Mantel-Haenszel and the Rasch model (MESA Psychometric Laboratory, Memorandum No. 39). Chicago, IL: University of Chicago, Department of Education. Markus, H. R., & Kitayama, S. (1991). Culture and the self: Implications for cognition, emotion, and motivation. Psychological Review, 98, Retrieved from doi: / x Marques, L., Alegria, M., Becker, A., Chen, C., Fang, A., Chosak, A., & Belo Diniz, J. (2011). Comparative prevalence, correlates of impairment, and service utilization for eating disorders across U.S. ethnic groups: Implications for reducing ethnic disparities in health care access for eating disorders. International Journal of Eating Disorders, 44, doi: /eat Parker, G., Cheah, Y. C., & Roy, K. (2001). Do the Chinese somatize depression? A cross-cultural study. Social Psychiatry and Psychiatric Epidemiology, 36, doi: /s Parker, G., Gladstone, G., & Chee, K. T. (2001). Depression in the planet s largest ethnic group: The Chinese. The Journal of Psychiatry, 158, doi: /appi.ajp Ryder, A. G., Yang, J., Zhu, X., Yao, S., Yi, J., Heine, S. J., & Bagby, R. M. (2008). The cultural shaping of depression: Somatic symptoms in China, psychological symptoms in North America? Journal of Abnormal Psychology, 117, doi: / x Santor, D. A., Ramsay, J. O., & Zuroff, D. C. (1994). Nonparametric item analyses of the Beck Depression Inventory: Evaluating gender item bias and response option weights. Psychological Assessment, 6, doi: / Simon, G. E., & Von Korff, M. (2006). Medical comorbidity and validity of DSM IV depression criteria. Psychological Medicine, 36, doi: /s Takeuchi, D. T., Alegría, M., Jackson, J., & Williams, D. (2007). Immigration and mental health: Diverse findings in Asian, Black, and Latino populations. Journal of Public Health, 97, doi: /AJPH Uebelacker, L. A., Strong, D., Weinstock, L. M., & Miller, I. W. (2009). Use of item response theory to understand differential functioning of DSM IV major depressive symptoms by race, ethnicity, and gender. Psychological Medicine, 39, doi: /s Weiss, B., Tram, J. M., Weisz, J. R., Rescorla, L., & Achenbach, T. M. (2009). Differential symptom expression and somatization in Thai versus U.S. children. Journal of Consulting and Clinical Psychology, 77, doi: /a Yang, F. M., & Kao, S. T. (2014). Item response theory for measurement validity. Shanghai Archives of Psychiatry, 26, doi: /j.issn Yen, S., Robins, C. J., & Lin, N. (2000). A cross-cultural comparison of depressive symptom manifestation: China and the United States. Journal of Consulting and Clinical Psychology, 68, doi: / x Yeung, A. S., & Chang, D. F. (2002). Adjustment disorder: Intergenerational conflicts in a Chinese immigrant family. Culture, Medicine, and Psychiatry, 26, doi: /a: Received October 23, 2013 Revision received September 4, 2014 Accepted September 5, 2014