Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Social Science & Medicine 56 (2003) 851–862 Ethnic differences in choices of health information by cancer patients using complementary and alternative medicine: an exploratory study with correspondence analysis Hisako Kakai, Gertraud Maskarinec*, Dianne M. Shumay, Yvonne Tatsumura, Katsuya Tasaki Cancer Research Center of Hawaii, University of Hawaii at Manoa, 1236 Lauhala Street, Honolulu, HI 96813, USA Abstract This study examined patterns in the use of health information among Caucasian, Japanese, and non-Japanese Asian Pacific Islander cancer patients in Hawaii and explored the relation of ethnicity and educational level to choices of health information sources. Information from 140 cancer patients, most of whom were users of complementary and alternative medicine (CAM), was analyzed using correspondence analysis. Three clusters of health information pertinent to the three ethnic groups emerged from the data. The results of this study revealed that Caucasian patients preferred objective, scientific, and updated information obtained through medical journals or newsletters from research institutions, telephone information services, and the internet. Japanese patients relied on media and commercial sources including television, newspapers, books, magazines and CAM providers. Non-Japanese Asians and Pacific Islanders used information sources involving person-to-person communication with their physicians, social groups, and other cancer patients. A higher educational level was closely related to a cluster of health information stressing objective, scientific and updated information, while a lower educational level was associated with interpersonally communicated information. The three ethnicity-specific patterns of health information use remained relatively stable at different educational levels, implying that the effect of patients’ ethnicity overrides their educational level in shaping their choices of health information. The results of this study indicate the importance of recognizing cancer patients’ culturally developed world views when understanding their health information-seeking behavior. For medical practice, these findings indicate the need for healthcare providers to assist cancer patients to obtain accurate health information in a culturally sensitive way. r 2002 Elsevier Science Ltd. All rights reserved. Keywords: Cancer; Health information; Ethnicity; Complementary medicine; Japanese-Americans; Asians; Pacific Islanders Introduction Cancer patients use various sources of health information to gain knowledge about their illness and prognosis, treatment options and side effects, ways to prevent recurrence, and psychological resources for coping (Cassileth, Zupkis, Sutton-Smith, & March, 1980; Fallowfield, Ford, & Lewis, 1994). These include, but are not limited to, healthcare providers (e.g., *Corresponding author. Tel.: +1-808-586-3078; fax: +1808-586-2984. E-mail address: [email protected] (G. Maskarinec). physicians and nurses), publications (e.g., books and magazines), phone assistance (e.g., Cancer Information Service), newspapers, television, radio, and word of mouth (Carlsson, 2000; Freimuth, 1993; Guidry, Aday, Zhang, & Winn, 1998; Ward, Anderson, Pundik, Redrick, & Kaufman, 1993; Worsley, 1989). Worsley (1989) categorized sources of health information into three major groups, ‘‘formal sources’’ (healthcare providers), ‘‘informal sources’’ (friends and relatives), and ‘‘commercial and media sources’’ (TV commercials, newspaper, and magazine articles). In a different approach, Carlsson (2000) distinguished an ‘‘active information-seeking strategy’’ (internet, medical books, 0277-9536/03/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 2 7 7 - 9 5 3 6 ( 0 2 ) 0 0 0 8 6 - 2 852 H. Kakai et al. / Social Science & Medicine 56 (2003) 851–862 narratives, and telephone information lines) and a ‘‘passive information-seeking strategy’’ (television, radio, newspapers, other cancer patients, and friends). Demographic variables associated with informationseeking behavior include age (Carlsson, 2000; TurkCharles, Meyerowitz, & Gatz, 1997; Worsley, 1989), gender (Freimuth, 1993; Rimer, Catoe, Graves, Burklow, & Anderson, 1993), education (Brown et al., 1993; Carlsson, 2000; Freimuth, 1993; Guidry et al., 1998; Jubelirer, Linton, & Magnetti, 1994; Turk-Charles et al., 1997; Ward et al., 1993), and ethnicity (Freimuth, 1993; Guidry et al., 1998; Rimer et al., 1993; Ward et al., 1993). The latter appears to be a critical contributor in shaping health information-seeking behavior although all ethnic groups perceive healthcare professionals as the most helpful source of information (Freimuth, 1993; Guidry et al., 1998; Rimer et al., 1993; Ward et al., 1993; Pennbridge, Moya, & Rodrigues, 1999). For example, Caucasian patients were more likely to use printed materials such as books and reference materials, whereas African-American patients tended to rely on pamphlets and television when seeking information about treatment options and side effects (Guidry et al., 1998). Use of the Cancer Information Service was dominated by Caucasians and underutilized by racial and ethnic minorities (Freimuth, 1993; Rimer et al., 1993; Ward et al., 1993). Educational background may also influence people’s health information-seeking behavior in combination with ethnicity (Brown et al., 1993; Carlsson, 2000; Guidry et al., 1998; Jubelirer et al., 1994; TurkCharles et al., 1997). For example, higher educational attainment may be responsible for the greater use of newspapers among Asians and Pacific Islanders (APIs) (Ward et al., 1993). Recently, with the advancement of information technology, the use of the internet as a source of health information has been increasing (Elliott & Elliott, 2000), particularly among cancer patients with higher education and younger age (Carlsson, 2000; Harris, 1998; Pennbridge et al., 1999). Whereas computer penetration and on-line access has been low among Hispanic- and African-Americans (24%), it has been high among APIs (51%). However, a breakdown for the individual ethnic groups within APIs is not available (National Telecommunication and Information Administration, 2000). The significant influences of culture on health behaviors have been discussed by scholars of various fields (Helman,1994; Kreps & Kunimoto, 1994; Spector, 2000). Although multiple definitions of ‘culture’ exist, many agree that the concept of culture refers to a set of guidelines, which include values, beliefs, attitudes, and behaviors, shared by members of a particular group (Ember & Ember, 1985; Helman, 1994; Matsumoto, 2000). The choice of guidelines is often influenced by individuals’ membership in various cultural groups defined by nationality, ethnicity, age, gender, and socioeconomic status (Kreps & Kunimoto, 1994). Internal locus of control (Rotter, 1954), which is the belief that individuals have control over events happening in their environments, is one concept that characterizes a tendency of individuals with Western cultural backgrounds who hold individualistic values (Dyal, 1984; Hamid, 1994). Individuals with non-Western cultural backgrounds, who generally cherish interpersonally oriented cultural values (Markus & Kitayama, 1991; Triandis, 1989), have been often described as possessing spiritual–social (Edman & Kameoka, 1997) or fatalistic (Domino & Lin, 1991; Domino & Lin, 1993) attributions to illness. This cognitive style is likely to be shaped by a world view that is ‘‘less dialectical, logical, or linguistic in character, less analytical, more synthetic, or more sensitive to family or community consensus than to individual autonomy, more virtue-based than principle-based’’ (Pellegrino, 1992, p. 14). The present study explored patterns of health information use among cancer patients of diverse ethnicity in Hawaii. Fifty-one percent of the state’s population are APIs, including Hawaiian/part-Hawaiian, Japanese, Chinese, Filipino, Korean, and Samoan (Hawaii State DBEDT’s Research and Economic Analysis Division, 2000). With 17% of the state’s population reporting exclusively Japanese ancestry and an additional 8% reporting some Japanese ancestry, Japanese-Americans represent the largest API group. Immigration from Japan to Hawaii occurred primarily during 1868–1924, when more than 200,000 Japanese came to work on plantations (Nordyke, 1989). JapaneseAmericans have been considered unique because of their rapid acculturation into mainstream America during and after World War II (Kitano, 1969). The pressure to assimilate and demonstrate their loyalty as ‘‘good Americans’’ during the war accelerated Japanese-Americans’ becoming very American (Ina, 1997). Compared to other APIs, more Japanese-American families speak only English and a smaller percentage is foreign-born (Uba & Sue, 1991). A number of cultural concepts are important to understanding behavioral patterns among Japanese-Americans. Confucianism which has greatly influenced Japanese culture (DeVos, 1985; Johnson & Marsella, 1978; Kitano, 1969; Marsella, 1993), emphasizes conformity, obedience, and the importance of legitimacy and authority associated with social roles and hierarchy. Therefore, Japanese-American patients, particularly elders, often defer to their physicians for medical decision-making (Saldov, Kakai, McLaughlin & Thomas, 1998), an accepted practice in traditional Japanese culture. At the same time, Japanese-Americans’ enryo, ‘‘modesty in the presence of one’s superior’’ (Kitano, 1969, p. 102) or ‘‘denial of self-importance’’ (Johnson, 1985, p. 123), discourages assertiveness in front of authorities such as physicians. H. Kakai et al. / Social Science & Medicine 56 (2003) 851–862 The major objective of this study was to investigate ethnic differences in health information-seeking behaviors among cancer patients in Hawaii who had participated in a survey on complementary and alternative medicine (CAM). In addition, we explored a possible association between patients’ education and ethnicity and choice of health information. Methods Participants Information for this analysis came from a larger study with 1168 patients who had participated in a mail survey on the use of CAM. As described in detail elsewhere (Maskarinec, Shumay, Kakai, & Gotay, 2000), the study subjects were identified through a state-wide populationbased cancer registry. All patients diagnosed with invasive cancer during 1996 and a subgroup of the 1995 cases received a questionnaire asking for selfreported ethnicity and types of alternative therapies for cancer (‘‘What kind of alternative therapies for cancer have you tried?’’). The following choices were offered: chiropractic, acupuncture, massage/body work, homeopathic medicine, relaxation techniques, hypnosis, yoga, religious healing or prayer, support groups, herbal/ botanical treatment, osteopathy, traditional Hawaiian healing, energy healing, meditation/mindfulness techniques, guided imagery/visualization, biofeedback, Tai Chi or Chi Gong, psychological counseling, special diet, megadoses of vitamins/other supplements, and other alternative therapy. We classified everyone who chose at least one of the categories as a CAM user. Although the response rate was only 47.6%, age, gender, place of residence, and ethnic distribution of the respondents was very similar to the distribution of all cancer patients recorded in the tumor registry. Only Filipino and Native Hawaiian patients responded at a slightly lower rate than the other groups. From the 439 patients who volunteered for an interview by indicating their interest on the questionnaire, we selected 143 individuals using a stratified sampling technique that attempted to include a variety of cancer sites, demographic backgrounds, residential areas, and types of CAM used. Due to our primary interest in interviewing CAM users, the majority of interviewees (76%) were those who had reported in the survey that they had used CAM for their cancers (Table 1). A smaller number (24%) of patients reporting no CAM use were interviewed to gain insight into their thinking about cancer and its treatment. Three patients were excluded from the present analysis because they did not provide information regarding their sources of health information. The initial survey study as well as the subsequent interview study were approved by the Hawaii Cancer Commis- 853 sion, the Committee of Human Studies at the University of Hawaii, and the Institutional Review Board of Kaiser Permanente. Interview procedure All interviews were conducted by two members of the research team at a location of the patients’ choice. All conversations were tape recorded with the patients’ permission and later transcribed verbatim. During the interview, the patient was asked various questions regarding his or her medical and psychological experiences with cancer and its treatment, including both conventional medicine and CAM. One of the questions specifically asked the patient to identify the primary sources of health information that he or she had used since receiving the cancer diagnosis. Health information was defined as any information related to the patient’s health including cancer, cancer treatment and prevention. Data coding The transcribed interviews were coded using a qualitative analysis software package, called NUD*IST 4 (N4) or Non-numerical Unstructured Data Indexing, Searching and Theorizing (Qualitative Solutions and Research Pty Ltd., 1997). We used N4 to develop a coding and classification system for type of CAM therapies, sources of health information, and other required variables by searching for a character or a sequence of characters and looking for key-word cooccurrences in the text. Sources of health information were coded into nine categories, including information provided by physicians (MDINF), telephone cancer information services (TELEP), professional references such as medical journals or newsletters from research institutions (PROF), television (TV), printed media materials such as newspapers, books, and magazines available at regular bookstores (PRINTM), CAM providers such as practitioners and product distributors (e.g., health food stores) (CAMPRO), family, friends, and co-workers (PEOPLE), other cancer patients (CAPT), and the internet (INTERNET). We assigned 1 to health information sources mentioned by patients and 0 otherwise. In a situation where the patient used one source of information with which s/he became acquainted through another source of information, we assigned 1 to both sources. For example, if the patient mentioned using a book recommended by his/her friend, both PRINTM and PEOPLE were coded as 1. Selfreported ethnic identities of the interviewees were categorized as Caucasian, Japanese, and non-Japanese API, which included Chinese, Korean, Filipino, Hawaiian, Samoan, and others. For reasons described in the introduction, Japanese-Americans were categorized as a H. Kakai et al. / Social Science & Medicine 56 (2003) 851–862 854 Table 1 Demographic characteristics of the interviewed participants (N ¼ 140) Variables Categories Caucasian (N ¼ 65) Japanese (N ¼ 33) Non-Japanese APIa (N ¼ 42) Gender Male Female 25 (38.5%) 40 (61.5%) 11 (33.3%) 22 (66.7%) 14 (33.3%) 28 (66.7%) Age group o40 years 40–49 years 50–59 years 60–69 years 70–79 years 80+ years 4 (6.2%) 8 (12.3%) 17 (26.2%) 19 (29.2%) 14 (21.5%) 3 (4.6%) 3 4 7 7 9 3 (9.1%) (12.1%) (21.2%) (21.2%) (27.3%) (9.1%) 5 (11.9%) 6 (14.3%) 15 (35.7%) 6 (14.3%) 9 (21.4%) 1 (2.4%) Education o=12 years 13–16 years 16+ years 33 (50.8%) 16 (24.6%) 16 (24.6%) 19 (57.6%) 8 (24.2%) 6 (18.2%) 23 (54.8%) 14 (33.3%) 5 (11.9%) Cancer site Breast Gastrointestinal Prostate All Other Sites 19 (29.2%) 8 (12.3%) 13 (20.0%) 25 (38.5%) 13 (39.4%) 6 (18.2%) 9 (27.3%) 5 (15.2%) 13 (34.2%) 10 (23.8%) 5 (11.9%) 14 (33.3%) Cancer state Localized Advanced Unknown 40 (61.5%) 24 (36.9%) 1 (1.5%) 22 (66.7%) 10 (30.3%) 1 (3.0%) 26 (61.9%) 15 (35.7%) 1 (2.4%) CAM use Yes No 55 (84.6%) 10 (15.4%) 22 (66.7%) 11 (33.3%) 30 (71.4%) 12 (28.6%) Vitamins (44.6%)b Religion (44.6%)b Diet (40.0%) Relaxation (35.4%) Meditation (33.8%) Herbs (29.2%) Vitamins (43.8%) Religion (31.3%) Diet (25.0%) Relaxation (18.8%) Massage (15.6%) Herbs (12.5%) Religion (31.3%) Herbs (26.2%) Diet (23.8%) Relaxation (19.0%) Massage (16.7%)b Vitamins (16.7%)b Six most popular CAMs a Other non-Japanese Asians and Pacific Islanders (API) include 15 Hawaiians, 14 Chinese, 9 Filipinos, 2 Koreans, 1 Samoan, and 1 Mixed Race. b Equal rank within the same ethnic group. distinct group from other APIs. Information on educational achievement obtained during the interviews was categorized into high (>12 years) and low (o12 years) educational attainment according to the completion of high school. Correspondence analysis For this study, we used correspondence analysis, an inductive statistical technique that is suitable for exploring relations among categorical variables but not for hypothesis testing (Clausen, 1998). Like principal components analysis, correspondence analysis allows researchers to reduce a complex data matrix into a more simple one without losing meaningful information. A significant advantage of correspondence analysis is that it permits results to be presented graphically. Categories of variables with similar frequencies are placed as points clustered together in space; on the other hand, categories with dissimilar frequencies are displayed as points that are far away from each other. The three important concepts of correspondence analysis are: (a) profiles, (b) masses, and (c) chi-square distances (Greenacre, 1994). After transforming the frequencies of variables into percentages relative to the row or column totals, also called profiles, the analysis makes it possible to compare the magnitude of occurrence among categories directly. While profiles, considered to be mathematical vectors, explain the proportional frequencies of categories only relative to the group total, masses are values that adjust different numbers in the group totals and are obtained by dividing each group total by the grand total. Distances between points in space are computed as chi-square distances, which are equivalent to Euclidean distances with an adjustment for proportions in frequencies of categories using profiles. H. Kakai et al. / Social Science & Medicine 56 (2003) 851–862 In this study, we used correspondence analysis in order to explore (a) the relation between cancer patients’ ethnic backgrounds and their choices of health information sources, and (b) the association between the patients’ educational levels and ethnicity-specific health information-seeking patterns. In the latter analysis, our purpose was not to observe differences in the choice of health information by educational level within each ethnic group, but to examine the association between ethnicity-specific patterns of health information-seeking behaviors and levels of educational attainment. Correspondence analysis allowed us to examine associations among the variables of interest visually. Results The three ethnic groups did not differ significantly by age (Caucasian M ¼ 60:66; Japanese M ¼ 62:09; nonJapanese API M ¼ 57:24; Fð2; 137Þ ¼ 1:37; p ¼ ns) and educational level, with 1 being high school degree or less, 2 bachelor’s degree, and 3 graduate degree (Caucasian M ¼ 1:74; Japanese M ¼ 1:61; non-Japanese API M ¼ 1:57; F ð2; 137Þ ¼ 0:67; p=ns). The members of all three ethnic groups typically reported using two or three sources of health information (Caucasian M ¼ 2:34; Japanese M ¼ 2:15; non-Japanese API M ¼ 2:26; F ð2; 137Þ ¼ 0:44; p=ns). Among all three groups, printed media sources were reported as the most frequently used source of health information (67.1%), followed by physicians’ information (42.1%). However, the rank order and the extent of the use for the other seven sources of health information differed across the three ethnic groups (Tables 2 and 3). Ethnicity and health information sources The results of the initial correspondence analysis with 140 cancer patients displayed 11 eigenvalues, indicating 855 that 11 independent dimensions had been decomposed. To raise the interpretability of the results, we decided to retain the first two dimensions, which explained 31.35% of the total inertia, i.e., variance (Dimension 1: 16.25%; Dimension 2: 15.10%). Table 4 depicts column coordinates, eigenvalues, and percentages of inertias for the retained two dimensions. Coordinates explain the magnitude and direction of the contribution of points to dimensions. The following included types of health information that were positively loaded on Dimension 1: MDINF, TELEP, PROF, PEOPLE, and CAPT. The types of health information that were negatively loaded on Dimension 1 included TV, PRINTM, CAMPRO, and INTERNET. Non-Japanese API had a positive loading on Dimension 1, whereas Caucasians and Japanese had a negative loading on this dimension. TELEP, PROF, and INTERNET were the only information sources that were positively loaded on dimension 2. Japanese and non-Japanese API were negatively and Caucasians positively loaded on Dimension 2. Three clusters of health information in relation to ethnicity could be identified (Fig. 1). As illustrated graphically, INTERNET, TELEP, and PROF were found to be close to Caucasians, indicating an association between the three sources of health information and this ethnic group. CAPT, MDINF and PEOPLE seemed to be another cluster with which non-Japanese API appeared to have a close relationship. The last cluster included TV, PRINTM, and CAMPRO and appeared to be related to the Japanese group. The identification of these three clusters of information types and their associations with certain ethnic groups were consistent with the trends described in Table 2. For example, 21.5% and 26.2% of Caucasians utilized PROF and INTERNET, respectively, while the use of these two types of health information among Japanese and nonJapanese API patients was much lower. For Japanese patients, the use of PRINTM (78.8%) and TV (24.2%) Table 2 Frequency (%) of Caucasian, Japanese, and non-Japanese API cancer patients in Hawaii using different sources of health information Sources of health informationa Caucasian (n ¼ 65) Japanese (n ¼ 33) Non-Japanese APIb (n ¼ 42) All (N ¼ 140) Physicians Telephone information Professional information TV Printed materials CAM providers People in general Other cancer Patients Internet 25 (38.5%) 8 (12.3%) 14 (21.5%) 13 (20.0%) 44 (67.7%) 11 (16.9%) 17 (26.2%) 5 (7.7%) 17 (26.2%) 11 (33.3%) 1 (3.0%) 4 (12.1%) 8 (24.2%) 26 (78.8%) 6 (18.2%) 8 (24.2%) 3 (9.1%) 4 (12.1%) 23 (54.8%) 5 (11.9%) 5 (11.9%) 8 (19.1%) 24 (57.1%) 3 (7.1%) 12 (28.6%) 9 (21.4%) 6 (14.3%) 59 14 23 29 94 20 37 17 27 a b Total adds to more than 100 because of multiple answers. Non-Japanese Asians and Pacific Islanders (API) includes 15 Hawaiians, 14 Chinese, 9 Filipinos, 2 Koreans, 1 Samoan, and 1 Mixed Race. (42.1%) (10.0%) (16.4%) (20.7%) (67.1%) (14.3%) (26.4%) (12.1%) (19.3%) 856 H. Kakai et al. / Social Science & Medicine 56 (2003) 851–862 Table 3 Frequency (%) of Caucasian, Japanese, and non-Japanese API cancer patients in Hawaii at different educational levels using different sources of health information Sources of health informationa Physicians Telephone information Professional information TV Printed materials CAM providers People in general Other cancer patients Internet Caucasian Japanese Non-Japanese APIb All Lowc (n ¼ 33) High (n ¼ 32) Low (n ¼ 19) High (n ¼ 14) Low (n ¼ 23) High (n ¼ 19) Low (n ¼ 75) High (n ¼ 65) 14 (42.4%) 5 (15.2%) 7 (21.2%) 8 (24.2%) 20 (60.6%) 5 (15.2%) 9 (27.3%) 1 (3.0%) 7 (21.2%) 11 (34.4%) 3 (9.4%) 7 (21.9%) 5 (15.6%) 24 (75.0%) 6 (18.8%) 8 (25.0%) 4 (12.5%) 10 (31.3%) 7 (36.8%) 0 (0.0%) 2 (10.5%) 3 (15.8%) 14 (73.7%) 2 (10.5%) 6 (31.6%) 1 (5.3%) 1 (5.3%) 4 (28.6%) 1 (7.1%) 2 (14.3%) 5 (35.7%) 12 (85.7%) 4 (28.6%) 2 (14.3%) 2 (14.3%) 3 (21.4%) 13 (56.5%) 4 (17.4%) 3 (13.0%) 4 (17.4%) 14 (60.9%) 1 (4.3%) 5 (21.7%) 5 (21.7%) 4 (17.4%) 10 (52.6%) 1 (5.3%) 2 (10.5%) 4 (21.1%) 10 (52.6%) 2 (10.5%) 7 (36.8%) 4 (21.1%) 2 (10.5%) 34 (45.3%) 9 (12.0%) 12 (16.0%) 15 (20.0%) 48 (64.0%) 8 (10.7%) 20 (26.7%) 7 (9.3%) 12 (16.0%) 25 (38.5%) 5 (7.7%) 11 (16.9%) 14 (21.5%) 46 (70.8%) 12 (18.5%) 17 (26.2%) 10 (15.4%) 15 (23.1%) a Total adds to more than 100 because of multiple answers. Non-Japanese Asians and Pacific Islanders (API) includes 15 Hawaiians, 14 Chinese, 9 Filipinos, 2 Koreans, 1 Samoan, and 1 Mixed Race. c Education: low—o12 years; high—X12 years. b Table 4 Column coordinates, eigenvalues, and percentages of inertias for retained two dimensions Column coordinates Source of information Dimension 1 Dimension 2 Information from physicians Telephone services Professional references Television Printed materials CAM providers Family and friends Cancer patients Internet Caucasian Japanese Non-Japanese API Eigenvalue Inertia percentage 0.55 0.79 0.11 0.46 0.25 0.88 0.09 0.64 0.06 0.23 1.29 1.28 0.41 16.25 0.19 0.74 0.60 0.24 0.11 0.01 0.24 0.31 0.92 0.97 1.16 0.73 0.38 15.10 Note: The total inertia with 11 dimensions is 2.53. was noticeably high compared to the other two ethnic groups. MDINF and CAPT were the most popular sources of information among non-Japanese API (54.8% and 21.4%, respectively). The following statements by patients capture the ethnic patterns in health information seeking behavior and illustrate Caucasians’ preferences for scientific information sources, non-Japanese API patients’ reliance on interpersonal communication, and Japanese patients’ tendency to rely on media and commercial information. Caucasian patients: ‘‘The number one is, of course, books. American Cancer Society, they have, on the internet, they have the latest. The internet has the latest information, too, and I will be using that. The daily newspaper. As soon as something new is printed in weekly magazines or the daily newspaper, so I cut them out all the time’’ (a 50 year-old male patient with lung cancer). ‘‘Newsletters, I had the Harvard newsletter and reading current books. And I’m learning lots from my computer, which I’ve had for about a year and half now ‘‘(a 90 year-old female patient with breast cancer). Non-Japanese API patients: ‘‘Relatives and friends that said they know my condition, so they say this would be good [for my cancer]...Then you take them, cause I feel that they are not going to do anything detrimental to my health. They were good enough to go out and I know they have to pay for it. Why not include them? The only thing they can do is [to] help me’’ (a 56 year-old part-Hawaiian female patient with colon cancer). ‘‘Actually, I guess our doctors and our friends, cause we also verified who we chose based on our friends’ recommendations. Our doctors and our friends, but now that, since my son is closer to the medical field now, I ask him a lot too’’ (a 53 year-old Chinese female patient with gastrointestinal cancer). Japanese patients: ‘‘I think it [Coenzyme Q-10] was on TV one time. They mentioned something about it; that it was not H. Kakai et al. / Social Science & Medicine 56 (2003) 851–862 857 Fig. 1. Profiles of health information sources and ethnic groups in a two-dimensional plot (n ¼ 140). Sources of information include INTERNET (the internet), PROF (professional information), TELEP (telephone information services), MDINF (physicians’ information), CAPT (other cancer patients’ information), PEOPLE (information from social group), TV (television), PRINTM (printed materials available at regular bookstores), and CAMPRO (CAM providers). only good for the heart, but they had something to do with cancer’’ (a 61 year-old male patient with prostate cancer). ‘‘I just rely upon the resource information that comes through [Company X who distributes nutritional supplements]. They cover everything, they don’t just focus on cancer. I just use that as my nutritional information base. That is all’’ (a 45 year-old female patient with breast cancer). Educational level and health information sources After adding educational levels, the results of the analysis (Table 3) indicated that 27.4% of the total inertia was explained by the first and second dimensions (Dimension 1: 14.0%; Dimension 2: 13.4%). In comparison to the original plot with sources of health information and ethnicity alone (Fig. 2), the inclusion of educational level did not affect the structures of the three clusters very much. The only change was that CAPT shifted its position from negative to positive loading on Dimension 2. As indicated by the plot, high educational level was close in distance to a cluster of health information including TELEP, PROF, and INTERNET. On the other hand, low educational level was close to another cluster of health information sources including CAPT, MDINF, and PEOPLE. Discussion As a result of the analysis, three clusters of health information sources closely related to the three different ethnic groups emerged. Caucasian background was associated with the use of health information containing scientific, objective, and updated knowledge, in particular, telephone cancer information services, professional information in medical journals and newsletters from research institutions, and the internet. Patients of Japanese ancestry appeared in the same cluster as commercial and media information sources that included television, printed media materials, and CAM providers. Non-Japanese API appeared close in space to the cluster of information involving interpersonal communication, such as other cancer patients, family members, friends, and co-workers, and physicians. The inclusion of educational level did not greatly change the initial clusters of health information sources. High educational level was closely related to the type of 858 H. Kakai et al. / Social Science & Medicine 56 (2003) 851–862 Fig. 2. Profiles of health information sources, ethnic groups, and educational levels in a two-dimensional plot (n ¼ 140). Sources of information include INTERNET (the internet), PROF (professional information), TELEP (telephone information services), MDINF (physicians’ information), CAPT (other cancer patients’ information), PEOPLE (information from social group), TV (television), PRINTM (printed materials available at regular bookstores), and CAMPRO (CAM providers). health information emphasizing objectivity and scientific evidence, while low educational level appeared close to the health information sources valuing word-of-mouth information given by the social group. Consistent with previous research findings, this study observed frequent use of the internet, professional health information (Brown et al., 1993; Carlsson, 2000; Guidry et al., 1998; Jubelirer et al., 1994; Turk-Charles et al., 1997), and telephone information services (Freimuth, 1993; Rimer et al., 1993; Ward et al., 1993) among Caucasian patients. The latter may be more effective for Caucasians because they are more open than other ethnic groups about disclosing personal problems and discussing them with strangers (Sue & Sue, 1990). These observations support the idea of an ‘‘active informationseeking strategy’’ (Carlsson, 2000) and reflect a desire to seek objective, scientific, and updated information, attributable to the Western epistemological emphasis on ‘‘empiricism, objectivity, quantification of experience, and attempting to control nature’’ (Pellegrino, 1992, p. 14). In addition, Western cultural values of independence, self-reliance, and self-determination (Johnson, 1985; Markus & Kitayama, 1991; Triandis, 1989) and the internal locus of control (Rotter, 1954) manifested in a desire to have some control over illness through active health information-seeking may contribute to the pattern noted among Caucasian patients. Unlike Caucasian patients, many of the non-Japanese APIs engaged in a ‘‘passive information-seeking strategy’’ (Carlsson, 2000), in which a patient’s social group plays a significant role in providing health information. Although we have no supporting data, we speculate that many non-Japanese APIs may see a minimal connection between actively seeking health information and controlling their illness because they do not believe that their own actions may greatly affect the outcome of their disease (Domino & Lin, 1991, 1993; Edman & Kameoka, 1997). Consequently, non-Japanese APIs may be less likely to seek information beyond what is given by their physicians or social groups. Patients with non-Japanese API backgrounds relied on physicians’ information more heavily than the other two ethnic groups. However, this preference for information from physicians does not rule out distrust towards the standard Western healthcare system due to perceived cultural incompetency, a frequently described issue among API populations (Braun, Look, & Tsark, 1995; Gotay, Issell, Hernandez, & Serxner, 1996; Uba & Sue, 1991). Because physicians’ information is generally readily available and orally communicated, the preference of receiving H. Kakai et al. / Social Science & Medicine 56 (2003) 851–862 information from physicians could be interpreted as an indication of API’s choice of person-to-person communication based on interpersonally oriented cultural values (McLaughlin & Braun, 1998; Uba & Sue, 1991). This experience-inductive approach among non-Japanese API patients, as opposed to the fact-inductive approach preferred by many Caucasian patients, may allow non-Japanese APIs to draw conclusions from the subjective experiences of people in their social group. For example, anecdotes about other cancer patients who coped successfully with cancer may be more persuasive than scientific evidence offered by professionals. The perceived qualities and trustworthiness of the individual who offers the information may be more important to a person holding collectivistic values than the quality of the information provided. Japanese patients greatly relied on ‘‘commercial and media sources’’ (Worsley, 1989) suggesting a preference for objective or scientific information over interpersonally acquired health information. In contrast to Caucasian patients, Japanese-Americans were less likely to use telephone information services, the internet, and professional information, possibly as a result of traditional Japanese values. Japanese-Americans may prefer solving personal problems within the family structure to disclosing them to strangers (Ina, 1997; Marsella 1993). Although there are generational differences, the stigma attached to cancer still makes it difficult for Japanese-Americans to discuss their illnesses openly. For example, Saldov et al. (1998) found that JapaneseAmerican elderly cancer patients had a resistance toward uttering the word ‘‘cancer’’. This observation was partly attributed to the elderly patients’ concerns about self-fulfilling prophecy through speech, as originated in the Japanese indigenous religious belief of kegare (Long & Long, 1982) and a concern of disclosing a family history of cancer because it may affect the offsprings’ chances of getting married (Saldov et al., 1998). The skepticism towards the internet may be a result of the importance attached to authority (DeVos, 1985; Johnson & Marsella, 1978; Kitano, 1969; Marsella, 1993). Similarly, self-effacing behavior, which discounts positive aspects of one’s own attributes, (Akimoto & Sanbonmatsu, 1999) may prevent Japanese-American patients from considering themselves legitimate users of professional information, such as medical journals. It is interesting to note that physicians’ information was least utilized by Japanese-American patients, despite their tendency to value authorities’ information. Perhaps, Japanese-Americans may hesitate to ask many questions, demand more information, or show assertiveness to avoid offending their physicians. The frequent use of vitamins and nutritional supplements among Japanese-American patients (Table 1) may explain the high utilization of information from CAM providers that is often found in health magazines 859 (Philen, Ortiz, Auerbach, & Falk, 1992), at health food stores (Calder, Issenman, & Cawdron, 2000; Gotay & Dumitriu, 2000), on TV, or in newsletters from distributors of these products. As to the association between education and ethnicity, it was unclear from the present analysis whether the identified patterns of health information use were due to ethnicity or due to education. Nonetheless, our data suggested that the observed ethnicity-specific trends were largely consistent across different educational levels with a few exceptions at a low educational level, where we observed a lower use of telephone information services among Caucasians than other non-Japanese APIs and a lower use of TV among the Japanese than all others (Table 3). This indicates that the effect of cultural beliefs overrides educational attainment in shaping cancer patients’ health information seeking behaviors. Valuing objective, scientific, and factual information for drawing conclusions is a predominant pattern of thought in the United States (Lieberman, 1994), and nurturing this epistemic mode among students is emphasized in the American educational system (Serpell & Hatano, 1997), an institution that largely reflects white middle class values (Heath, 1989). As a result, it seems reasonable that we found a close association between high educational attainment and types of health information emphasizing objectivity, scientific evidence, and updated knowledge among Caucasians. Selection bias arising from the fact that only a proportion of survey participants volunteered for an interview may limit the generalizability of this study. However, the population-based recruitment strategy included all cancer cases diagnosed in the state during one year and provided for the inclusion of diverse cancer patients. The results should also be interpreted cautiously because of the exploratory nature of correspondence analysis; it is not designed to make statistical decisions. Nonetheless, the fact that our results are largely consistent with previous research findings and logically congruent with cross-cultural theories strengthens our interpretation. This study highlights CAM users’ sources of health information—important knowledge in light of consistently rising CAM use (Eisenberg et al., 1998; Ernst & Cassileth, 1998)—and adds new information about APIs to a literature primarily focused on comparisons between Caucasians and African-Americans (Freimuth, 1993; Guidry et al., 1998). It is important to understand the critical role that cultural values and beliefs play in shaping cancer patients’ health information seeking behaviors because patients’ choices of health information may affect their survival and quality of life. Barriers due to the cultural incompatibility of healthcare providers and patients in oncology practice may be problematic when negotiating the course of cancer treatments. Our findings may help healthcare providers to develop educational 860 H. Kakai et al. / Social Science & Medicine 56 (2003) 851–862 interventions that enhance patients’ understanding of cancer and its treatment. Healthcare providers need to make conscious efforts to eliminate cultural barriers facing cancer patients in their acquisition of health information. They should avoid stereotypes and offer diverse resources to all patients while discussing the benefits and limitations of each. Despite a commonly observed lack of communication, an active exchange of health information between patients and their physicians is essential and should be encouraged. Physicians should remind their patients that it is all right to ask questions or to share their ideas or desires rather than deferring all medical decisions to their physicians. Physicians may also need to encourage patients, Japanese-American patients in particular, to speak about their CAM use rather than assuming physicians’ lack of knowledge in CAMs or unwillingness to discuss them (Tasaki, Maskarinec, Shumay, Tatsumura, & Kakai, in press). All cancer patients, regardless of their ethnicities, may need to be informed about the availability of a variety of health information sources. Educational brochures and cancer seminars in the health care facilities may be of value. Patients and their families with strong interpersonal values, often of non-Caucasian backgrounds, may particularly benefit from the establishment of trusting and personal relationships with healthcare providers or health educators while gaining accurate information about cancer and general health through such educational activities. These educational efforts at a community level may also alleviate rising issues involving the digital divide in today’s highly technological society. Acknowledgements We would like to show our appreciation to all participants for taking the time and effort to respond to our questionnaire and to participate in the interviews. The help of Carolyn Gotay, Ph.D., Marc Goodman, Ph.D. and the staff of the Hawaii Tumor Registry are greatly appreciated. We would also like to thank our research team, including Tammy Brown, Carole Prism, and David Henderson for their help with transcription and analysis. This research was supported by a special study grant from the National Cancer Institute, Surveillance, Epidemiology, and End Result program under contract number N01-PC67001 and by a medical research grant from the Hawaii Community Foundation. References Akimoto, S. A., & Sanbonmatsu, D. M. (1999). Differences in self-effacing behavior between European and Japanese Americans: Effect on competence evaluations. Journal of Cross-Cultural Psychology, 30, 159–174. Braun, K. L., Look, M. A., & Tsark, J. A. (1995). High mortality rates in native Hawaiians. Hawaii Medical Journal, 54(9), 723–729. Brown, P., Ames, N., Mettger, W., Smith, T. J., Gramarossa, G. L., Friedell, G. H., & McDonald, S. S. (1993). Closing the comprehension gap: Low literacy and the cancer information service. Journal of National Cancer Institute Monograph, 14, 157–163. Calder, J., Issenman, R., & Cawdron, R. (2000). Health information provided by retail health food outlets. Canadian Journal of Gastroenterology, 14(9), 767–771. Carlsson, M. (2000). Cancer patients seeking information from sources outside the health care system. Support Care Cancer, 8(6), 453–457. Cassileth, B. R., Zupkis, R. V., Sutton-Smith, K., & March, V. (1980). Information and participation preferences among cancer patients. Annals of Internal Medicine, 92(6), 832–836. Clausen, S. E. (1998). Applied correspondence analysis: An introduction. Thousand Oaks, CA: Sage. DeVos, G. (1985). Dimensions of the self in Japanese culture. In A. J. Marsella, & F. L. K. Hsu (Eds.), Culture and self: Asian and Western perspectives (pp. 141–183). New York: Tavistock Publications. Domino, G., & Lin, J. (1991). Images of cancer: China and the United States. Journal of Psychosocial Oncology, 9(3), 67–78. Domino, G., & Lin, W-Y. (1993). Cancer metaphors: Taiwan and the United States. International Journal of Psychology, 28(1), 45–56. Dyal, J. A. (1984). Cross-cultural research with the locus of control construct. In H. M. Lefcourt (Ed.), Research with the locus of control construct (pp. 209–306). New York: Academic Press. Edman, J. L., & Kameoka, V. A. (1997). Cultural differences in illness schemas: An analysis of Filipino and American illness attributions. Journal of Cross-Cultural Psychology, 28(3), 252–265. Eisenberg, D. M., Davis, R. B., Ettner, S. L., Appel, S., Wilkey, S., Van, R. M., & Kessler, R. C. (1998). Trends in alternative medicine use in the United States, 1990–1997: Results of a follow-up national survey. Journal of the American Medical Association, 280(18), 1569–1575. Elliott, B., & Elliott, G. (2000). High volume medical web sites. Delaware Medical Journal, 72(1), 21–29. Ember, C. R., & Ember, M. (1985). Anthropology (4th ed). Englewood Cliffs, NJ: Prentice-Hall. Ernst, E., & Cassileth, B. R. (1998). The prevalence of complementary/alternative medicine in cancer: A systematic review. Cancer, 83(4), 777–782. Fallowfield, L., Ford, S., & Lewis, S. (1994). Information preferences of patients with cancer. Lancet, 344(8936), 1576. Freimuth, V. S. (1993). Narrowing the cancer knowledge gap between whites and African Americans. Journal of National Cancer Institute Monograph, 14, 81–91. Gotay, C. C., & Dumitriu, D. (2000). Health food store recommendations for breast cancer patients. Archives of Family Medicine, 9(8), 692–699. Gotay, C. C., Issell, B. F., Hernandez, B. Y., & Serxner, S. (1996). Barriers to mammography in a low income, H. Kakai et al. / Social Science & Medicine 56 (2003) 851–862 multiethnic clinic population. Hawaii Medical Journal, 55(8), 136–140. Greenacre, M. (1994). Correspondence analysis and its interpretation. In M. Greenacre, & J. Blasius (Eds.), Correspondence analysis in the social science (pp. 3–22). San Diego, CA: Academic Press. Guidry, J. J., Aday, L. A., Zhang, D., & Winn, R. J. (1998). Information sources and barriers to cancer treatment by racial/ethnic minority status of patients. Journal of Cancer Education, 13(1), 43–48. Hamid, P. N. (1994). Assertiveness and personality dimensions in Chinese students. Psychological Reports, 75(1), 127–130. Harris, K. A. (1998). The informational needs of patients with cancer and their families. Cancer Practice, 6(1), 39–46. Hawaii State DBEDT’s Research and Economic Analysis Division (2000). The state of Hawaii data book. Honolulu, HI: Hawaii State Department of Business, Economic Development & Tourism. Available: http://www.hawaii.gov/dbedt/db00/index.html. Heath, S. B. (1989). Ways with words: language, life, and work in communities and classrooms. New York: Cambridge University Press. Helman, C. G. (1994). Culture, health and illness (3rd ed.). Jordan Hill, Oxford: Butterworth-Heinemann. Ina, S. (1997). Counseling Japanese Americans: from internment to repatriation. In C. C. Lee (Ed.), Multicultural issues in counseling (pp. 189–206). Alexandria, VA: American Counseling Association. Johnson, F. (1985). The western concept of self. In A. J. Marsella, & F. L. K. Hsu (Eds.), Culture and self’: Asian and Western perspectives (pp. 91–138). New York: Tavistock Publications. Johnson, F. A., & Marsella, A. J. (1978). Differential attitudes toward verbal behavior in students of Japanese and European ancestry. Genetic Psychology Monographs, 97, 43–76. Jubelirer, S. J., Linton, J. C., & Magnetti, S. M. (1994). Reading versus comprehension: Implications for patient education and consent in an outpatient oncology clinic. Journal of Cancer Education, 9(1), 26–29. Kitano, H. L. (1969). Japanese-americans: evolution of a subculture. Englewood Cliffs, NJ: Prentice-Hall. Kreps, G. L., & Kunimoto, E. N. (1994). Effective communication in multicultural health care settings. Thousand Oaks, CA: Sage. Lieberman, D. A. (1994). Ethnocognitivism, problem solving, and hemisphericity. In L. A. Samovar, & R. E. Porter (Eds.), Intercultural communication: A reader (pp. 178–193). Belmont, CA: Wadsworth. Long, S. O., & Long, B. D. (1982). Curable cancers and fatal ulcers. Attitudes toward cancer in Japan. Social Science and Medicine, 16(24), 2101–2108. Markus, H. R., & Kitayama, S. (1991). Culture and the self: Implications for cognition, emotion, and motivation. Psychological Review, 98, 224–253. Marsella, A. J. (1993). Counseling and psychotherapy with Japanese Americans: Cross-cultural considerations. American Journal of Orthopsychiatry, 63(2), 200–208. Maskarinec, G., Shumay, D. M., Kakai, H., & Gotay, C. C. (2000). Ethnic differences in complementary and alternative 861 medicine use among cancer patients. Journal of Alternative and Complementary Medicine, 6(6), 531–538. Matsumoto, D. (2000). Culture and psychology: People around the world (2nd ed.). Belmont, CA: Wadsworth/Thomson Learning. McLaughlin, L. A., & Braun, K. L. (1998). Asian and Pacific Islander cultural values: Considerations for health care decision making. Health Social Work, 23(2), 116–126. National Telecommunication and Information Administration. (2000). Falling through the net II: New data on the digital divide. Washington, DC: US Bureau of the Census, US Department of Commerce. Nordyke, E. C (1989). The peopling of Hawaii. Honolulu, HI: University of Hawaii Press. Pellegrino, E. D. (1992). Prologue: Intersections of western biomedical ethics and world culture. In E. Pellegrino, P. Mozzarella, & P. Corsi (Eds.), Transcultural dimensions in medical ethics (pp. 13–19). Frederick, MD: University Publishing Group. Pennbridge, J., Moya, R., & Rodrigues, L. (1999). Questionnaire survey of California consumers’ use and rating of sources of health care information including the Internet. Western Journal of Medicine, 171(5–6), 302–305. Philen, R. M., Ortiz, D. I., Auerbach, S. B., & Falk, H. (1992). Survey of advertising for nutritional supplements in health and bodybuilding magazines. Journal of the American Medical Association, 268(8), 1008–1011. Qualitative Solutions and Research Pty Ltd. (1997). QSR NUD*IST 4 User Guide. Victoria, Australia: SAGE Publications. Rimer, B. K., Catoe, K. E., Graves, C., Burklow, J., & Anderson, D. M. (1993). Older callers to the Cancer Information Service. Journal of the National Cancer Institute Monograph, 14, 165–169. Rotter, J. B. (1954). Social learning and clinical psychology. Englewood Cliffs, NJ: Prentice-Hall. Saldov, M., Kakai, H., McLaughlin, L., & Thomas, A. (1998). Cultural barriers in oncology: Issues in obtaining medical informed consent from Japanese-American elders in Hawaii. Journal of Cross-Cultural Gerontology, 13, 265–279. Serpell, R., & Hatano, G. (1997). Education, schooling, and literacy. In J. W. Berry, P. R. Dasen, & T. S. Saraswathi (Eds.), Handbook of cross-cultural psychology: Vol 2. Basic processes and human development (pp. 339–376). Boston, MA: Allyn & Bacon. Spector, R. E. (2000). Cultural diversity in health and illness (5th ed.). Upper Saddle River, NJ: Prentice-Hall. Sue, D. W., & Sue, D. (1990). Counseling the culturally different: Theory and practice (2nd ed.). New York: Wiley. Tasaki, K., Maskarinec, G., Shumay, D.M.Tatsumura, Y., & Kakai, H. (in press). Communication between physicians and cancer patients about complementary and alternative medicine: Exploring patients’ perspectives. PsychoOncology. Triandis, H. C. (1989). The self and social behavior in differing cultural contexts. Psychological Review, 96, 506–520. Turk-Charles, S., Meyerowitz, B. E., & Gatz, M. (1997). Age differences in information-seeking among cancer patients. 862 H. Kakai et al. / Social Science & Medicine 56 (2003) 851–862 International Journal of Aging and Human Development, 45(2), 85–98. Uba, L., & Sue, S. (1991). Nature and scope of services for Asian and Pacific Islander Americans. In N. Mokuau (Ed.), Handbook of social services for Asian and Pacific Islanders (pp. 3–19). Westport, CT: Greenwood Press. Ward, J. A., Anderson, D. M., Pundik, C. G., Redrick, A., & Kaufman, R. (1993). Cancer Information Service utilization by selected US ethnic groups. Journal of the National Cancer Institute Monograph, 14, 147–156. Worsley, A. (1989). Perceived reliability of sources of health information. Health Education Research, 4, 367–376.