Acculturation and Perceptions of a Good Death Among Japanese Americans and Japanese Living in the U.S.

Accepted Manuscript Acculturation and Perceptions of a Good Death among Japanese Americans and Japanese Living in the United States Masanori Mori, M.D., Yuichiro Kuwama, M.D., Takamaru Ashikaga, Ph.D., Henrique A. Parsons, M.D., M.Sc., Mitsunori Miyashita, R.N., Ph.D. PII:

S0885-3924(17)30415-3

DOI:

10.1016/j.jpainsymman.2017.08.010

Reference:

JPS 9552

To appear in:

Journal of Pain and Symptom Management

Received Date: 13 February 2017 Revised Date:

11 August 2017

Accepted Date: 12 August 2017

Please cite this article as: Mori M, Kuwama Y, Ashikaga T, Parsons HA, Miyashita M, Acculturation and Perceptions of a Good Death among Japanese Americans and Japanese Living in the United States, Journal of Pain and Symptom Management (2017), doi: 10.1016/j.jpainsymman.2017.08.010. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Acculturation and Perceptions of a Good Death among Japanese Americans and Japanese Living in the

RI PT

United States.

Masanori Mori, M.D.1, Yuichiro Kuwama, M.D.2, Takamaru Ashikaga, Ph.D.3, Henrique A. Parsons, M.D., M.Sc.4, Mitsunori Miyashita, R.N., Ph.D.5

Palliative Care Team, Seirei Mikatahara General Hospital, 2 Mount Sinai Beth Israel, Japanese Medical

SC

1

Practice, 3Department of Medical Biostatistics, University of Vermont College of Medicine, 4Department of

M AN U

Medicine/Division of Palliative Care, University of Ottawa, Ontario, Canada, 5Department of Palliative

Corresponding author: Masanori Mori, M.D.

TE D

Nursing, Health Sciences, Tohoku University Graduate School of Medicine

EP

Palliative Care Team, Seirei Mikatahara General Hospital

AC C

3453 Mikatahara-cho, Kita-ku, Hamamatsu, Shizuoka 433-8558, Japan Tel: +81-53-436-1251, Fax: +81-53-438-2971 [email protected]

Word counts: 3 tables; 24 references; 3,594 words

1

ACCEPTED MANUSCRIPT

ABSTRACT Context Acculturation is the phenomenon of the attitudinal changes of individuals who come into continuous contact

RI PT

with another culture. Despite the long history of Japanese immigration to America, little is known about the impact of acculturation on perceptions of a good death. Objectives

living in America (J/A), and the Japanese living in Japan (J/J).

M AN U

Methods

SC

To examine differences in perceptions of a good cancer death among Japanese Americans (JA/A), Japanese

We administered surveys among JA/A and J/A, and used historical J/J data for reference. Primary endpoint was the proportion of respondents who expressed the necessity of core and optional items of the Good Death Inventory. Group differences ≥20% were deemed clinically important. Results

TE D

441 survey responses in America and 2,548 in Japan were obtained. More than 80% of respondents consistently considered 9 of 10 core items necessary without significant group differences. No core item reached a ≥20% group difference. Three of the 8 optional items reached ≥20% group difference: ‘fighting against disease until one’s last moment’ (49%, p<0.0001; 52%, p<0.0001; and 73% in JA/A; J/A; and J/J;

EP

respectively), ‘knowing what to expect about one’s condition in the future’ (83%, p<0.0001; 80%, p<0.0001; and 58%, respectively), and ‘having faith’ (64%, p=0.0548; 43%, p=0.0127; and 38%, respectively).

AC C

Conclusions

While most core items of a good death were preserved throughout the levels of acculturation, perceptions of some optional items shifted away from Japanese attitudes as individuals became more acculturated. Understanding of different levels of acculturation may help clinicians provide culturally-sensitive end-of-life care.

2

ACCEPTED MANUSCRIPT

Keywords: Acculturation; Japanese American; good death; prognostic disclosure

AC C

EP

TE D

M AN U

SC

RI PT

Running Title: Acculturation and perception of a good death

3

ACCEPTED MANUSCRIPT

INTRODUCTION Achieving a good death is one of the essential goals of palliative care for cancer patients (1-3). As perceptions of a good death can vary markedly among different cultures, providing culturally-sensitive end-of-

RI PT

life (EOL) care that respects both cultural and individual values is critical to help terminally-ill patients achieve a good death (4-6).

As the population in western countries becomes more diverse, clinicians commonly face ethical

SC

dilemmas due to cultural variations in perceptions of a good death and preferences regarding EOL cancer care (4-6). Particularly in the United States (US), complex immigration patterns have resulted in various levels of

M AN U

acculturation among a single ethnicity. Acculturation is the phenomenon of the value, attitudinal, and behavioral changes of individuals who come into continuous contact with another culture (7). Variations in acculturation can affect health perceptions and behaviors including perceptions of a good death and end-of-life (EOL) decision making, complicating the understanding of cultural norms (8). An exploration of the association of acculturation with perceptions of a good death within a single ethnic population, such as

TE D

Japanese Americans, may serve as a model to understand the effect of acculturation in other ethnic groups from both western and eastern countries.

The long history of Japanese immigration to the US has resulted in more than 1.4 million Japanese Americans spanning multiple generations (9). While most Americans tend to value individualism,

EP

independence, and autonomy as individual, Japanese people generally value collectivism, inter-dependency, and autonomy as family, resulting in marked differences as well as similarities in perceptions of a good death

AC C

between Americans and Japanese (1-3, 10). Furthermore, multiple levels of acculturation among Japanese Americans can impact on their preferences regarding EOL cancer care, although little is known if the levels of acculturation could actually impact on their perception of a good death. Previous studies revealed that preference for autonomy and disclosure, willingness to forgo care, and views of advance care planning shift toward American values as Japanese Americans acculturate (8, 11). More acculturated Japanese Americans preferred prognostic disclosure, though how their preferences regarding physician’s initiating prognostic communication are influenced by acculturation is unknown (11).

4

ACCEPTED MANUSCRIPT

Prior nationwide studies in Japan identified 10 core, 8 optional, and 3 unique perceptions (‘dying as one sleeps’, ‘having a sudden death’ (‘pokkuri’) and ‘leaving the decision to a medical expert’ (‘omakase’) contributing to a good death among the general public (2, 3, 12). However, little is known whether these

RI PT

commonly-held perceptions in Japan would shift toward the American values as people of Japanese ancestry acculturate in the US. To establish individual goals of EOL care for Japanese Americans and Japanese

presently living in the US, it is imperative to understand what constitutes a good death as well as preferences regarding prognostic disclosure in this distinct population.

SC

Thus, the primary aim of our study was to explore differences in general perceptions of a good death among the Japanese Americans living in America (JA/A), Japanese living in America (J/A) relative to

M AN U

Japanese living in Japan (J/J). Our secondary aims were to examine differences in general perceptions of culturally specific Japanese items for a good death (‘dying as one sleeps’, ‘having a sudden death’, and ‘leaving the decision to a medical expert’), and prognostic disclosure preferences.

Participants and procedures

TE D

METHODS

Cross-sectional anonymous self-completion questionnaire surveys were conducted among individuals aged 20 years or older with Japanese ancestry living in the US (JA/A and J/A). Several modes were used to

EP

deliver surveys to maximize the representativeness of subjects with various levels of acculturation. First, we identified participants of the “Senior Week” event that took place in the New York City in September, 2010.

AC C

“Senior Week” was an annual, two-week event organized by the Japanese American Association in New York (NY). The event included a number of lectures and small group sessions that covered various topics related to Japanese culture. The event staff and investigators hand distributed the questionnaires without seeking the selfidentification status of the participants and collected the completed questionnaires during the event. In addition, questionnaires were mailed to those members who had not attended the Senior Week events with selfaddressed envelopes. Second, we distributed questionnaires to participants in a Japanese social meeting in West Virginia in October 2010. Third, we conducted a web-based survey with the same questionnaire contents

5

ACCEPTED MANUSCRIPT

among members of the Japanese American Citizens League via its mailing list and a banner on its official journal’s website in March 2011. In all cases, participants were asked to complete the questionnaires only once. The return of a completed questionnaire or the completion of the web-based survey was considered consent to

possible.

RI PT

participate in the study. Due to the nature of the sampling frames, formal response rate calculations were not

With respect to J/J, the data of previous Japanese nationwide survey studies were used for a historical comparison (3, 13). These studies were conducted by our team in 2004, and included 2,548 J/J participants

SC

who responded to a large number of questions including those in our study.

The ethical and scientific validity of this study was verified by the Institutional Review Board of the

M AN U

University of Vermont Committee on Human Subjects Research for the Medical Sciences.

Measurements

The authors conducted forward and back translations of the questionnaire that included the following

the questionnaire.

TE D

measurements. Both English and Japanese translated versions of each of the question items were integrated in

Items of the Good Death Inventory

EP

The Good Death Inventory (GDI) is a valid, reliable, and widely used questionnaire developed in Japan that contains a total of 54 items covering 18 domains that would contribute to a good death in Japan (2,

AC C

3). The GDI consists of 10 core domains that most Japanese consistently rated as important, and 8 optional domains that were not as consistently rated as important by Japanese individuals (3). Content/factor validity of the GDI was previously confirmed among Japanese general population and bereaved families of cancer patients (3). In order to reduce respondent burden, we chose one representative item from each of 18 domains that could best reflect cultural differences through discussions among the authors including the researcher who originally developed the GDI: ‘being free from physical distress’, ‘being able to stay at one’s favorite place’, ‘trusting physician’, ‘having some pleasure in daily life’, ‘not being a burden to family members’, ‘spending

6

ACCEPTED MANUSCRIPT

enough time with one’s family’, ‘being independent in daily activities’, ‘living in calm circumstances’, ‘not being treated as an object or a child’, ‘feeling that one’s life was completed’, ‘dying a natural death’, ‘seeing people whom one wants to see’, ‘maintaining one’s role in family or occupational circumstances’, ‘dying

RI PT

without awareness that one is dying’, ‘fighting against disease until one’s last moment’, ‘not exposing one’s physical and mental weakness to family’, ‘knowing what to expect about one’s condition in the future’, and ‘having faith’ (2, 3).

In addition, we asked three unique items of a good death which have traditionally been considered

SC

important to the Japanese culture (‘dying as one sleeps’, ‘having a sudden death’ (‘pokkuri’) and ‘leaving the decision to a medical expert’ (‘omakase’)) (3, 12).

M AN U

We provided respondents a scenario where they had incurable cancer, and asked them to rate the relative importance of each item in order to achieve a good death on a seven-point Likert scale (1: absolutely unnecessary, 2: unnecessary, 3: somewhat unnecessary, 4: unsure, 5: somewhat necessary, 6: necessary, and 7: absolutely necessary) (3).

TE D

Preferences of prognostic disclosure

Utilizing a valid questionnaire originally developed and used in a Japanese nationwide study (13), we provided two scenarios where respondents had an incurable cancer, and life expectancy of 6 or 1-2 months. In

EP

the respective scenarios, we asked respondents to state their preference about the way a discussion of prognosis should be initiated from the following: ‘not to discuss at all’, ‘physician to inform me only if I ask’, ‘physician

(13).

AC C

to check with me first whether I want to know’ or ‘physician to initiate a discussion and inform me in detail’

Participant Characteristics

We asked participant’s age, sex, length of stay in the US, presence of family members who died within 10 years as well as their terminal disease (cancer or non-cancer), and presence of chronic illnesses (3, 13). We also obtained variables of the Brief Acculturation Scale which is a valid and reliable tool to measure

7

ACCEPTED MANUSCRIPT

the level of acculturation among Japanese Americans: self-identification (“Japanese American” or “Japanese”), language preference (English or Japanese), extent of multiple generations born in the US (self, parents, and grandparents), and country in which the respondent was raised (US or Japan) (14). We divided the participants

RI PT

living in the US into "JA/A" and "J/A" based on their self-identification ("Japanese American" and "Japanese", respectively).

Statistical Analyses

SC

The primary endpoint was the proportion of respondents who answered “somewhat necessary”,

“necessary”, and “absolutely necessary” for each of the 18 good death items (3). Our secondary endpoints

M AN U

included the following; 1) the proportion of respondents who answered “somewhat necessary”, “necessary”, and “absolutely necessary” for the 3 culturally specific Japanese items for a good death on a seven-point Likert scale (3), and 2) the proportion of respondents who preferred each prognostic disclosure item (13). As our interest was in how JA/A and J/A may differ relative to the J/J group respondents, we implemented a multivariate logistic regression model for the dichotomous survey response using the three self-identified

TE D

group categories as a single item with the J/J group serving as the reference. From a clinical point-of-view, we assumed that a 20% group difference or more in the proportion of subjects who would rate “somewhat necessary”, “necessary” and “absolutely necessary” for each of 18 good

EP

death items would be of value. A simplified planning sample size calculation had been conducted assuming an equal number of survey respondents from the three groups (JA/A, J/A and J/J). The simplified scenario used a

AC C

(50%, 60% and 70%) response set for the groups to get the overall 20% difference for a given item for the primary endpoint. A chi-square test using a 5% significance level for three groups with these anticipated effects would have an 80% power to detect these differences if the sample sizes for each group were 116 each. Given the similarity of most secondary analysis effort to those for the primary endpoint, these calculations were considered adequate for these secondary endpoints. Analysis of baseline characteristics used contingency table methods for categorical items and the Kruskal-Wallis non-parametric rank test for ordinal items. Analysis of each of the primary measures and

8

ACCEPTED MANUSCRIPT

secondary measures were examined using 2x3 contingency tables and the chi-square homogeneity statistic. The four prognostic disclosure preference categories were examined using a 4x3 contingency table approach, where the four preferences would be crossed with the three survey subject groups. Adjustment for age and sex

RI PT

differences between the three groups were implemented using multivariate logistic regression models for binary outcome measures and multiple linear regression models for ordinal items with indicator variables to implement group membership.

Due to the multiple comparisons with a total of 23 outcome items and relatively large sample size

SC

differences among the three self-identified groups, Bonferroni adjusted p-value of less than 0.0021 was

M AN U

considered significant in this report (0.05/23).

RESULTS Participant characteristics

A total of 441 surveys were returned in the US and analyzed with the previous Japanese data of 2,548 participants. Baseline characteristics of the respondents were shown in Table 1. One hundred and forty and 301

TE D

of the respondents identified themselves as JA/A and J/A, respectively. JA/A were significantly more likely than J/A to prefer English, have multiple generations of families who were born in the US, and have been

EP

raised in the US. There were statistically significant differences in age and sex among the three groups.

Perceptions of a good death

AC C

Table 2 summarizes the proportion of respondents who answered “somewhat necessary”, “necessary” and “absolutely necessary” in order to achieve a good death if they had developed an incurable cancer. More than 80% of respondents in all three groups consistently considered the core items important without significant intergroup differences. The only exception was the item ‘spending enough time with one’s family’ which 84%, 74%, and 84% in JA/A, J/A, and J/J, respectively, considered important. Here the odds ratio (OR) for JA/A was 1.04 (95% confidence interval (CI), 0.64–1.68; adjusted p-value=0.8886), and the OR for J/A

9

ACCEPTED MANUSCRIPT

was 0.57 (95% CI, 0.43–0.76; adjusted p-value=0.0002) compared to J/J respondents. None of the 10 core items reached greater than 20% differences between any two groups. On the other hand, the significantly different group differences were seen for 5 of 8 optional items.

RI PT

Items that reached greater than 20% differences between any two groups included ‘fighting against disease until one’s last moment’ (49%, 52%, and 73% in JA/A, J/A, and J/J, respectively; OR for JA/A, 0.34 (95% CI, 0.24–0.49; adjusted p-value <0.0001), and OR for J/A, 0.39 (95% CI, 0.31–0.51; adjusted p-value <0.0001), respectively, as compared to J/J), ‘knowing what to expect about one’s condition in the future’ (83%, 80%, and

SC

58% in JA/A, J/A, and J/J, respectively; OR for JA/A, 3.51 (95% CI, 2.21–5.56; adjusted p-value <0.0001), and OR for J/A, 3.01 (95% CI, 2.21–4.09; adjusted p-value <0.0001), respectively, as compared to J/J), and

M AN U

‘having faith’ (64%, 43%, and 38% in JA/A, J/A, and J/J, respectively; OR for JA/A, 1.97 (95% CI, 0.99–3.94; adjusted p-value=0.0548), and OR for J/A, 0.66 (95% CI, 0.48–0.91; adjusted p-value=0.0127), respectively, as compared to J/J).

Likewise, the significantly different proportion of respondents rated all the three good death items specific to the Japanese culture important (Table 2). Items that reached greater than 20% differences between

TE D

any two groups included ‘dying as one sleeps’ (68%, 80%, and 92% in JA/A, J/A, and J/J, respectively; OR for JA/A, 0.15 (95% CI, 0.10–0.23; adjusted p-value <0.0001), and OR for J/A, 0.30 (95% CI, 0.21–0.42; adjusted p-value <0.0001), respectively, as compared to J/J), and ‘having a sudden death’ (‘pokkuri’) (49%, 68%, and

EP

79% in JA/A, J/A, and J/J, respectively; OR for JA/A, 0.23 (95% CI, 0.16–0.33; adjusted p-value <0.0001),

AC C

and OR for J/A, 0.52 (95% CI, 0.40–0.69; adjusted p-value <0.0001), respectively, as compared to J/J).

Preferences of prognostic disclosure Table 3 shows the participants’ preferences for prognostic disclosure. In a scenario where they had

developed an incurable cancer and would have a life expectancy of 6 months, JA/A respondents were the most likely to prefer that their physician initiate discussion and inform them in detail, followed by J/A and J/J (61%, 59% and 40%, respectively); on the other hand, JA/A were least likely to prefer not to discuss their prognosis at all, followed by J/A and J/J (0.8%, 2.5%, and 11%, respectively), or having their physician inform them

10

ACCEPTED MANUSCRIPT

only if they ask (9.6%, 12%, and 27%, respectively) (p <0.0001). Multiple linear regression models revealed that the age-gender adjusted regression coefficients for JA/A, J/A, and J/J were 0.59 (95% CI, 0.41–0.78; adjusted p-value <0.0001), 0.53 (95% CI, 0.40–0.65; adjusted p-value <0.0001), and 3.24 (95% CI, 3.08–3.41;

RI PT

adjusted p-value <0.0001), respectively. The results of the respondents’ preferences in a scenario where they would have a life expectancy of 1 – 2 months were essentially similar (Table 3).

DISCUSSIONS

SC

To the best of our knowledge, this is the first study to explore differences in perceptions of a good death among Japanese Americans with different levels of acculturation and Japanese living in Japan. The

M AN U

present study generated several major findings in a hypothetical scenario of an incurable cancer. Although we examined the sample of people with Japanese ancestry, our findings may serve as a model to better understand the potential impact of acculturation on perceptions of a good death among other minor populations. The first important finding is that values for almost all core items derived from the Japanese Good Death Inventory were preserved throughout the levels of acculturation. This is in line with the findings of

TE D

previous ethnographic and national survey studies showing that most American patients share with the Japanese similar values for a good death such as ‘being free from pain and shortness of breath’, ‘trusting one’s physician’, ‘not being a burden to family’, and ‘feeling that one’ life is complete’(1, 10). In fact, these core

EP

items are considered important across different cultures and countries, and may thus represent unanimously valuable factors (1-3, 15-19). Put together, clinicians may safely assume that patients with Japanese ancestry

AC C

value most core good death domains irrespective of the levels of acculturation, and help terminally-ill patients achieve a good death by ensuring that these domains are satisfied at their end of life. On the other hand, caution should be taken in addressing the importance of ‘spending enough time with one’s family’ in the J/A population. Although the majority of J/A considered this item important, the proportion who valued it was slightly lower than the other two groups. This finding is in part consistent with a previous study in which less acculturated Japanese Americans were less likely to prefer group surrogate decision making compared to more acculturated Japanese Americans and Japanese people living in Japan (11). These results may be explained by

11

ACCEPTED MANUSCRIPT

the uniqueness of less acculturated population as early immigrants, who are less likely to have many families around them (11), and whose weak relation to their family in Japan may have made them immigrate to the US. It may be important for clinicians to sensitively ask terminally-ill patients with Japanese ancestry about their

RI PT

expectation of their family. The second important finding is that perceived values for most optional good death items and all the culturally specific Japanese items significantly increased or decreased proportionally as people with Japanese ancestry acculturate to the US culture. In particular, the proportion of participants who valued 3 optional and 2

SC

culturally specific Japanese items differed by more than 20% among the groups, which had been set a priori as clinically important difference. As people with Japanese ancestry acculturate, their perceptions shift away from

M AN U

the Japanese traditional attitudes which value ‘fighting against disease until one's last moment’, ‘dying as one sleeps’, and ‘having a sudden death (“pokkuri”)’ (2, 3, 20, 21). On the other hand, they become more likely to consider ‘knowing what to expect about one's condition in the future’ and ‘having faith’ important for a good death. These tendencies reflect prevailing perceptions of a good death in the US (1, 10, 11). Previous national survey in the US indicated that only less than half of patients considered ‘using all available treatments no

TE D

matter what the chance of recovery’ important at the end of life, while most patients valued ‘feel prepared to die’, ‘knowing what to expect about one’s physical condition’, ‘being at peace with god’, ’praying’, and ’meeting with a clergy member’ (1). These findings suggest that the level of acculturation may have

EP

impact on patients’ perceived good death and their preferences about aggressive EOL care, explicit EOL discussions including prognostic disclosure, and religious and spiritual care. By assessing the level of

AC C

acculturation and their specific values, goals, and preferences about these culture-dependent topics, clinicians may better provide culturally-sensitive, individualized care to terminally-ill patients with Japanese ancestry. The third important finding is the clarification of the differences in preferences regarding detailed

process of initiating prognostic disclosure in case they had developed an incurable cancer. The more people with Japanese ancestry living in the US acculturated, the more they preferred their physicians taking active role in initiating prognostic disclosure. Very few people with higher acculturation preferred nondisclosure. This is consistent with previous studies showing that highly acculturated Japanese Americans preferred more explicit

12

ACCEPTED MANUSCRIPT

terminal disclosure than less acculturated Japanese Americans and Japanese living in Japan, and the predominant majority of patients in western countries including the US preferred full prognostic disclosure (11, 22-24). These observations suggest that clinicians should take an active role in assessing how much

RI PT

prognostic information patients with Japanese ancestry prefer, bearing in mind that the substantial minority may prefer prognostic disclosure only if patients ask.

This study has several limitations. First, the majority of the study sample was derived from limited geographic areas in the US. We attempted to minimize this bias by adding web-based survey, though the

SC

response rate was unclear. These convenient sampling methods with potentially limited representativeness of our participants and unknown response rate may hamper the generalizability of our findings. Second, as the

M AN U

data of J/J were obtained from the previous study conducted 6 years prior to our survey in the US, caution should be taken in interpreting the results from the direct comparisons. Third, perceptions and preferences of the general public may not reflect those of terminally-ill patients. Moreover, our findings in the Japanese American sample may not be applicable to the impact of acculturation of other minor population. Further studies are needed to determine the impact of acculturation on perceptions of a good death among terminally-

TE D

ill patients of differing ethnic self-identification groups. Fourth, although we adjusted the results for age and sex, other factors may have contributed to the outcomes. Future studies should examine the association between comprehensive background factors and perceptions of a good death.

EP

In conclusion, we showed that while most core items of a good death perceived by the Japanese were preserved throughout the levels of acculturation in Japanese Americans, the majority of optional items and

AC C

culturally specific Japanese items for a good death were similar to the perceptions of a good death held by the Americans as the Japanese Americans become acculturated. Understanding of the levels of acculturation of patients of ethnic minority may help clinicians provide culturally-sensitive, individualized EOL care to terminally-ill patients, especially when they discuss aggressive EOL care, prognostic information, imminent death, and religiosity and spirituality.

ACKNOWLEDGMENT

13

ACCEPTED MANUSCRIPT

Authors would like to express special appreciation to the following organizations, programs and colleagues: Mitsui-USA/Japanese Medical Society of America and the University of Vermont College of Medicine Hematology/Oncology Fellowship Research Grant for their generous financial support for this research; the

RI PT

Japanese American Association in New York, Japanese American Citizens League, Pacific Citizen (official newspaper of the Japanese American Citizens League), and Dr. Michiko Iwasaki, the Assistant Professor in the Department of Counseling, Rehabilitation Counseling, and Counseling Psychology at West Virginia

University, for their cooperation in patient accrual; and Mrs. Linda Norton and Mrs. Gail Berry at University

SC

of Vermont and Ms. Michiyo Noda and Miss Yuki Isokado at the Japanese American Association for their secretarial assistance. We also thank Dr. Tatsuya Morita for his valuable advice on the manuscript. Lastly, we

AC C

EP

TE D

M AN U

express our special appreciation to Dr. Steven Grunberg for his mentorship on this study.

14

ACCEPTED MANUSCRIPT

REFERENCES 1. Steinhauser KE, Christakis NA, Clipp EC, et al. Factors considered important at the end of life by patients, family, physicians, and other care providers. JAMA 2000;284:2476-82.

RI PT

2. Miyashita M, Morita T, Sato K, et al. Good death inventory: a measure for evaluating good death from the bereaved family member's perspective. J Pain Symptom Manage 2008;35:486-98.

3. Miyashita M, Sanjo M, Morita T, Hirai K, Uchitomi Y. Good death in cancer care: a nationwide quantitative study. Ann Oncol 2007;18:1090-7.

SC

4. Blackhall LJ, Murphy ST, Frank G, Michel V, Azen S. Ethnicity and attitudes toward patient autonomy. JAMA 1995;274:820-5.

M AN U

5. Carrillo JE, Green AR, Betancourt JR. Cross-cultural primary care: a patient-based approach. Ann Intern Med 1999;130:829-34.

6. Ersek M, Kagawa-Singer M, Barnes D, Blackhall L, Koenig BA. Multicultural considerations in the use of advance directives. Oncol Nurs Forum 1998;25:1683-90.

7. Cheung-Blunden VL, Juang LP. Expanding acculturation theory: Are acculturation models and the

Development 2008;32:21-33.

TE D

adaptiveness of acculturation strategies generalizable in a colonial context? International Journal of Behavioral

8. Bito S, Matsumura S, Singer MK, et al. Acculturation and end-of-life decision making: comparison of

EP

Japanese and Japanese-American focus groups. Bioethics 2007;21:251-62. 9. Bureau USC. 2015 American Community Survey 1-Year Estimates. 2015. Available from:

AC C

https://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=ACS_15_1YR_B02018&pr odType=table. Accessed February 8th, 2017. 10. Long SO. Cultural scripts for a good death in Japan and the United States: similarities and differences. Soc Sci Med 2004;58:913-28.

11. Matsumura S, Bito S, Liu H, et al. Acculturation of attitudes toward end-of-life care: a cross-cultural survey of Japanese Americans and Japanese. J Gen Intern Med 2002;17:531-9.

15

ACCEPTED MANUSCRIPT

12. Voltz R, Akabayashi A, Reese C, Ohi G, Sass HM. End-of-life decisions and advance directives in palliative care: a cross-cultural survey of patients and health-care professionals. J Pain Symptom Manage 1998;16:153-62.

RI PT

13. Sanjo M, Miyashita M, Morita T, et al. Preferences regarding end-of-life cancer care and associations with good-death concepts: a population-based survey in Japan. Ann Oncol 2007;18:1539-47.

14. Meredith LS, Wenger N, Liu H, Harada N, Kahn K. Development of a brief scale to measure acculturation among Japanese Americans. Journal of Community Psychology 2000;28:103-113.

SC

15. Steinhauser KE, Clipp EC, McNeilly M, et al. In search of a good death: observations of patients, families, and providers. Ann Intern Med 2000;132:825-32.

M AN U

16. Vig EK, Pearlman RA. Good and bad dying from the perspective of terminally ill men. Arch Intern Med 2004;164:977-81.

17. Murray SA, Grant E, Grant A, Kendall M. Dying from cancer in developed and developing countries: lessons from two qualitative interview studies of patients and their carers. BMJ 2003;326:368. 18. Grant E, Murray SA, Grant A, Brown J. A good death in rural Kenya? Listening to Meru patients and their

TE D

families talk about care needs at the end of life. J Palliat Care 2003;19:159-67. 19. Morita T, Oyama Y, Cheng SY, et al. Palliative Care Physicians' Attitudes Toward Patient Autonomy and a Good Death in East Asian Countries. J Pain Symptom Manage 2015;50:190-9 e1.

EP

20. Miyashita M, Kawakami S, Kato D, et al. The importance of good death components among cancer patients, the general population, oncologists, and oncology nurses in Japan: patients prefer "fighting against

AC C

cancer". Support Care Cancer 2015;23:103-10. 21. Ando M, Somchit S, Miyashita M, Jamjan L. The perception for Good Death of community dwelling Japanese and Thailand respondents. Asian/Pacific Island Nursing Journal 2016;1:91-96. 22. Johnson M, Tod AM, Brummell S, Collins K. Prognostic communication in cancer: A critical interpretive synthesis of the literature. Eur J Oncol Nurs 2015;19:554-67. 23. Hagerty RG, Butow PN, Ellis PA, et al. Cancer patient preferences for communication of prognosis in the metastatic setting. J Clin Oncol 2004;22:1721-30.

16

ACCEPTED MANUSCRIPT

24. Hagerty RG, Butow PN, Ellis PM, Dimitry S, Tattersall MH. Communicating prognosis in cancer care: a

AC C

EP

TE D

M AN U

SC

RI PT

systematic review of the literature. Ann Oncol 2005;16:1005-53.

17

ACCEPTED MANUSCRIPT

Table 1. Baseline characteristics JA/A (n=140) n 140

n

%

J/J (n=2548) n

301

%

<0.0001 a

2501

30

21%

32

11%

0

0%

40-49

11

7.9%

49

16%

613

25%

50-59

20

14%

49

16%

758

30%

60-69

23

16%

79

26%

710

28%

70-79

23

16%

78

26%

420

17%

80-

33

24%

14

4.7%

0

0%

M AN U

SC

<40

Sex

138

301

<0.0001 b

2512

Male

48

35%

66

22%

1186

47%

Female

90

65%

235

78%

1326

53%

Length of stay in the US

11-20 years 21-30 years

EP

31 years Language preference

AC C

English Japanese

Self: Born in the US?

<0.0001 a

299

0

0%

47

16%

NA

NA

3

2.2%

52

17%

NA

NA

14

10%

70

23%

NA

NA

120

88%

130

43%

NA

NA

TE D

0-10 years

137

133

<0.0001 b

295

91

68%

16

5.4%

NA

NA

42

32%

279

95%

NA

NA

139

<0.0001 b

301

Yes

93

67%

5

1.6%

NA

NA

No

46

33%

296

98%

NA

NA

Parents: Born in the US? None

138 73

<0.0001 a

299 53%

18

294

p-value

RI PT

Age

%

J/A (n=301)

98%

NA

NA

ACCEPTED MANUSCRIPT

One parent

22

16%

4

1.3%

NA

NA

Both parents

43

31%

1

0.3%

NA

NA

Grandparents: Born in the US?

74%

296

99%

One grandparent

8

5.8%

0

0%

NA

NA

Two grandparents

15

11%

1

0.3%

NA

NA

Three or more grandparents

13

9.4%

0

0%

NA

NA

140

299

NA

NA

RI PT

102

Place of being raised

<0.0001 b

29

21%

218

73%

NA

NA

Mostly in Japan

12

8.6%

65

22%

NA

NA

Mostly in the US

20

14%

6

2.0%

NA

NA

In the US only

76

54%

2

0.7%

NA

NA

Other

3

2.1%

8

2.7%

NA

NA

135

294

0.3185 b

2534

43

32%

84

29%

649

26%

Died of diseases other than cancer

51

38%

103

35%

934

37%

No bereavement experience

41

30%

107

36%

951

38%

No

140

299

<0.0001 b

2473

64

46%

76

25%

1023

41%

76

54%

223

75%

1450

59%

AC C

Yes

EP

Having chronic disease

TE D

Died of cancer

M AN U

In Japan only

Bereavement experience within 10 years

a

<0.0001 a

297

SC

None

138

: p-values are based on Kruskal-Wallis nonparametric rank tests.

b

: p-values are based on chi-square tests.

Abbreviations: JA/A, Japanese American; J/A, Japanese living in America; J/J, Japanese living in Japan; US, United States; NA, Not applicable

19

ACCEPTED MANUSCRIPT

Table 2. Perceptions of a good death among Japanese Americans (JA/A, n = 140), Japanese living in America (J/A, n = 301), and Japanese living in Japan (J/J, n = 2548).

n (%)

OR a

95% CI a

Adjusted

Core 10 domains Being free from pain and physical distress

RI PT

p-value a

123 (92%)

1.25

0.65, 2.43

0.5012

J/A

268 (93%)

1.28

0.79, 2.06

0.3193

J/J

2,236 (90%)

JA/A

1 (Ref)

M AN U

Being able to stay at one's favorite place

SC

JA/A

118 (89%)

0.54

0.30, 0.94

0.0302

275 (94%)

0.99

0.60, 1.65

0.9790

2,355 (93%)

1 (Ref)

127 (95%)

0.85

0.36, 1.99

0.7117

285 (98%)

1.47

0.67, 3.21

0.3394

2,409 (96%)

1 (Ref)

122 (92%)

0.95

0.50, 1.79

0.8719

277 (95%)

1.39

0.82, 2.36

0.2240

2,239 (91%)

1 (Ref)

JA/A

117 (89%)

1.06

0.58, 1.91

0.8588

J/A

267 (92%)

1.33

0.85, 2.10

0.2116

J/J

2,193 (89%)

1 (Ref)

J/A J/J

JA/A J/A J/J

TE D

Trusting physician

JA/A

J/J

AC C

J/A

EP

Having some pleasure in daily life

Not being a burden to family members

Spending enough time with one's family

20

ACCEPTED MANUSCRIPT

JA/A

108 (84%)

1.04

0.64, 1.68

0.8886

J/A

215 (74%)

0.57

0.43, 0.76

0.0002

J/J

2,098 (84%)

1 (Ref)

JA/A

119 (90%)

1.18

0.65, 2.13

0.5927

J/A

274 (94%)

1.78

1.10, 2.91

0.0199

J/J

2,169 (88%)

1 (Ref)

1.98

0.99, 3.94

0.5848

0.66

0.48, 0.91

0.0127

123 (93%)

J/A

241 (82%)

J/J

M AN U

JA/A

SC

Living in calm circumstances

RI PT

Being independent in daily activities

2,147 (87%)

1 (Ref)

126 (95%)

2.81

1.21, 6.49

0.0158

273 (94%)

1.87

1.13, 3.08

0.0149

2,218 (89%)

1 (Ref)

106 (83%)

1.08

0.67, 1.73

0.7534

260 (89%)

1.70

1.16, 2.48

0.0065

2,031 (81%)

1 (Ref)

JA/A

100 (76%)

0.33

0.22, 0.51

< 0.0001

J/A

256 (87%)

0.72

0.49, 1.04

0.0802

J/J

2,197 (90%)

1 (Ref)

114 (86%)

1.96

1.18, 3.28

0.0097

Not being treated as an object or a child JA/A J/A

TE D

J/J

Feeling that one's life was completed JA/A

J/J

AC C

Optional 8 domains

EP

J/A

Dying a natural death

Seeing people whom one wants to see JA/A

21

ACCEPTED MANUSCRIPT

J/A

243 (83%)

1.45

1.04, 2.00

0.0264

J/J

1,917 (76%)

1 (Ref)

JA/A

91 (70%)

0.74

0.50, 1.10

0.1358

J/A

232 (80%)

1.31

0.96, 1.79

0.0834

J/J

1,853 (75%)

1 (Ref)

0.46

0.31, 0.68

0.0001

0.91

0.70, 1.17

0.4518

Maintaining one's role in family or

50 (38%)

J/A

153 (52%)

J/J

M AN U

JA/A

SC

Dying without awareness that one is dying

RI PT

occupational circumstances

1,299 (53%)

1 (Ref)

65 (49%)

0.34

0.24, 0.49

< 0.0001 b

150 (52%)

0.39

0.31, 0.51

< 0.0001 b

1,799 (73%)

1 (Ref)

58 (44%)

0.54

0.37, 0.78

0.0011

129 (45%)

0.57

0.44, 0.74

< 0.0001

1,414 (58%)

1 (Ref)

JA/A

110 (83%)

3.51

2.21, 5.56

< 0.0001 b

J/A

230 (80%)

3.01

2.21, 4.09

< 0.0001 b

J/J

1,415 (58%)

1 (Ref)

Fighting against disease until one's last moment

J/A J/J

TE D

JA/A

Not exposing one's physical and mental

JA/A

J/J

AC C

J/A

EP

weakness to family

Knowing what to expect about one's condition in the future

22

ACCEPTED MANUSCRIPT

Having faith 86 (64%)

1.97

0.99, 3.94

0.0548 b

J/A

124 (43%)

0.66

0.48, 0.91

0.0127

J/J

917 (38%)

1 (Ref)

JA/A

90 (68%)

0.15

0.10, 0.23

< 0.0001 b

J/A

231 (80%)

0.30

0.21, 0.42

< 0.0001

J/J

2,278 (92%)

RI PT

JA/A

Japanese-unique items

JA/A J/A J/J

1 (Ref)

M AN U

Having a sudden death ('pokkuri')

SC

Dying as one sleeps

66 (49%)

0.23

0.16, 0.33

< 0.0001 b

193 (68%)

0.52

0.40, 0.69

< 0.0001

1,953 (79%)

1 (Ref)

('omakase') JA/A J/A

54 (41%)

0.37

0.24, 0.56

< 0.0001

119 (42%)

0.45

0.34, 0.59

< 0.0001

1,408 (57%)

1 (Ref)

EP

J/J

TE D

Leaving the decision to a medical expert

a

AC C

Percentage of ‘somewhat necessary’, ‘necessary’, and ‘absolutely necessary’. : Odds ratios, 95% confidence intervals, and p-values are based on multivariate logistic regression model,

adjusted for age and sex due to the baseline differences in these covariates’ distributions. Items with statistically significant intergroup differences are shown in bold (p<0.0021). b

: Items with clinically significant intergroup differences.

Abbreviations: OR, odds ratio; CI, odds ratio; JA/A, Japanese American; J/A, Japanese living in America; J/J, Japanese living in Japan; Ref, reference.

23

ACCEPTED MANUSCRIPT

Table 3. Preferences of prognostic disclosure among Japanese Americans (JA/A, n = 140), Japanese living in America (J/A, n = 301), and Japanese living in Japan (J/J, n = 2548). J/A (n=301)

J/J (n=2548)

n

n

n

%

125

278

1

0.8%

7

Physician to inform me only if I ask

12

9.6%

34

Physician to check with me first

36

29%

73

76

61%

inform me in detail Coefficient (95% CI) b

1 – 2 month life expectancy (p<0.0001) a Not to discuss at all

EP

Physician to inform me only if I ask Physician to check with me first

2.5%

270

11%

12%

681

27%

26%

540

22%

59%

1003

40%

0.59 (0.41, 0.78)

0.53 (0.40, 0.65)

3.24 (3.08, 3.41)

<0.0001

<0.0001

<0.0001

125

TE D

Adjusted p-value b

164

M AN U

whether I want to know

%

2494

SC

Not to discuss at all

Physician to initiate discussion and

%

RI PT

6 month life expectancy (p<0.0001) a

JA/A (n=140)

281

2489

2

1.6%

11

3.9%

415

17%

12

9.6%

30

11%

619

25%

33

26%

64

23%

418

17%

78

62%

176

63%

1037

42%

AC C

whether I want to know

Physician to initiate discussion and inform me in detail

Coefficient (95% CI) b Adjusted p-value b

0.68 (0.48, 0.88)

0.65 (0.51, 0.79)

3.25 (3.06,3.43)

<0.0001

<0.0001

<0.0001

a

: p-values are based on Kruskal-Wallis nonparametric rank tests.

24

ACCEPTED MANUSCRIPT

b

: Coefficients, 95% confidence intervals, and p-values are based on indicator variables in multiple linear

regression models, adjusted for age and sex due to the baseline differences in these covariates’ distributions. Pvalues and coefficients with statistically significant intergroup differences are shown in bold (p<0.0021).

RI PT

Abbreviations: JA/A, Japanese American living in America; J/A, Japanese living in America; J/J, Japanese

AC C

EP

TE D

M AN U

SC

living in Japan; CI, confidence interval.

25