Opinion survey of the Hong Kong general public regarding genomic science and technology and their ethical and social implications

Abstract

The study's objective is to survey the attitudes of Chinese people living in Hong Kong toward genomic science and technology (GST) and their ethical and social implications. Using a 24-item questionnaire, 877 Cantonese-speaking residents between age 18 and 64 with minimum high school education are interviewed by telephone. Multiple regression analysis identifies education level as the most important demographic variable. Overall, respondents have mild agreement with genetic determinism and the use of GST for disease prevention but not for non-therapeutic genetic enhancement and production of “genetically modified” crops or meat. Respondents strongly believe that GST tampers with nature and resources should be used to solve other healthcare problems first. Respondents also show little concern that personal genetic information may be abused by their employers or schools and have only a minimal willingness to share personal genetic information with their family members.

Keywords:

• biotechnology
• ethics
• Chinese culture

Introduction

Since the completion of the Human Genome Project and successful cloning of Dolly and other animals within the past decade, people have become more familiar with the rapid development of genomic science and technology (GST). At the same time, people have also become aware of the ethical, legal and social implications (ELSI) that GST has for society, including health and environmental threats of genetically modified (GM) crops, food animals and microbes, informed consent procedures in genetic testing, genetic discrimination, confidentiality of genetic information (GI), uses and abuses of GI, etc. This has led to many public surveys in Western countries to assess the general public's attitudes towards GST (Gaskell et al. 2003, Sturgis et al. 2004). Cross-cultural studies comparing different attitudes in different countries have also been conducted in New Zealand, Australia, Japan, India and Israel (Macer et al. 1997, 2000, Ikeda 2008), but studies about the Chinese people are rare. The few published studies in Mainland China, Taiwan and Singapore are focused on agricultural biotechnology (Subrahmanyan and Cheng 2000, Li et al. 2002, Chen and Li 2007), and in Hong Kong, only a telephone survey of public attitudes towards GM food and labeling is available (http://www.greenpeace.org.hk). Little is known about how Hong Kong people perceive the development of GST and its potential risks. The purpose of this study is to increase our understanding of the Hong Kong public's attitude towards GST and the ELSI. This is not only crucial for the development of GST in this community which has extensive international collaborative networks, but the study will also make a significant contribution to the GST literature regarding Chinese cultural factors that may facilitate or deter the development of GST in a Chinese community since over 95% of the 8 million people living in Hong Kong are ethnically Chinese.

Material and methods

(a) Telephone survey of the general public

The Research Centre of University of Hong Kong (HKU) was commissioned to conduct the telephone survey targeting Cantonese-speaking residents between age 18 and 64 with at least high school education as respondents (hereafter as “General Public” and “Phone Survey”), using a survey questionnaire containing 24 questions in addition to demographic information. Household telephone numbers were randomly sampled and only one qualified person from each residence was interviewed. A total of 1545 eligible households were contacted with 877 (56.8%) surveys completed. The maximum sampling error is±3.3%, at 95% level of confidence (α = 0.05). The phone interviews were conducted on the basis of anonymity and confidentiality and the study was approved by the local Institutional Review Board.

(b) Data management and statistical analysis

All responses are summarized by descriptive statistics. Factor analysis, derived by principal component factor analysis with varimax rotation, is performed on the 24 questions, supplemented by manual adjustment to improve the integrity of subject content of each theme, with identification of five independent themes from 17 questions: (A) Use of GST to prevent disease; (B) Human genetic enhancement; (C) Genetic modification of food and micro-organisms; (D) GST and negative reproductive implications; (E) Abusive use of genetic information. Internal reliability of themes is evaluated by Cronbach's alpha coefficient which ranges from 0.73 to 0.78 (Table 3). Each of the remaining seven ungrouped questions is each treated as an individual theme. Mean scores of each theme are given from 1 to 5 based on a five-point Likert scale with 1 = strongly disagree and 5 = strongly agree, and “neutral” set at 3. Demographic characteristics are given in numbers and percentages. Multiple regression modeling is used to determine statistically significant demographic variable(s) in association with responses. ANOVA is also used to evaluate differences within each demographic variable, followed by Scheffe's test or Fisher's LSD test to explore any pair-wise difference(s) between different “sub-groups” within the variable. Statistical significance is set at p < 0.05. All the statistical procedures are carried out with SPSS Version 15.0 for Windows.

Table 1. Grouping of different kinds of dialing responses in Phone Survey.

Dialing responses	n	%
Ineligibility status was confirmed	7708	46.4
Fail qualify	849	5.1
Language problem	134	0.8
Business	1033	6.2
Facsimile	779	4.7
Invalid phone number	4913	29.6
Eligibility status was undetermined	3954	23.8
Refusal before screening	21	0.1
No answer	3632	21.9
Busy	301	1.8
Interviews incomplete	668	4.0
Partially done cases	71	0.4
Refusal after screening	597	3.6
Not available	3405	20.5
Successful cases	877	5.3
Total	16612	100.0
Notes: Response rate = done cases/(done+partial+refusal after screening) = 56.8%.
The sampling error for the survey is ±3.3%. It is calculated by the formula: where α = 0.05; zα = 1.96; response rate (rr) = 0.0568; sample size (ss) = 877.

Table 2. Demographic characteristics of “General Public” (n = 870).

	n	%*
Gender
Male	428	49.2
Female	442	50.8
Age
18–24	260	29.9
25–44	444	51.1
45–64	165	19.0
Mean±S.D^§	33.9±11.74
Education
High school (F5 graduate) & matriculation	464	53.6
College & undergraduate	356	41.1
Postgraduate	46	5.3
Religion
No religion	623	71.8
Christian	204	23.5
Buddhism	36	4.1
Others	5	0.6
Occupation
Pre-university students	78	9.0
University students	87	10.1
Employees without university education	284	32.9
Employees with university education	287	33.3
Jobless/housewife/retired	126	14.6
Notes: *Excluded non-response cases. ^§Means and S.D.s of age groups are evaluated from age classes.

Table 3. Mean scores of themes and its related questions from “General Population”.

Theme	Content	Respondents of “General Public” from phone survey
		Mean	Overall mean	Cronbach's alpha
A. Use of genomic science and technologies (GST) to prevent disease	(3) GST will reduce the number of disabled children born	3.50	3.21	0.76
	(4) GST will be used to stop children from inheriting the usually fatal genetic disease, e.g. Thalassemia Anemia, major, even when the risks involved are not yet completely known	3.38
	(5) GST will be used to stop children from inheriting non-fatal birth defects, e.g. cleft lip, even when risks not yet completely known	3.12
	(7) GST will be used to stop children from inheriting mental retardation, even when risks completely known	3.04
	(9) GST will be used to test if a person is at increased risk of having diseases such as cancer, heart diseases and dementia, even when risks not completely known	3.01
B. Human genetic enhancement	(6) If GST can help to raise children to become successful people, people will use them to improve the next generation, even when risks are not completely known	1.92	2.05	0.78
	(8) GST will be used to improve the intelligence level of an unborn child, even when risks not completely known	2.18
C. Non-human genetic enhancement	(12) People will try to increase food production by genetically modified crops, even when risks not completely known	2.83	2.56	0.74
	(13) People will increase food production by genetically modified fish or animals, even when risks not completely known	2.34
	(14) Human DNA will be transferred to other organisms for medical purposes, e.g. transfer human genes to microbes to produce insulin, even when risks not completely known	2.52
D. GST and negative reproductive implications	(18) Genetic testing will encourage people to have nothing less than perfect babies and this will lead to excessive, unnecessary and unjustified abortions	2.94	2.91	0.73
	(23) Many pregnant women may consent to genetic testing without being fully informed of the nature and implications of the tests being done	2.90
	(24) Many pregnant women may consent to genetic testing and make a decision of abortion after the test without being fully informed of the nature and implications of the tests being done	2.87
E. Abusive uses of genetic information	(19) Insurance companies may refuse to provide insurance coverage on the basis of a person's genetic profile	3.26	2.68	0.74
	(20) Employers may discriminate against employees on the basis of their known genetic information	2.83
	(21) Children may be rejected from school on the basis of their known genetic information	2.47
	(22) Genetic information will be stored in computers (databanks) and it may be accessed by third parties, e.g. schools, police, health departments or other government departments etc without the owners' permission	2.16
F.  (1) Health and diseases are determined by genes		3.23	3.23	Nil
G. (2) GST reduces personal responsibility for health		2.87	2.87	Nil
H. (10) Society cannot guarantee GI used fairly		3.40	3.40	Nil
I. (11) Family members are obliged to share GI with each other		3.32	3.32	Nil
J. (15) Many other healthcare problems have priority over GST		3.63	3.63	Nil
K. (16) Genomic science ultimately tampers with nature		4.00	4.00	Nil
L. (17) GST should be prohibited before oversight mechanisms in place		3.30	3.30	Nil
Note: Themes F to L contain single question.

Results

Demographic characteristics of respondents

The “General Public” consists of 877 respondents in the Phone Survey. Male and female respondents are almost equal and 51% of respondents are between 25 and 44 years old. Respondents' mean age is about 34 years; 54% respondents have high school or matriculation and approximate 40% have university education; 66% are employed and about 70% have no religious beliefs (see Table 2).

Multiple regression modeling of themes

Multiple regression modeling with stepwise approach is applied to various themes to evaluate the significant demographic variables(s). Stepwise model-building techniques with a single dependent variable are described in numerous sources (see e.g. Darlington 1990, Morrison 1990, Hocking 1996). Respondents' education level is the demographic variable that is significantly associated with the largest number of themes. It affects positively the agreement scores of themes D, E and H (the regression coefficients (β) are given by 0.23, 0.18 and 0.23 respectively; see Table 4) and negatively to themes B, G, I and J (β = –0.16, –0.34, –0.14 and –0.14 respectively; see Table 4). This shows that respondents with higher education tend to have higher agreement scores in the first group of themes but lower agreement scores in the second group of themes. Age and religious affiliation also affect six themes each. Age has a positive regression coefficient in themes B and I (β = 0.10 and 0.21, respectively) but a negative coefficient in themes C, D, G and K (β = –0.11 for both themes C and D, –0.25 and –0.15 for themes G and K respectively). This suggests that older respondents tend to have a higher agreement in themes B and I but have a lower agreement in themes C, D, G and K. The different religions within the “religion” variable are listed in the following order: (1) “Non-religious”; (2) Christianity; (3) Buddhism. Respondents with no religious affiliation have higher agreement scores in themes A and B (β = –0.15 and –0.16 respectively) and lower agreement scores in themes H, J, K and L (β = 0.14, 0.16, 0.19 and 0.25 respectively). Gender is listed as (1) male; and (2) female, and it affects the responses of three themes. Male respondents have higher agreement scores than female respondents in theme C (β = –0.17) but have lower agreement score in themes J and K (β = 0.14 and 0.17 respectively). Occupation only affects theme G and “Pre-U students” are listed as first in the list (Table 2) and they have lower agreement score in theme G (β = 0.13).

Table 4. The results of multiple regression of various themes.

Independent variables	Regression coefficients (standard error) in themes
	A	B	C	D	E	G	H	I	J	K	L
Gender			−0.17 (0.062)**						0.14 (0.066)*	0.17 (0.071)*
Age		0.10 (0.046)*	−0.11 (0.045)*	−0.11 (0.047)*		−0.25 (0.068)***		0.21 (0.057)***		−0.15 (0.052)**
Education		−0.16 (0.053)**		0.23 (0.054)***	0.18 (0.053)**	−0.34 (0.070)***	0.23 (0.064)***	−0.14 (0.066)*	−0.14 (0.056)*
Religion	−0.15 (0.049)**	−0.16 (0.058)**					0.14 (0.070)*		0.16 (0.061)*	0.19 (0.065)**	0.25 (0.066)***
Occupation						0.13 (0.043)**
Notes: Stepwise method is adopted for the analysis of multiple regression of age, gender, education, religion and occupation as defined in Table 1. None of the variables are significant to the modeling of theme F.
*Demographic variable(s) significant to the regression model at 5% level.
**Demographic variable(s) significant to the regression model at 1% level.
***Demographic variable(s) significant to the regression model at 0.1% level.

Attitudes towards the use of genomic science and technologies

(A) Use of GST to prevent disease

Respondents have an overall mild agreement for this theme (mean: 3.21) (see Table 3), with about 40% having “agree” scores (combining “agree” and “strongly agree”), and approximately 35% having “neutral” scores (see Figure 1 and Table 5). Religion is the only demographic variable with a statistically significant association with response (p = 0.014) with non-religious respondents having the highest mean score (3.26), followed by Christians (3.09) and Buddhists (3.04) (see Table 6B); this is also confirmed by the regression model (see Table 4) that shows the endorsement to use GST to prevent disease is weakened by religious beliefs (p < 0.01). By ANOVA, there is no difference among “Non U-Employees”, “U-Employees” and “U-Students” (Table 6B), indicating that they have fairly similar responses, providing mild support for use of GST for disease prevention, particularly for child disability, fatal genetic diseases, birth defects and mental retardation; support is attenuated by religious beliefs and if GST is used to predict chronic diseases such as cancer, heart disease and dementia.

Figure 1. The distribution of responses (in %) of all questions in various themes.

Table 5. Different responses (in %) for various themes (A to L) by telephone respondents.

Responses	Percentage
	A	B	C	D	E	F	G	H	I	J	K	L
Strongly disagree	9.1	36.7	21.3	15.3	22.7	5.4	12.8	6.3	8.0	1.8	4.0	4.5
Disagree	15.0	32.5	27.7	20.6	23.8	11.1	27.3	13.7	13.7	8.1	3.9	16.6
Neutral	34.8	22.5	30.3	31.9	25.5	50.1	29.8	32.8	33.3	36.9	20.0	40.0
Agree	28.0	5.5	15.2	22.7	18.9	21.9	20.4	28.2	28.4	31.7	31.9	22.1
Strongly agree	13.1	2.7	5.6	9.6	9.1	11.5	9.6	18.9	16.7	21.4	40.2	16.9

(B) Human genetic enhancement

Respondents have strong disagreement with use of GST for human genetic enhancement such as creating more intelligent or successful children (mean: 2.05) (see Table 3), with approximately 70% disagreeing and  < 10% agreeing (see Figure 1 and Table 5). The regression model shows that age (p < 0.05), education (p < 0.01) and religious belief (p < 0.01) are significant variables for this theme (see Table 4). By ANOVA, respondents between age 25 and 44 have stronger disagreement (2.00) than age 45 to 64 (2.24) (p = 0.015; see Table 6A); university-educated respondents have stronger disagreement (1.93) than respondents with high school and matriculation education (2.16) and Christians (1.85) have stronger disagreement than non-religious respondents (2.12) (p = 0.001; see Table 6B). Significant difference is not seen between “U-Students” and other occupation groups, but is seen between “NonU-Employees” (2.20) and “U-Employees” (1.96) (p = 0.013; see Table 6B).

Table 6A. Comparison of mean score by gender, age and education of “General Public” in Phone Survey.

Item		Gender			Age					Education
		Mean		p-value of t-test	Mean			p-value of ANOVA	Pair-wise difference	Mean			p-value of ANOVA	Pair-wise difference
		Male	Female		18–24	25–44	45–64			High school & matriculation	College & undergraduate	Postgraduate
Theme	A	3.20	3.21	0.864	3.23	3.20	3.21	0.884		3.25	3.15	3.25	0.202
	B	2.11	1.99	0.044	2.02	2.00	2.24	0.015	(2,3)	2.16	1.93	1.92	0.001	(1,2)
	C	2.65	2.47	0.005	2.66	2.53	2.46	0.059		2.57	2.56	2.45	0.686
	D	2.97	2.85	0.063	2.97	2.95	2.73	0.021	(1,3),(2,3)	2.79	2.99	3.39	<0.001	(1,2),(2,3),(1,3)
	E	2.71	2.64	0.294	2.73	2.69	2.58	0.235		2.58	2.79	2.86	0.002	(1,2)
	F	3.21	3.25	0.560	3.17	3.23	3.36	0.129		3.22	3.21	3.50	0.151
	G	2.78	2.95	0.031	2.98	2.88	2.72	0.085		2.99	2.78	2.27	<0.001	(1,2),(2,3),(1,3)
	H	3.37	3.43	0.434	3.42	3.42	3.31	0.506		3.30	3.46	4.00	<0.001	(1,3),(2,3)
	I	3.26	3.38	0.104	3.22	3.25	3.68	<0.001	(1,3),(2,3)	3.41	3.23	3.20	0.052
	J	3.55	3.70	0.025	3.55	3.67	3.65	0.280		3.69	3.58	3.37	0.042	(1,3)*
	K	3.92	4.09	0.021	4.09	4.05	3.78	0.007	(1,3),(2,3)	4.01	4.05	3.73	0.163
	L	3.26	3.34	0.283	3.20	3.37	3.29	0.112		3.34	3.29	3.11	0.381

Table 6B. Comparison of mean score by occupation and religious belief of “General Public” in Phone Survey.

Item		Religious belief^‡					Occupation
		Mean			p-value of ANOVA	Pair-wise difference	Mean					p-value of ANOVA	Pair-wise difference
		No religion	Christian	Buddhism			Pre-U Students	U-Students	Non U-Employees	U-Employees	Unemployed
Theme	A	3.26	3.09	3.04	0.014	(1,2)	3.28	3.13	3.24	3.18	3.20	0.704
	B	2.12	1.85	2.13	0.001	(1,2)	2.04	1.91	2.20	1.96	2.05	0.013	(3,4)
	C	2.60	2.50	2.31	0.099		2.86	2.53	2.54	2.58	2.40	0.012	(1,5)
	D	2.89	2.94	2.96	0.744		3.05	2.93	2.78	3.07	2.71	0.000	(3,4),(4,5)
	E	2.65	2.75	2.75	0.331		2.71	2.84	2.54	2.76	2.67	0.023	(2,3),(3,4)
	F	3.25	3.19	3.17	0.641		3.29	3.13	3.18	3.27	3.28	0.578
	G	2.87	2.85	2.86	0.968		2.67	3.01	3.00	2.68	3.03	0.002	(3,4)
	H	3.34	3.49	3.69	0.070		3.56	3.37	3.27	3.57	3.32	0.011	(3,4)
	I	3.30	3.33	3.57	0.385		3.22	3.18	3.38	3.24	3.57	0.039	(1,5),(2,5),(4,5)*
	J	3.57	3.77	3.81	0.021	(1,2)	3.51	3.57	3.70	3.52	3.80	0.036	(1,5),(3,4),(4,5)*
	K	3.94	4.21	4.03	0.006	(1,2)	4.17	4.14	3.98	3.95	3.99	0.387
	L	3.24	3.47	3.61	0.005	(1,2)	3.10	3.16	3.36	3.29	3.41	0.169
Notes:
Age: (1) 18–24; (2) 25–44; (3) 45–64; Education: (1) High school & matriculation; (2) College & undergraduate; (3) Postgraduate; Occupation: (1) Pre-U students (F5 & matriculation students); (2) Non U-employees (employees without university education); (3) U-employees (employees with university education); (4) Unemployed (jobless/housewife/retired); Religion: (1) No religion; (2) Christian; (3) Buddhist.
‡“Others” have been omitted from the comparison due to insufficient sample size of the group.
(1,2): Significant pair-wise difference of group 1 and 2 by Scheffe's test.
*Fisher LSD test.

(C) Genetic modification (GM) of food and other organisms

Respondents have an overall mild disagreement with this theme (mean: 2.56) with about 50% disagreeing and 30% remaining “neutral”. There is stronger disagreement in respect of GM meat (2.34) than crops (2.83) (see Table 3). The regression model shows that gender (p < 0.01) and age (p < 0.05) are significant demographic variables (see Table 4), with female respondents (2.47) having stronger disagreement than male (2.65), and older respondents (age 45 to 64) having stronger disagreement than younger respondents (18 to 24). ANOVA shows a significant mean difference between different occupation groups (p = 0.012), with significant pair-wise difference seen between “Pre-U Students” and “Unemployed” respondents but not among “U-Students”, “Non U-Employees” and “U-Employees”. The responses of the latter three groups are found in a narrow range between 2.53 and 2.58, suggesting very similar “disagree” responses to GM food.

(D) GST and negative reproductive implications

Respondents mildly disagree (mean: 2.91) with the claim that GST has negative reproductive implications related to abortion (see Table 3). The number of respondents “agreeing” (32.3%) and “disagreeing” (35.9%) is roughly equal (Figure 1 and Table 5). The regression model shows that age (p < 0.05) and education (p < 0.001) are significant demographic variables, with older respondents tending to have stronger disagreement than younger respondents (β for age = –0.11; see Table 4), and less-educated respondents tending to have stronger disagreement (β for education = 0.23). ANOVA shows significant mean difference between different occupation groups (p < 0.001), with pair-wise differences between “Non U-Employees” (2.78) and “U-Employees” (3.07) and between “U-Employees” and “Unemployed” (2.71) (see Table 6B).

(E) Abuses of genetic information

Respondents have an overall mild disagreement (mean = 2.68) with potential abuse of genetic information by insurance companies, employers and schools (see Table 3), with about 28% “agreeing” and 47% “disagreeing” (Figure 1 and Table 5). The regression model shows that education (p < 0.01) is the only significant demographic variable. ANOVA confirms that the less-educated respondents would have stronger disagreement (2.58) than university-educated respondents (2.79) (p = 0.002; see Table 6A). Among different occupational groups, there are significant pair-wise differences between “U-Students” (2.84) and “Non U-Employees” (2.54), and between “Non U-Employees” and “U-Employees” (2.76) (p = 0.023; see Table 6B). This suggests that university education is associated with higher suspicion of GI abuses.

(F) Genetic determinism

Respondents have an overall mild agreement (mean: 3.23) that health and disease are determined by genes (see Table 3) with about 50% scoring “neutral”, 33% agreeing and 17% disagreeing (Figure 1 and Table 5). Responses are not affected by demographic variables (see Tables 4, 6A and 6B).

(G) GST reduces personal responsibility for health

Respondents have an overall weak disagreement (mean: 2.87) with this theme (see Table 3), with approximately 40% disagreeing, 30% agreeing and 30% neutral (Figure 1 and Table 5). The regression model shows that age (p < 0.001), education (p < 0.001) and occupation (p < 0.01) are significant demographic variables. Older respondents tend to have stronger disagreement than younger respondents (β for age = –0.25; see Table 4) and less-educated respondents tend to have higher agreement than more educated (β for education = –0.34; see Table 4). From ANOVA, the response is also affected by gender (p = 0.031), education (p < 0.001) and occupation (p = 0.002), with female (2.95) and less-educated respondents (2.99) having weaker disagreement than male (2.78) or university-educated respondents (2.78) (see Tables 6A and 6B). Comparisons among occupation groups are significant (p = 0.002) with “U-Employees” having stronger disagreement than “Non U-Employees”.

(H) Society cannot guarantee strictly fair use of genetic information

Respondents have an overall agreement (mean: 3.40) that it is impossible for society to guarantee the strictly fair use of genetic information (see Table 3) with approximately 47% agreeing and 20% disagreeing (Figure 1 and Table 5). The regression model shows that education (p < 0.001) and religion (p < 0.05) are significant demographic variables. ANOVA shows postgraduate respondents (4.00) having stronger agreement than university-educated (3.46) and high school-educated (3.30) respondents (β for education = 0.23; see Table 4), and “U-Employees” (3.57) having stronger agreement than “Non U-Employees” (3.27) (p = 0.011) (see Tables 6A and 6B).

(I) Sharing genetic information (GI) among family members

Respondents have an overall agreement (mean: 3.32) that people have responsibility and obligation to share GI with family members (see Table 3), with approximately 45% agreeing and 20% disagreeing (Figure 1 and Table 5). The regression model shows that age (p < 0.001) and education (p < 0.05) are significant variables; older respondents tend to have stronger agreement (β for age = 0.21; see Table 4) and less-educated respondents have higher agreement scores (β for education = –0.14, see Table 4). ANOVA shows differences among different occupation groups (p = 0.039) with significant pair-wise difference between “Pre-U Students” (3.22) and “Unemployed” (3.57), between “U-Students” (3.18) and “Unemployed”, and between “U-Employees (3.24) and “Unemployed” (Table 6B), indicating that “Unemployed” respondents (consisting of housewives and retirees) have the strongest agreement for this theme.

(J) Limited healthcare resources and GST

Respondents have an overall strong agreement (mean: 3.63) that limited healthcare resources should be used to solve other healthcare problems first rather than to develop GST (see Table 3), with approximately 53% agreeing and 10% disagreeing (Figure 1 and Table 5). The regression model shows that gender (p < 0.05), education (p < 0.05) and religion (p < 0.05) are significant variables. Therefore, less-educated respondents would have higher agreement score than other education groups (β for education = –0.14; see Table 4). T-test shows that female respondents have stronger agreement than male (3.70 and 3.55 respectively, p = 0.025; see Table 6A). Less-educated respondents have stronger agreement than postgraduate respondents (3.69 and 3.37 respectively, p = 0.042; see Table 6A). ANOVA shows differences in response due to gender, education, religion and occupation (p = 0.025, 0.042, 0.021 and 0.036 respectively; see Tables 6A and 6B). Female respondents have stronger agreement than male (3.70 and 3.55 respectively); significant pair-wise differences are seen between “high school” (3.69) and “postgraduate” respondents (3.37), and between “non-religious” (3.57) and “Christian” (3.77) respondents, suggesting that better educated and non-religious respondents have weaker agreement with this theme. Among different occupational groups, pair-wise differences are found between “Pre-U Students” (3.51) and “Unemployed” (3.80), between “Non U-Employees” (3.70) and “U-Employees” (3.52) and between “U-Employees” and “Unemployed”. This suggests that “Unemployed” respondents and “Non U-Employees” have higher agreement for this theme and “Pre-U Students” and “U-Employees” have lower agreement.

(K) GST and tampering with nature

Respondents have a strong overall agreement (mean: 4.00) that GST ultimately tampers with nature (Table 3), with approximately 72% agreeing and 8% disagreeing (Figure 1 and Table 5). The regression model shows that gender (p < 0.05), age (p < 0.01) and religion (p < 0.01) are significant variables. Younger respondents would have higher agreement scores (β for age =−0.15; see Table 4) than the other two age groups. ANOVA shows that female respondents have stronger agreement than male (4.09 and 3.92 respectively, p = 0.021); younger respondents from 18 to 24 (4.09) and 25 to 44 (4.05) have stronger agreement than older respondents from 45 to 64 (3.78, p = 0.007); and Christians have stronger agreement than non-religious respondents (4.21 and 3.94 respectively, p = 0.006) (Tables 6A and 6B). There is no significant difference among occupational groups.

(L) GST should be prohibited before regulatory mechanisms are in place

Respondents have a mild overall agreement (mean: 3.30) with this theme (see Table 3). There are approximately 39% agreeing, 21% disagreeing and 40% staying “neutral” (Figure 1 and Table 5). The regression model shows that religion (p < 0.001) is the only demographic variable that affects response. ANOVA shows that the difference is between Christians (3.47) and non-religious respondents (3.24) (Table 6B).

Discussion

Dissemination and abuses of genetic information

The proliferation and management of genetic information (GI) is one of the most important ramifications of GST, which has far-reaching ethical, legal and social implications. In this study, respondents agree that it is impossible for society to guarantee strict fairness when GI is used. On the other hand, excepting insurance companies, respondents disagree that employers or schools would abuse GI. Less-educated respondents have stronger disagreement than university-educated respondents and this is consistent with findings in the West that better-educated people are more suspicious of abuses of GI (Human Genetics Commission 2001) Furthermore, respondents show no concern that personal GI stored in databanks may be accessed by third parties without permission. In the West, people are much more suspicious of potential abuses of personal GI and this possibility has been used as an important argument to either ban employers, insurance companies and government agencies from accessing GI (Gaskell et al. 2003) or ban genetic testing altogether (Hietala et al. 1995). Further studies are needed to investigate why our respondents have so much confidence in schools, employers and databanks.

If management of GI in the public domain is a difficult problem, the dissemination of GI within a family is equally challenging. When a person has inherited disease-causing genes, it implies that the mutated genes can potentially be found in other family members or passed on to the next generation. In the present study, respondents have a mild agreement that people have responsibility and obligation to share GI with their family members, with older (between age 45 and 64) and less-educated respondents having the highest agreement and university students being the least enthusiastic. This is consistent with an earlier finding that >50% of Mainland Chinese would disclose GI to relatives at risk (Mao and Wertz 1997). In the West, findings are also quite inconsistent, with some findings affirming family members' right to know and patients' willingness to share GI (Julian-Reynier et al. 1996, Lehmann et al. 2000), while others showing a general reluctance to share GI except with one's spouse and children (Henneman et al. 2004). The right to know and the duty to inform increases if at-risk family members are part of a nuclear family or are threatened by a preventable disease (Pentz et al. 2005); willingness to share decreases if the disease is non-preventable and the at-risk person is a distant biologic relative (Peterson et al. 2003). Hence, sharing GI with family members in the West entails maintaining a delicate balance between several competing values: between a family member's “need to know” and the patient's right to privacy and confidentiality; between members of a nuclear family and extended family; and between “family-owned GI”, e.g. information of a gene mutation running in the family and “individual-owned GI”, e.g. an individual's own genetic test result (Pentz et al. 2005). If we consider the importance and dominance of the family in Chinese culture, these considerations seem to be less relevant. In a Chinese context, even members of the extended family would be entitled to own and therefore to know GI related to the family, as long as they share the genetic (blood) ties. From this perspective, the findings of the present study may not be representative of a general Chinese response because they reflect a relatively weak endorsement for sharing GI among family members. This is probably related to the relatively high education level of our respondents (60% high school graduates and 40% university graduates) and in Hong Kong, to be well educated implies strong exposure to Western influences that tend to “dilute” traditional Confucian sentiments. This is confirmed by our findings that university students in our sample have the weakest agreement to sharing GI with family members. More research is needed to explore the factors that may influence sharing GI with family members in Hong Kong since it is expected that more families will be confronted with genetic risk information in the future.

Genetic determinism, genetic testing and negative reproductive implications

Genetic testing presupposes a degree of genetic determinism and in our study, respondents have a mild agreement that diseases are genetically determined. Many lay people mistakenly believe that genes are rigidly deterministic, when in fact most genes are only mildly deterministic because of developmental and environmental factors that affect gene expressions, as proved by identical twin studies. The fact that 50% of respondents remain “neutral” in this question suggests considerable uncertainty and ambivalence, and this may also be the reason why their support for use of GST in disease prevention is very reserved. Support is highest for the purpose of preventing serious diseases and lowest for predicting risks of chronic diseases. This is in contrast to the widespread enthusiasm Westerners have for genetic testing of a wide spectrum of human diseases ranging from severe genetic diseases to psychiatric conditions and common disorders (Hietala et al. 1995, Jallinoja et al. 1998, Milner et al. 1998, Iredale and Longley 2000). A low view of genetics in disease causation and prevention held by the respondents probably explains why they also disagree that GST is related to people feeling less responsible for their own health. At the same time, respondents have an overall neutral score regarding certain negative reproductive implications of GST including an increase in unjustified abortions. This reflects respondents' general ambivalence towards the medical application of GST. Interpreters of public surveys tend to explain ambivalent attitudes as a reflection of the respondents' incomplete comprehension or inadequacies of the questionnaire. While this may be true in some cases, more commonly it is due to the very nature of the technology (GST in this case) being very complex and offering both advantages and disadvantages. Ambivalent responses then reflect the respondents' efforts to balance pros and cons of GST and account for the simultaneous coexistence of approval and support of certain aspects of GST, and doubt and disapproval of other aspects of GST (Furu et al. 1993, Jallinoja et al. 1998). The ambivalent attitudes of our respondents likely reflect the same predicament.

The ambivalence towards GST is even more manifested in respondents' disagreement with genetic modification (GM) of food. This finding is consistent with the findings of a government survey in 2000 that >50% of Hong Kong residents would not buy GM food (FEHD 2001), a 2004 Hong Kong Greenpeace survey that >75.0% of respondents rejected food containing GM ingredients (Greenpeace 2004) and findings from Mainland Chinese studies that GM foods are generally not accepted by Chinese consumers (Zhang 2002, Wang 2003). In this regard, Chinese consumers' attitudes are closer to European consumers' strongly negative attitudes than US consumers' more accepting attitudes (Gaskell et al. 1999). Our study also shows a trend that GM crops are slightly more acceptable than GM meats, probably because the former are perceived to be safer (Magnusson and Koivisto Hursti 2002). In contrast to GM food, respondents show a mild acceptance for the transfer of human DNA to microorganisms for medicinal production. This reflects the Chinese people's general pragmatic approach in decision-making since GM medicine is seen as more beneficial than GM food. A Taiwanese study also concludes that Chinese consumers are more positive about using GST to produce medicine than food (Chen and Li 2007). Chinese people are well known for their obsession with eating and food, and their concern with GM food probably reflects the cultural idiosyncrasies about food as well as a general reservation regarding GST.

Genetic enhancement and tampering with nature

Given the general reserved attitude towards GST, it is not surprising to find our respondents strongly rejecting the use of GST for human enhancement. GST can serve the purpose of therapy and enhancement: genetic therapy attempts to prevent or treat diseases by correcting the genes with harmful mutations and genetic enhancement attempts to alter the genetic material to improve non-pathological physical and behavioral characteristics. Despite the argument that it may be impossible to distinguish between the two (Miller 1994, Marteau et al. 1995), there is more support for genetic therapy than enhancement in the West based mainly on safety considerations (Frewer et al. 1997, Kerr et al. 1998, Milner et al. 1998, Calnan et al. 2005). But our study indicates that there may be another important underlying reason for Hong Kong people's conservatism regarding GST in general and rejection of genetic enhancement in particular. Our respondents hold the strong belief that GST ultimately tampers with nature. This corroborates the finding of another Mainland Chinese study that GM food threatens the natural order of things and is fundamentally unnatural (Lu 2006). This sentiment is not unique among Chinese since some consumers in the West, particularly Christians, also feel that GM foods are unnatural (Hill et al. 1998) and eating GM foods or taking gene tests gives rise to a sense of guilt and immorality for disturbing the order of nature (Hietala et al. 1995, Magnusson 2002). The reason why the concept of tampering with nature is shared by the culturally diverse Westerners and Chinese people is probably because they share a common worldview of a “natural order”: for Westerners the order of nature is ordained by the Christian God, and for the Chinese prescribed by Dao (path) – a concept shared by the two most influential philosophical/religious schools in China: Confucianism and Daoism.

The concern that GST tampers with nature may also affect public policy related to future development of GST. Respondents in this study agree that limited healthcare resources should be used to solve other healthcare problems in Hong Kong rather than developing GST and that GST should be prohibited before strict regulatory and oversight mechanisms are in place. Between the two stereotypes of either “genetic optimism” or “discourse of concern” towards a new technology such as GST (Durant et al. 1996, Conrad 2001), how people approach it is largely influenced by beliefs and attitudes (Catz et al. 2005). Until one's inherited beliefs are reconciled with the scientific merits of GST, people remain conservative or even skeptical about the innovation. For this reason, in order to improve people's attitude towards GST, education programs must be designed to tailor-fit the ethnic, cultural and linguistic contexts of the local community for them to be effective.

Conclusion

The present study of a sample of the general public of Hong Kong reveals a strong conservative attitude in this community about GST. Respondents have a limited agreement to the use of GST for prevention of serious inheritable diseases, but they disagree with most of the other applications including genetic testing for chronic diseases, production of GM foods and genetic enhancement of natural traits. They are also concerned about the negative implications of GST, particularly about consumers' increased desire for “designer babies” and choices for abortion. They believe that limited resources are better spent to provide other needed medical services than developing GST. It is possible that the overall negative response of our respondents may have been influenced by the fact that about half of our questions are cast in a “negative” tone; but more probably, this conservative attitude is related to the Chinese belief that GST basically tampers with nature. With regard to management of GI, our respondents show little concern about potential abuses by employers or schools, although they distrust insurance companies. Respondents are lukewarm in their willingness to share GI with other family members, and university students are even less willing to commit themselves; these responses contradict the strong Confucian familism that underpins the cultural ethos of the local community and requires further studies for their clarification.

Acknowledgement

The authors wish to thank Professor Chad Hansen of the Department of Philosophy, University of Hong Kong for his contribution in the conception phase of this project. This study was supported in part by a grant from the University Strategic Research Theme: Genomics, Proteomics, Bioinformatics & ELSI, The University of Hong Kong, grant No.10206152.11222.21700.302.01. The authors declare no conflict of interest of any kind.