The development of a Chinese version of the tobacco use subscale of the
behavioral risk factor surveillance system (BRFSS)
Yi-Hua Chen
a, Hung-Yi Chiou
a, Ping-Ling Chen
b,⁎
a
School of Public Health, College of Public Health, Taipei Medical University, No. 250, Wu-hsing Street, Taipei (110), Taiwan
b
School of Nursing, College of Nursing, Taipei Medical University, No. 250, Wu-hsing Street, Taipei (110), Taiwan Available online 14 February 2008
Abstract
Objectives. To provide a culturally and linguistically specific survey instrument for tobacco consumption in Taiwan, we evaluated the reliability
and validity of a Chinese translation of the 1993 US Tobacco Use Subscale of Behavioral Risk Factor Surveillance System (TU-BRFSS).
Methods. An integrative translation was followed by a pilot study of 100 randomly selected adults from throughout Taiwan. Telephone
interviews took place in July, 2004. Validity was assessed by Content Validity Index (CVI) computed on the basis of expert review and the
averaged scores of back-translation.
Results. Of 29 questions, 25 met the CVI criteria for retention in the instrument. In the back-translation assessment, 85% of the average scores
taken from the expert evaluations were above 4 (scale of 1
–5). Three of four percent agreements between the referent question and 4 other
questions were 100%.
Conclusions. The Chinese version of the TU-BRFSS, with appropriate content, semantics, and conceptual equivalence, appears valid and
reliable for future surveillance and research in Taiwan and other Chinese populations.
© 2008 Elsevier Inc. All rights reserved.
Keywords: BRFSS; Tobacco use subscale; Reliability; Validity; Taiwan; Smoking
Introduction
Tobacco use is a global epidemic and is considered to be the
single most preventable cause of premature morbidity and
mortality among men and women (
Peto and Lopez, 2001
). In
Taiwan, tobacco control is considered one of the most important
health policies by the administration.
Based upon U.S. tobacco control from 1965 to 2001, the
percentage of adult smokers aged 18+ in the general population
declined from 42.4% to 22.8%, both for males (52.0% to
25.5%) and for females (34.0% to 21.5%;
Centers for Disease
Control and Prevention, 2003
). This indicates that it takes a
sustained effort and a long period of time to reduce smoking
rates. According to surveys conducted by the Taiwan Wine and
Tobacco Monopoly Bureau between 1973 and 1996, the overall
adult smoking rate was 55%–63% for males and 2.3%–4.6%
for females (
Department of Health, 2006
). In a 2002 national
survey, approximately 50.0% of men and 5.8% of women
reported that they smoke every day, which is higher than what
has been observed in other countries (
Department of Health,
2002
).
World Health Organization (WHO) urged member states to
immediately take action on nationwide strategies of tobacco
control to prevent further morbidity and mortality caused by
tobacco use. With the establishment of national tobacco
sur-veillance systems toward tobacco use prevention and control,
a standardized and reliable structure and capacity could be
applied to assess and monitor tobacco profiles within a country.
To promote tobacco control actions globally, it is essential for
countries to be aware of similarities and differences among
groups. Key to the assessment of international comparisons and
the evaluation of a country's tobacco control efforts in light of
the experiences of other countries is a set of standardized
instruments and administration for surveillance data collection.
Preventive Medicine 46 (2008) 591–595
www.elsevier.com/locate/ypmed
⁎ Corresponding author. Fax: +886 2 2738 4831.
E-mail addresses:yichen@tmu.edu.tw(Y.-H. Chen),hychiou@tmu.edu.tw
(H.-Y. Chiou),plchen@tmu.edu.tw(P.-L. Chen).
0091-7435/$ - see front matter © 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.ypmed.2008.02.004
As a member in the global village, previous surveys in Taiwan
on smoking prevalence and behaviors might not be sufficient
for effective international comparison because of differences in
the instruments and methods utilized in Taiwan and in other
countries. Thus, a more appropriate assessment instrument
should be applied in Taiwan to facilitate comparison with the
global population.
The Behavioral Risk Factor Surveillance System (BRFSS,
available for download at
www.cdc.gov/brfss
), which is the
most widely used random digit dial telephone survey, was
designed to measure prevalence and time trends for
health-related behaviors in the U.S. population in 1993 (
Centers for
Disease Control and Prevention, 2006
). Telephone surveys
are an appropriate method for investigating the prevalence of
health risk behaviors among populations. Thus, given the
cost advantage and feasibility for long-term administration,
the Tobacco Use Subscale of BRFSS (TU-BRFSS), a widely
adopted instrument in surveillance system in literature, was
selected for translation and use for tobacco surveillance and
research in Taiwan.
The TU-BRFSS was designed to evaluate adult smoking
prevalence, age of smoking initiation, behaviors for smoking
cessation, exposure to and policy for environmental tobacco
smoke. Reliability testing has been reported for the core BRFSS
questions (
Shea et al., 1991; Stein et al., 1993, 1995; Brownson
et al., 1994
). Although an original instrument might have been
established with sound psychometric properties, it is critical to
evaluate the reliability and validity of translated versions (
Polit
and Hungler, 1999
).
Thus, the aims of this study were to translate and evaluate
the Chinese version of the TU-BRFSS, to conduct a
com-prehensive pilot study, and to assess the validity and reliability
of using this instrument to assess smoking behaviors of adults
aged 18+ years in Taiwan.
Methods
In order to evaluate the content, semantics, and conceptual equivalence of the instruments in both the source and target languages,Guillemin et al. (1993)
recommended the adoption of translation, back-translation, expert review and a pilot study as guidelines for cross-cultural adaptation of health-related measures. This study was administered based upon these guidelines.
Translation, back-translation, and expert review
Back-translation, the most commonly recommended and adopted manner of instrument translation (Brislin et al., 1973; Flaherty et al., 1988; Jones and Kay, 1992), was used to assess the semantics and content equivalence of the target language version. While the semantic equivalence assesses whether the meaning of each question remains the same compared with the source language version, the content equivalence ensures that the content of each item maintains consistent cultural relevance between the two versions. The validity and conceptual equivalence of the target language was assessed through expert review.
Thus, an integrative translation method was developed based upon the methods of translation, back-translation, and expert review as follows.
Step I: Two bilingual experts translated the TU-BRFSS instrument from English into Chinese, with Chinese Version I generated from the combined agreement on translation.
Step II: Five experts in tobacco-related fields assessed the validity, suitability, cultural equivalence, conceptual equivalence, and applicability of the
instrument for international comparisons of Chinese Version I. They were asked to rate each question on a four-point scale based on relevance, clarity, and importance.
Step III: Suggestions for questions and wording modifications that were collected from expert review were then taken into consideration while editing the BRFSS into Chinese Version II.
Step IV: Two bilingual English instructors of university-level English classes translated Chinese Version II back into English to ensure that the Chinese Version II retained the same meaning as the English version. Step V: Two additional bilingual English instructors independently compared the original English instrument and the version translated back from Chinese Version II to ensure the equivalence and cultural relevance. Each question was thus scored from“1” meaning “very inappropri-ate” to “5” meaning “very appropriate.”
Step VI: The results from Step V were edited and further modified based upon expert evaluation. This Chinese version III of the TU-BRFSS was used in a pilot study to assess reliability.
Pilot study
Sample/location
A pilot study was conducted in July, 2004. The sample was selected from adults in Taiwan with at least one residential telephone in the household, as the household telephone coverage rate in Taiwan in 1999 was up to 99% (Ministry of the Interior, 1999). A sample size of at least 25 (preferably 50) is required if the purpose of a pilot study is to examine whether measurements are reliable and valid (Lancaster et al., 2004). Because our study was a nationwide investigation, 100 adults were predetermined for investigation. Phone numbers were propor-tionately selected and called by random digit sampling. While about 39% of calls were not answered, 33% were not eligible (e.g., non-residential household), and 10% refused to participate, 100 adults were questioned, with an overall average completion rate of 18% (17.2%, 18.7%, 20.8% and 15.8% in the Northern, Central, Southern, and Eastern areas of Taiwan, respectively). Procedure
For each of the Northern, Central, Southern, and Eastern areas of Taiwan, 25 area codes were randomly selected. Then, the last four digits of the telephone number were randomly selected to ensure that all residents of Taiwan with a home phone had a chance of being selected, regardless of their number's presence in a phone book. Up to four calls during three different calling periods were made to contact a selected household before a replacement number was
Table 1
The content validity index for the Chinese Version I of the BRFSS, the Tobacco Use Subscale Instrument (TU-BRFSS)
Traits No. of questions Percentage (%) CVI score for keeping
1 22 75.9
0.8 3 10.3
0.6 4 13.8
CVI score for modification
1 4 13.8 0.8 6 20.7 0.6 9 31.0 0.4 2 6.9 0.2 7 24.1 0 1 3.4
Keep the question?
No 4 13.8
Yes 25 86.2
Unmodifieda (10) (40) Modifieda (15) (60) Data from nationwide Taiwanese respondents (2004).
aThe numbers in parentheses represent number and percentages of retained
generated and dialed. Each interview was administered by trained Public Health graduate students and lasted approximately 5–10 min. All interviewees who completed the phone call provided full responses.
Statistical analyses
Data from the integrative translation method
The validity of the instrument was assessed by both the Content Validity Index (CVI) calculated from expert review and the averaged scores estimated from back-translation. The CVI is derived from a four-point ordinal rating scale with“1” meaning a very inappropriate element and “4” a very appropriate item (Lynn, 1986). The CVI score for retaining a question was defined as the proportion of questions that received a rating of 3 or 4 by the experts, while the CVI for modification was calculated as the proportion that received a rating of 4. Generally, questions that had a CVI over 0.8 remained, while those with CVI scores for modification lower than 0.6 were further edited or deleted based upon the experts' opinions.
Data from the pilot study
Because the instrument was not designed with a set of questions with consistent ordinal or continuous coding, Cronbach's alpha was not applicable. Alternative methods, such as percent agreement, were adopted for reliability examination in the pilot study. SAS 8 (SAS Institute, Cary, NC) statistical software was used for data analysis.
Results
The CVI assessment of the Chinese TU-BRFSS version I is
presented in
Table 1
. Of 29 questions, those with CVI for
keeping over 0.8 were retained, and the rest were discarded,
resulting in a 25-question scale. By modifying 15 questions
using experts' opinions, the edited Version II instrument was
then used for back-translation examination.
In
Table 2
, 85% of average scores taken from the expert
evaluations comparing the two versions of the instrument were
above 4 (scale of 1
–5), indicating acceptable content and
cultu-ral equivalence between the original and translated versions.
Questions, especially those with average scores below 4, were
further modified using experts' opinions. The revised Chinese
version III was used in the pilot study that examined reliability.
In the pilot study, the gender ratio within the community
sample was approximately equal. Over 60% of respondents
were married, and over 75% had at least a high school
diplo-ma. Sixty-seven percent currently held a job, and more than half
earned a family income of 1 million NT (i.e., equivalent to
$30,000 US) or less per year (not shown in table).
Table 3
compares the sociodemographic distribution of national data,
drawn from governmental statistics reported by the Department
of Statistics, Ministry of the Interior, the Executive Yuan, and the
data from our study. No statistically significant difference was
found regarding the distribution of gender, age, employment
status, and marital status (all p-values
N0.05) between groups.
Further data analysis assessed the logical consistency of an
answer to one question with the response to a comparable
question. If interviewees' answers were reliable, their responses
to the referent question and the other four questions in
Table 4
should have been highly consistent. Results indicated that the
percent agreement between the referent question and 3 out of
the 4 questions was 100%. This high consistency demonstrated
good reliability.
In summary, based upon the reliability and validity
assessment, the TU-BRFSS instrument was further edited and
modified as follows. Questions of
“last smoked regularly” and
“has sought medical care” were deemed unsuitable and were
removed. For cultural applicability, four more questions were
added; these include questions on
“how often have you quit
smoking?” “in the past 7 days, when you were at home, how
many days has someone else smoked in front of you?” “in the
Table 2
The percentage distribution of the averaged scores reported by two experts who compared the two versions of the instrument (i.e., the source version and English version translated back from the Chinese version II) in the back-translation Score average No. of questionsa Percentage (%)
5 6 22.2 4.5 9 33.3 4 8 29.6 3.5 1 3.7 3 2 7.4 2.5 1 3.7 Total 27 100
Data from nationwide Taiwanese respondents (2004).
aThis indicates“number of questions receiving the given score average.”
Table 3
The sociodemographic distribution of national data and our pilot study data for the BRFSS instrument, the Tobacco Use Subscale (TU-BRFSS)
National data Sample Chi-squarea No. (%) Gender Male 11,541,585 (50.9%) 51 (51%) 0.97 female 11,147,537 (49.1%) 49 (49%) Age 18–29 3,886,690 (23.3%) 27 (27.8%) 0.52 30–64 10,672,210 (63.9%) 57 (58.8%) 65+ 2,150,475 (12.9%) 13 (13.4%)
Employment statusb Employed 9,786,000 (95.6%) 67 (95.7%) 0.95 Unemployed 454,000 (4.4%) 3 (4.3%)
Marital status Single 4,429,153 (39.3%) 28 (28%) 0.08 Married 6,022,846 (53.4%) 63 (63%)
Divorced/widowed 819,500 (7.3%) 9 (9%) Data are from nationwide Taiwanese respondents (2004).
a Chi-square tests were carried out for the comparison of distributions between national data and sample data. b
past 7 days, when you were working at your job, how many
days has someone else smoked in front of you?
” and “are you a
native resident?
” Question order and wording were modified to
better fit the logical thought flow of the interview. The Chinese
version was thus developed to assess adult smoking behaviors
in Taiwan.
Discussion
The integrative translation method in this study presented
a systematic and valid approach to translate an instrument.
Empirical evidence from our study suggests that the Chinese
version of the TU-BRFSS is valid and reliable for international
comparison and native investigation. It possessed the
appro-priate semantics, content (assessed by back-translation), and
conceptual (assessed by expert review) equivalence of the
original language.
Cigarette smoking is a major health risk behavior worldwide.
In an analysis of national vital statistics in developed countries
from 1950 to 2000,
Peto et al. (1994)
estimated that more than
10 million people will die from smoking by 2025. In Taiwan,
13.9% of male mortality and 3.3% of female mortality could
be attributed to cigarette smoking (
Liaw and Chen, 1998
).
Reducing tobacco-related risk behaviors is a priority among the
national and international health objectives.
Reliability and validity are essential characteristics to
en-sure that instruments are valid for smoking surveillance and
monitoring. Previous studies have demonstrated that the
BRFSS questionnaire is a valid tool to survey and conduct
research in the US (
Shea et al., 1991; Stein et al., 1993, 1995;
Brownson et al., 1994
). For example,
Stein et al. (1993)
demon-strated that reliability coefficients were over 0.7 for behavioral
risk factors including smoking. In another study by
Shea et al.
(1991)
, Pearson or kappa correlations for questions concerning
demographics and behavioral risk factors were more than 0.6
(p
b0.001), except for questions about diet. Test-retest reliability
demonstrated acceptable to high question reliability on an
indi-vidual level. In our study, the logical consistency of an answer
to one question with a response to another comparable item was
assessed. There was 100% agreement between smoking status
and smoking habits, smoking quantity per day, and smoking
rules at home. This high consistency demonstrated that the
Chinese version of the TU-BRFSS instrument is appropriate and
reliable.
Content validity measures the comprehensiveness and
re-presentativeness of the content of a scale and could contribute to
support the construct validity. The CVI is the most widely
adopted approach to quantify content validity. Based upon
recommendations by
Lynn (1986)
, items in this study that did
not achieve the required minimum agreement of the experts
were eliminated or further edited. The final Chinese version
of the TU-BRFSS thus demonstrated good content validity.
Further, when comparing the two versions of the instrument (i.e.,
the source version and English version translated back from
the Chinese version II), 85.1% of the average scores taken
from the expert evaluations were above 4 and items were
modified based upon experts' review. Thus, the Chinese version
of the TU-BRFSS displayed good content and cultural
equi-valence with the source and target versions.
Study limitations and strengths
Results of the study directed the need for a culturally and
linguistically specific survey instrument to address tobacco
control issues in Taiwan. A procedure recommended by
Guillemin et al. (1993)
was adopted for an appropriate reliability
and validity assessment of TU-BRFSS. A nationwide sample of
100 community adults was drawn for the administration of a
pilot study. However, there were some limitations to our study.
First, internal consistency reliability as assessed by Cronbach's
alpha was inapplicable because the instrument was not designed
with a series of questions that had a consistent ordinal or
continuous scale of response choices. In this study, logical
consistency was measured instead by percent agreement. In
addition, test-retest reliability was not possible because of the
anonymous nature of the telephone interview in the pilot study.
Second, the CVI procedure to assess content validity might have
been limited by the possibility of chance inflation (agreement)
(
Waltz and Bausell, 1981
). However, five experts in our study
were capable of providing a sufficient level of control for chance
agreement (
Lynn, 1986
). Finally, a relatively small sample size
and rather low response rate might restrict inferences from
broader generalization. However, our pilot study recruited an
appropriate sample size as a pilot and was representative of the
national data reported by government statistics regarding the
distributions of gender, age, employment, and marital status.
Future studies are needed to evaluate the Chinese version of
the TU-BRFSS in a more demographically and geographically
diverse population. With the recruitment of more participants,
broader generalizations may be possible.
Conclusions
In summary, these preliminary findings support applying the
Chinese version of the TU-BRFSS for surveillance and research
to measure and monitor the prevalence of smoking-related
issues in Taiwan and other Chinese populations as well. Based
Table 4
Logical analysis of questions on current smoking status and other smoking-related items in the Chinese version III instrument
Items Percent
agreement (%) Reference question: Do you now smoke cigarettes every day,
some days, or not at all? To be compared with the following four questions:
–
Question 1: About how long has it been since you last smoked cigarettes regularly?
100 Question 2: During the past 30 days (1 month), on how many
days did you smoke cigarettes?
90 Question 3: During the past 30 days (1 month), how many
cigarettes did you usually smoke on the days you smoked? 100 Question 4: Which statement best describes the rules about
smoking inside your home?
100
upon the reliability and validity of the instrument, future work
should be performed to provide the nation with a more accurate
estimate of smoking prevalence and evaluation of effects on
tobacco control intervention.
Acknowledgments
This research was supported by a grant from the Bureau of
Health Promotion, Department of Health, Executive Yuan,
Taiwan.
References
Brislin, R.W., Lonner, W.J., Thorndike, R.M., 1973. Cross-cultural Research Methods. John Wiley & Sons, New York.
Brownson, R.C., Jackson-Thompson, J., Wilkerson, J.C., Kiani, F., 1994. Reliability of information on chronic disease risk factors collected in the Missouri Behavioral Risk Factor Surveillance System. Epidemiology 5, 545–549.
Centers for Disease Control and Prevention, 2003. Cigarette smoking among adults— United States, 2001 [erratum appears in MMWR Morb Mortal Wkly Rep. 2003 Oct 24;52(42):1025]. MMWR Morb. Mort. Wkly. Rep. 52, 953–956.
Centers for Disease Control and Prevention, 2006. Behavioral Risk Factor Surveillance Factor Operational and User's Guide. Version 3.0. In: National Center for Chronic Disease Prevention and Health Promotion, (Ed.). Division of Adult and Community Health, Behavioral Surveillance Branch, Centers for Disease Control and Prevention, USA.
Department of Health, 2002. Reports from knowledge, attitude, and behavior about health promotion in Taiwan. Bureau of Health Promotion, Department of Health, Executive Yuan, Taiwan.
Department of Health, 2006. Taiwan Tobacco Control Annual Report, 2006. Bureau of Health Promotion, Department of Health, Executive Yuan, Taiwan.
Flaherty, J.A., Gaviria, F.M., Pathak, D., et al., 1988. Developing instruments for cross-cultural psychiatric research. J. Nerv. Ment. Dis. 176, 257–263. Guillemin, F., Bombardier, C., Beaton, D., 1993. Cross-cultural adaptation of
health-related quality of life measures: literature review and proposed guidelines.[see comment]. J. Clin. Epidemiol. 46, 1417–1432.
Jones, E.G., Kay, M., 1992. Instrumentation in cross-cultural research. Nurs. Res. 41, 186–188.
Lancaster, G.A., Dodd, S., Williamson, P.R., 2004. Design and analysis of pilot studies: recommendations for good practice. J. Eval. Clin. Pract. 10 (2), 307–312.
Liaw, K.M., Chen, C.J., 1998. Mortality attributable to cigarette smoking in Taiwan: a 12-year follow-up study. Tob. Control 7, 141–148.
Lynn, M.R., 1986. Determination and quantification of content validity. Nurs. Res. 35, 382–385.
Ministry of the Interior, 1999. Investigation on the Telecommunication Industry. The Ministry of the Interior, the Executive Yuan, Taiwan.
Peto, R., Lopez, A.D., 2001. Future worldwide health effects of current smoking patterns. In: Koop, C.D., Pearson, C., Schwarz, M.R. (Eds.), Critical Issues In Global Health. Jossey-Bass, New York, NY.
Peto, R., Lopez, A.D., Boreham, J., Thun, M., Health, C., 1994. Mortality from smoking in developed countries, 1950–2000: indirect estimates from national statistics. Oxford University Press, Oxford.
Polit, D.E., Hungler, B.P., 1999. Nursing research: Principles and methods, 6th ed. Lippincott, Philadelphia.
Shea, S., Stein, A.D., Lantigua, R., Basch, C.E., 1991. Reliability of the behavioral risk factor survey in a triethnic population. Am. J. Epidemiol. 133, 489–500.
Stein, A.D., Courval, J.M., Lederman, R.I., Shea, S., 1995. Reproducibility of responses to telephone interviews: demographic predictors of discordance in risk factor status. Am. J. Epidemiol. 141, 1097–1105.
Stein, A.D., Lederman, R.I., Shea, S., 1993. The Behavioral Risk Factor Surveillance System questionnaire: its reliability in a statewide sample [see comment] Am. J. Public Health 83, 1768–1772.
Waltz, C.W., Bausell, R.B., 1981. Nursing research: Design, statistics and computer analysis. F. A. Davis, Philadelphia.