Volume 2: No.
4, October 2005
Twelve Essentials of Science-based Policy
Bernard C.K. Choi, PhD, MSc
Suggested citation for this article: Choi BCK. Twelve essentials of science-based
policy. Prev Chronic Dis [serial online] 2005 Oct [date cited].
Available from: URL: http://www.cdc.gov/pcd/issues/2005/
This article presents a systematic framework of 12 essentials, or basic
elements, of science-based policy. The 12 essentials are grouped into three
categories, or areas, as follows: 1) knowledge generation, which
includes credible design, accurate data, sound analysis, and comprehensive
synthesis; 2) knowledge exchange, which includes relevant content,
appropriate translation, timely dissemination, and modulated release; and 3)
knowledge uptake, which includes accessible information, readable
message, motivated user, and rewarding outcome.
Back to top
The relationship between science and policy is an important topic in
evidence-based public health policy and practice (1). It seems logical to
assume that as scientific research generates more quality findings,
policymakers will make better decisions. However, numerous underlying obstacles
A systematic framework can be used to describe the key components that link
science to policy. The framework, which consists of three areas that are
subdivided into 12 essentials (basic elements), reveals issues and
solutions related to science-based decision making. In this article, policy
is defined broadly to include not only legislation but also “prudence or
wisdom in the management of affairs” and “a definite course or method of
action selected from among alternatives in light of given conditions to guide
and determine present and future decisions” (3). Therefore, the term policymakers
may encompass public health practitioners, public health researchers, and even
the general public, because members of the general public make health
decisions for themselves and their families.
Science-based policy involves producing high-quality scientific evidence,
building bridges between the producers and users of scientific evidence, and
incorporating scientific evidence into health policy and practice (4). Accordingly, the three primary areas in science-based policy are knowledge generation, knowledge exchange, and knowledge uptake
1). Within these three areas, the 12 essentials are categorized as follows: knowledge
generation — 1) credible design, 2) accurate data, 3) sound
analysis, and 4) comprehensive synthesis; knowledge exchange —
5) relevant content, 6) appropriate translation, 7) timely dissemination, and
8) modulated release; and knowledge uptake — 9) accessible
information, 10) readable message, 11) motivated user, and 12) rewarding
outcome (Table 1).
The names of the three areas described in this framework vary in other
articles. For example, knowledge generation (5,6) has also been
called knowledge acquisition (7) and knowledge creation (8); knowledge
exchange (6,9-11) has been called knowledge dissemination
(7,8,12), knowledge transfer (9,11), knowledge brokering
(10), knowledge translation (13), and knowledge access (5); and
knowledge uptake (6,9) has been referred to as knowledge
application (7,8), knowledge utilization (8,12), and knowledge
use (5). The meanings of the terms vary slightly. For example, the
term dissemination implies a one-way transmission of knowledge, whereas
the terms transfer and exchange imply a two-way transfer of
information (14). The term brokering seems to be associated with a
process, the objective of which is to exchange information (10).
Back to top
Ideally, evidence for policy decisions should be generated from scientific
research based on high-quality study designs. In general, the strength of data
generated by various study designs results in a hierarchical pattern.
Experimental studies such as clinical trials and field trials provide strong
evidence; community trials and observational studies such as cohort studies
and case-control studies provide moderate evidence; other observational
studies such as historical cohort studies, cross-sectional studies, and
ecological studies provide weak evidence; and case reports and news reports
provide minimal evidence (15-18).
However, the scientific evidence hierarchy is often turned
upside down when policy decisions are being made. News reports and case
reports often play an important role in policy decisions, because decision
makers, including those in the general public, often do not have the time,
ability, or expertise to access and synthesize the evidence from high-quality
studies. For example, in 1999, the newspaper USA Today published the
following health-related headlines: “‘Scars’ May Be Cancer Predictor,”
“Persistent Heartburn Is a Cancer Warning Sign,” “Two Drinks a Day Keep
Stroke Away,” “Study: High-Fiber Diets Don’t Cut Colon Cancer,” and
“No Link Found Between Fat, Breast Cancer” (19). News headlines can be
based on inconclusive evidence (e.g., “may be”), scare tactics
(e.g., “warning sign”), disregard of details (e.g., the health
risks of drinking, such as liver disease), and conflicting messages (e.g.,
reporting results that are different from numerous other studies).
Even when scientific evidence is produced from adequately designed studies,
current knowledge generation can be hindered by a false-positive research cycle
(Figure) (20). Consider the following scenario. Evidence
relating cellular telephone use and brain tumors is still inconclusive,
despite the multiple studies that have been done and the widespread attention
given to the topic (21). Assume the null hypothesis is true — that cellular telephone use
does not cause brain tumors. In addition, assume that as a result of
chance or bias (and not a high-quality study), some researchers report finding an association between cellular
telephone use and brain tumors (a false-positive study). Publication of the
false-positive study creates concern, and the problem becomes a hot topic (hot
topic bias) which results in more studies — perhaps even 100 — that are
investigate the potential problem (22). At the conventional significance
a level, of .05, five of the 100
studies are expected to have positive results (23); in other
words, five of the studies in this example would be expected to find that cellular
telephone use causes brain tumors. The researchers who obtain positive results
are more likely to document their results and submit papers to a journal
(positive results bias), and journal editors are more likely to publish
studies with positive results (editor’s bias) (20). In other
words, the five positive studies (which are actually false-positive studies)
are more likely to be published, and few of the other 95 studies that did not
find an association between cellular telephone use and brain tumors will be published (publication bias). The five additional false-positive
studies make the topic even more urgent in the research community, and the
false-positive research cycle begins again as more studies are designed to
assess the problem. Through this biased process, researchers can often “prove”
something out of nothing.
The false-positive research cycle.
Bias is defined as the “deviation of results or inferences from
the truth, or processes leading to such deviation” (24). The best way to
identify bias is by comparing results with the truth, or a gold standard. For
example, researchers conducted a study to determine the baseline accuracy of
dentists’ readings of dental radiographs (bitewings) (25). The study’s
methodology involved constructing 15 models of the posterior part of a natural
dentition. The models had extracted teeth mounted in a medium with a radiodensity
similar to that
of human bone. Bitewing radiographs were taken of the simulated dentitions. The teeth
used in the model mounts were removed from the models, serially sectioned, and
examined with a microscope for caries. The results of the microscopic
examination were established as the gold standard. Dentists were asked to read
independently the 15 sets of bitewing radiographs and make treatment decisions
about the teeth. The agreement between the
dentists’ readings and the gold standard established by the microscope
results was poor (mean κ = 0.35) (25).
Even laboratory tests cannot guarantee the accuracy of a study’s data.
For example, many physicians use four different types of laboratory tests to
diagnose leukemia (routine morphology testing, electron microscopy, cell surface
marker identification, and cancer cytogenetics), and the four test results often seem
contradictory. An interrater agreement study was conducted, with each of the
four laboratories being classified as a rater. The interrater reliability
result confirmed that the results were
contradictory. Results from the four diagnostic laboratories correlated poorly
for cell type identification in leukemia (pairwise κ = 0.17–0.40) (26).
Health data often come from health surveys using questionnaires, and the
accuracy of the data is likely affected by questionnaire biases. For example,
questions or answers may be phrased in a way that misleads respondents and
causes them to make an incorrect choice (framing bias) (27). An example
of framing bias follows:
Which operation would you prefer?
An operation that has a 5% mortality
[ ] An operation that 90% of the patients will survive
People may choose the second option when they read “90%” and “survive,”
even though a 90% survival rate (which is a 10% mortality rate) is actually
worse than a 5% mortality rate.
According to a comprehensive assessment of 109 instances of bias that were
found in scientific research (literature review, 4; study design, 31; study
execution, 3; data collection, 46; analysis, 15; interpretation, 7;
publication, 3), most of the instances of bias were found in the data
collection phase of research (46 of 109, or 42%, of the total instances) (22).
Failure to control for confounding effects is a common problem in data
analysis. A confounder is a factor “that can cause or prevent the
outcome of interest, is not an intermediate variable, and is associated with
the factor under investigation” (24). For example, if researchers were
studying the association between drinking alcohol and lung disease, they would
need to treat smoking as a potential confounder because 1) smoking is known to
cause lung disease, and 2) smoking and drinking alcohol are often associated
Techniques to control for confounding include stratification and mathematical
Failing to conduct a sound data analysis could completely change the
results of a study. In a mass screening for colorectal cancer, Zheng et al
evaluated the accuracy of occult blood testing, using rectoscopy as the gold
standard for comparison (30). Clinical and epidemiological data from 60,496 individuals were
collected. It was found that of the 477 individuals who had colorectal cancer
diagnosed by rectoscopy (the gold standard), 437 were identified as having
colorectal cancer by the occult blood test. This corresponded to a test
sensitivity of 92% (437/477), which indicated that the occult blood test was a
good screening test for colorectal cancer. The results were submitted to a
scientific journal, and comments from two reviewers were received. One
reviewer was pleased with the study and recommended publication. The other
reviewer pointed out a gross error in the calculations and mentioned “work-up
bias.” According to the original paper on work-up bias, it is not an easy
issue to address (31). An appropriate mathematical procedure was subsequently
developed to address the work-up bias (32). Using the new procedure, the
occult blood test sensitivity was recalculated to be 28%, indicating that it
was not a good screening test for colorectal cancer (30). Therefore, the
proper analysis completely reversed the study’s conclusion.
Scientific papers are being published constantly. Approximately 30,000
biomedical journals are being published currently, and 17,000 new
biomedical books are published every year.
On average, physicians would have to read 19 articles each day to
stay knowledgeable about new developments in their field (33,34).
Comprehensive syntheses of current information are needed to address potential
problems such as lack of time and lack of expertise (35). Comprehensive
syntheses include narrative reviews, systematic reviews, meta-analyses,
meta-databases, inventories of best practices, and public health
observatories. They make life easier for consumers of scientific material,
such as scientists, physicians, and policymakers, collectively referred to as knowledge
users in this article.
A narrative review is a summary of the literature that exists on a
particular topic; informal and subjective methods are used to collect and
interpret information (33,36). A systematic review is a summary that is
written after a comprehensive search for relevant studies and then evaluated
and synthesized according to a predetermined and an explicit method
(33,37,38). A meta-analysis (an analysis of several analyses) takes a
systematic review one step further by mathematically aggregating available
data from independent studies to yield a more statistically powerful estimate
(33,36,39). A meta-database (a database of several databases) includes
information about the location, source, content, and other details of the
relevant databases (40). An inventory of best practices (or better practices)
is created using an approach based largely on less rigorous study designs of
practices and programs. The inventory often focuses on particular organizational
behaviors for which conclusive quantitative evaluations are difficult to
design and execute (41). A public health observatory is more detached from
actual health phenomena and events, provides objective descriptions and analyses, and
provides forecasting of patterns,
interrelationships, processes, and public health outcomes (42,43).
Comprehensive syntheses can be a major undertaking. For example, a lifestyle
modification guide was created to prevent and control hypertension. It was a
50-page supplementary issue of a scientific journal based on a review of 37
years (1960 to 1996) of scientific literature on weight, alcohol, exercise,
sodium, calcium, magnesium, potassium, and stress and their effects on the
Some comprehensive syntheses require a review of not only contemporary
literature but also historical literature. For example, one analysis was
composed of 12 lessons for public health surveillance in the twenty-first
century. The lessons were created after conducting a broad review of the
historical documents on major epidemics during the past 5000 years (since 3180
BC) and included the plague, smallpox, dancing mania, cholera, the Spanish
flu, and lung cancer (45).
Back to top
Information should not be disseminated all at once and should not be
provided to everyone. Only relevant information needs to be disseminated. For
example, depending on the audience, one of two information dissemination
approaches can be used: the encyclopedia approach or the fire-alarm approach
(46). The encyclopedia approach involves conveying all available
information in the form of reports, atlases, Web sites, and other methods.
This type of information is needed by knowledge users such as scientists and
certain policymakers who need extremely detailed information.
policymakers and the general public, the fire-alarm approach may be more
appropriate. This approach involves only conveying information when selected
indicators are not in the normal range and indicate a potential problem. For example, it has been proposed that new composite indicators for public
health, similar to economic indicators such as the Dow Jones average or the
consumer price index, be developed to document the relevant health information
needed for public health decisions (20,47). Many stockholders trade
successfully by buying or selling their stock holdings based on the
performance of economic composite indicators. In a similar way, indicators
such as a national health index, national heart health index, and national
diet index could be helpful to health policymakers.
As scientists make new discoveries, more sophisticated methods and theories
are developed. At some point, the average policymaker and even some scientists
cannot understand the information. The key is to strike a balance between providing all available information and providing what is needed by knowledge users. “Complex models with simple model-user
interface” can be used to achieve this goal (48). Following is an example of
how such a model-user interface was created for public health practitioners.
During the first months of the severe acute respiratory syndrome (SARS)
epidemic in 2003, a mathematical model was developed to predict the spread of
SARS (49). The mathematical model can be thought of as a machine, with the
engine of the machine comprising a series of four mathematical equations:
Cti = R0t
C = ∑ Cti
Dti + d = Cti
D = ∑ Dti + d
indicates the predicted number of incident cases on day ti,
and t is time expressed in the number of incubation periods; C, the predicted
total number of cases; Dti + d, the
predicted number of deaths on day ti + d, and t is
time expressed in the number of incubation periods; D, the predicted total
number of deaths; R0, the basic reproductive number (i.e., the
expected number of new infectious cases per infectious case); F, the
case-fatality rate (i.e., the proportion of cases who die within the
symptomatic period); i, the incubation period (i.e., the time from infection
to symptoms); and d, duration of disease (i.e., the time from symptoms to
recovery or death).
These equations are complex but do not have to be understood to be used,
just as a person who drives a car does not have to understand how the engine
works. The model-user interface is simple. The required information for using the
previous SARS model to predict the number of SARS cases and deaths consists of
only R0, F, i, and d, and the result is a set of several
line graphs showing the predicted and observed numbers of SARS cases and
deaths. The deviation of the observed numbers from the predicted numbers
indicates the success of infection control measures (49).
For the general public, an effective yet simple and basic way to convey, or translate, complex information is
by using health proverbs (50). Sayings such as “an apple a day keeps the
doctor away” (51) have helped convey important health messages through
the years. They
were created by our ancestors, and we
have the responsibility to create new science-based health proverbs for future
Public health practitioners can learn about knowledge translation
weather forecasters (52), who use symbols (such as a sun partly covered by
clouds) and maps to explain the weather. Symbols could be used to denote
public health events, and the public could receive short- and long-term public
health forecasts and public health alerts, complete with color-coded maps to
illustrate public health problems in space and time.
Timely dissemination of information requires an ongoing information
distribution mechanism. For example, 365 health indicators relevant to the
general public could be developed, with one per day being discussed on the
evening news (20). After the news and the weather forecasts, the reporter
could discuss one of the indicators, such as air pollution during the previous
5 years and its predicted relationship to asthma in the next 3 years. The
public would not be expected to watch the news without fail, but if the
information dissemination occurred daily, the public’s awareness and
knowledge would increase with time (53).
In Canada, approximately 167,456 deaths result from chronic diseases each year.
A chronic disease clock was developed by the Public Health Agency of
Canada to disseminate information in real
time on its Web site (54). The chronic
disease clock is a digital clock with two categories: chronic-disease–related
deaths so far this year and chronic-disease–related
deaths so far today (as of 12:00 midnight). People can actually watch the number of
deaths attributable to chronic disease increase every few minutes because one
death occurs every 3 minutes in Canada. The clock keeps running 24 hours per
day, 365 days per year.
The general public is overwhelmed by health information. The end result is
that they do nothing to improve their health because they do not know how to
begin the process. The various types of available information need to be prioritized and
disseminated in stages.
For example, the World Health Organization’s The World
Health Report contains an immense amount of information (55). Chapter 4 of
The World Health Report 2002 is about major health risks. In
industrialized countries, the leading risk factors for chronic diseases are
tobacco use, high blood pressure, excessive alcohol consumption, high cholesterol,
overweight, low fruit and vegetable intake, and physical inactivity. The four
major chronic diseases in terms of resulting disability are
cardiovascular disease, cancer, chronic respiratory diseases, and neuropsychiatric disorders (55). The information in the chapter can be prioritized
for modulated release (56) in three steps. To promote health, the public is told to play it
SAFE (with the acronym SAFE representing smoking, alcohol, food, and
exercise) — refrain from smoking, drink alcohol in moderation, eat a
balanced diet, and increase physical activity. If they do not play it SAFE,
they have to call a COP to assess the situation (with COP
representing cholesterol, obesity, and pressure) — go for annual medical
examinations to assess blood cholesterol levels, weight, and blood pressure.
If they do not play it SAFE and call a COP, they have to expect
HARM (with HARM representing heart disease, abnormal growth, respiratory
disease, and mental disorders) — in the form of chronic conditions such as
heart disease, cancer, lung disease, and mental disorders. They would then
have to seek treatment. SAFE–COP–HARM concisely
summarizes the important information (56).
Back to top
Scientific findings must be published in accessible formats. For example,
information posted on a Web site may be considered accessible; however, some
people do not have access to the Internet. Even people who do have Internet
access may have difficulty retrieving a specific piece of information. For example, a Google search of the Internet using the key
words health information resulted in 13,200,000 Web sites (53).
Various unique information dissemination tools have been invented. For
example, executives at Xerox’s Palo Alto Research Center (Palo Alto, Calif)
can monitor the company’s overall share price by watching an office
fountain. The water flow is controlled through an Ethernet connection to a
computer that has the latest stock data. Flow strengthens when the price
New ways to actively market information and make it accessible to
various populations are needed. A group of experts at an occupational health
workshop for Latin Americans suggested unique ideas such as writing folk songs
for the radio on the health effects of pesticides and organizing concerts with
themes related to healthy living (58). The Brazilian Ministry of Health
distributes a free package of two decks of playing cards, and one health
message is written on each card, for a total of 104 health messages. Messages
include tips such as “Take a walk with your dog for 30 minutes to burn up to
200 calories” and “Increase your fruit and vegetable consumption to five
times a day.” Other ways to make information accessible include
incorporating messages into theatrical performances and story-telling
To be understood by different audiences, a message must be conveyed in
relevant terms. For example, Canadian policymakers
readily understand the economic and health impact of smoking on society. For
them, a relevant message would be that eliminating tobacco use for 1 year
in Canada would save $16.5 billion and prevent 47,000 deaths per year (59).
This message may not be relevant to members of the general public who are
not interested in policy and economics but are passionate about sports.
Instead of telling them about how much society will save if they quit smoking,
you could tell them how many important sports events, such as Stanley Cup hockey
playoffs, World Cup international soccer games, or Super Bowl football
playoffs, they would miss in their lifetime if they continued smoking (59).
For younger audiences, a relevant message such as “smoking makes you ugly”
could be an innovative way to convey smoking-related information (60). Teenage
smokers who do not care about the long-term health effects of tobacco
smoking may be able relate to the more immediate effects on
appearance, such as smoking-induced facial wrinkles (61,62) and baldness (63).
It is important to raise awareness of how scientific evidence can be used
to make health policy decisions. The key is to create an atmosphere in which
knowledge users are interested in and seeking out scientific knowledge rather
than being inundated with unwanted information. Knowledge users can be
motivated in many ways, and education plays an important role. Presenting
facts is not enough. For example, after returning home from a doctor’s office,
a colleague’s teenage son told her that his doctor told him he was obese. The
boy then said that he really did not need to worry about the problem because
obesity was so common. The boy had the facts but was not motivated to do
anything about them.
Educating people by teaching them about the severity and consequences of a health problem helps motivate
For example, obese people need to know that they have a higher risk of
developing chronic diseases such as diabetes and heart disease. Knowing the number of people who became blind or had limbs
amputated because of diabetes would be a better way to drive home the
ramifications of diabetes than simply stating the number of people who had
Policymakers and the general public must be convinced that using science
for making health decisions will be beneficial and have a noticeable impact on
their health — in other words, that it will have a rewarding outcome. For
example, mathematical prediction models could help policymakers evaluate how
various policies will affect a particular situation. To help show the general
public how scientific evidence can be used to make health decisions and
improve their health, computer software could be used to calculate the
probability of disease risks or overall health outcomes based on input related
to personal lifestyle choices, demographics, diet, and smoking (20). For
example, a 20-year-old man in excellent health may find out that he is
expected to live 75 years. The computer program could be used to show him
that if he were to start smoking, he would only be expected to live 67
years (64). The 8-year difference may be rewarding enough for him to decide
not to start smoking.
Back to top
The science-based policy framework of knowledge generation, knowledge
exchange, and knowledge uptake has similarities to Boyer’s research (65).
Boyer studied the concept of scholarship and distinguished four
kinds of scholarly pursuits: discovery, integration, application, and teaching
(65). Many parallels exist between Boyer’s work and the framework described
in this article: Boyer’s discovery category parallels the framework’s
knowledge generation area, his integration category parallels the knowledge exchange area,
and his application category parallels the knowledge uptake area. Overall, education is important in all three areas of the framework.
Corresponding to the 12 essentials are 12 recommendations for the future
(Table 2). It is hoped that these recommendations will stimulate additional
research and provide evidence for the necessity of a strong evidence base in
public health policy.
Back to top
Views presented in this article are those of the author and cannot be
attributed to the Public Health Agency of Canada, the University of Ottawa, or
the University of Toronto. The article is based on an invited presentation at
the Evidence-based Decision Making seminar on January 27, 2004, jointly
sponsored by Health Canada and the University of Ottawa, Ottawa, Ontario; an invited presentation at Johns Hopkins University on September 17,
2004, Baltimore, Md; a seminar at the University of Toronto on February
24, 2005, Toronto, Ontario; and a seminar at the University of British
Columbia on May 13, 2005, Vancouver, British Columbia.
Back to top
Corresponding Author: Bernard C.K. Choi, PhD, MSc, Centre for Chronic Disease
Prevention and Control, Public Health Agency of Canada, PL 6701A, 120
Colonnade Rd, Ottawa, Ontario, Canada K1A 1B4. Telephone: 613-957-1074. E-mail: Bernard_Choi@phac-aspc.gc.ca.
Author Affiliations: Dr Choi is also affiliated with the Department of Epidemiology
and Community Medicine, University of Ottawa, Ottawa, Ontario, and the
Department of Public Health Sciences, University of Toronto, Toronto, Ontario.
Back to top
- Tang KC, Ehsani JP, McQueen DV.
Evidence based health promotion:
recollections, reflections, and reconsiderations. J Epidemiol Community
- Scheel IB, Hagen KB, Oxman AD.
The unbearable lightness of healthcare policy making: a description of a
process aimed at giving it some weight. J Epidemiol Community Health
- Merriam-Webster Dictionary On-Line [database online] Springfield (MA):
Merriam-Webster, Inc [cited 2004 Oct 20]. Available
from: URL: http://www.m-w.com/cgi-bin/dictionary*.
- Haynes B, Haines A.
Barriers and bridges to evidence based clinical practice. BMJ 1998;317:273-6.
- CAB International. Knowledge for development. Oxfordshire (UK): CAB
International; 2004 [cited 2005 Mar 3]. Available from: URL: http://www.cabibioscience.org/Html/K4DYellow.htm*.
- Centre for Global eHealth Innovation. The Global eHealth Innovation
Network. Ontario, Canada: Centre for Global eHealth Innovation; 2005
[cited 2005 Mar 3]. Available from: URL: http://www.ehealthinnovation.org/splash/ehealthflash*.
- Knowledge acquisition, dissemination and application [cited 2005 Mar
3]. Available from: URL: http://www.ilabsgroup.com/rcenter/Knowledge%20
- The Knowledge Management Forum. What is knowledge management
[Internet] [cited 2005 Mar 3]? Available
from: URL: http://www.km-forum.org/what_is.htm*.
- Lavis JN. Enhancing the uptake of research knowledge in health policy.
Presentation slides; 2004 May 6 [cited 2005 Mar 3]. Available from: URL: http://www.researchtopolicy.ca/presentations/
- Canadian Health Services Research Foundation. Knowledge brokering.
Ottawa, Canada: Canadian Health Services Research Foundation; 2004 [cited
2005 Mar 3]. Available from: URL: http://www.chsrf.ca/keys/use_knowledge_e.php*.
- Institute for Work and Health. Knowledge transfer and exchange.
Toronto, Canada: Institute for Work and Health; 2004 [cited 2005 Mar 3]. Available from: URL: http://www.iwh.on.ca/kte/kte.php*.
- National Center for the Dissemination of Disability Research. NIDRR’s
long range plan — knowledge dissemination and utilization.
Austin (TX): National Center for the Dissemination of Disability Research;
2001 [cited 2005 Mar 3]. Available from: URL: http://www.ncddr.org/relativeact/kdu/lrp_ov.html*.
- Canadian Institutes of Health Research. Knowledge translation overview.
Ottawa, Canada: Canadian Institutes of Health Research; 2004 [cited 2005
Available from: URL: http://www.cihr-irsc.gc.ca/e/7518.html*.
- Health Canada. Getting the right knowledge to the right people at the
right time: an invitational roundtable on knowledge transfer. Ottawa,
Canada: Health Canada; 2004 [cited 2004 Aug 6]. Available from: URL: http://www.hc-sc.gc.ca/hpfb-dgpsa/nhpd-dpsn/get_right_ knowledge_roundtable_e.pdf*.
- Choi BCK, Pak AWP, Leake JL. Epidemiologic design options for health
risk studies on dental amalgam. Can J Community Dentistry 1996;11:7-12.
- Beaglehole R, Bonita R, Kjellstrom T. Basic epidemiology. Geneva,
Health Organization; 1993.
- Corbin SB. Concepts of modern risk assessment and management. J Am
College Dentists 1994;61:17-25.
- Centre for Evidence Based Medicine. Levels of evidence and grades of
recommendation, 2001 May. Oxford, England: Centre for Evidence Based
Medicine; 2001 [cited 2004 Oct 22]. Available from: URL: http://www.cebm.net/levels_of_evidence.asp*.
- Harvard University. Health insight – take charge of health
information. Boston (MA): Harvard University; 2004 [cited 2004 Oct 20]. Available
from: URL: http://www.health-insight.harvard.edu/*.
- Choi BCK. Perspectives on epidemiologic surveillance in the 21st
century. Chron Dis Canada 1998;19:145-51.
- Microwave News. Mobile phone — brain tumor risk in the limelight
again: focus on Swedish and U.S. epi studies. New York: Microwave News; 1999 May/Jun [cited 2004 Oct
20]. Available from: URL: http://www.microwavenews.com/m-j99vws.pdf*.
- Choi BCK, Pak AWP. Bias, overview. In: Armitage P, Colton T, editors. Encyclopaedia of
biostatistics. 1st ed. Chichester, Sussex (UK): John Wiley & Sons Ltd;
1998. p. 331-8.
- Western Kentucky University. Course BIOL 483. Lecture on “Hypothesis
Testing, Type I and Type II Errors.” Bowling Green (KY): Western Kentucky
University [cited 2004 Oct 20]. Available from: URL: http://bioweb.wku.edu/courses/Biol483/483lects3.htm*.
- Last JM. A dictionary of epidemiology. New York (NY): Oxford University
- Choi BCK, Jokovic A, Kay EJ, Main PA, Leake JL.
Reducing variability in
treatment decision making: effectiveness of educating clinicians about
uncertainty. Medical Educ 1998;32(1):105-11.
- Choi BCK, de Harven E, Bailey DJ, Dube ID, Pantalony D.
identification in leukemia: the level of agreement among four independent
diagnostic methods. Cancer Detect Prev 1994;18(5):383-91.
- Choi BCK, Pak AWP.
A catalog of biases in questionnaires. Prev Chronic
Dis [serial online] 2005 Jan.
- Choi BCK, de Guia N, Walsh P.
Look before you leap: stratify before you
standardize. Am J Epidemiol 1999;149(12):1087-96.
- Rothman K, Greenland S. Modern epidemiology. Hagerstown (MD): Lippincott-Raven;
- Zheng G, Choi BCK, Yu X, Zou R, Shao Y, Ma X.
Mass screening for rectal
neoplasm in Jiashan County, China. J Clin Epidemiol 1991;44:1379-85.
- Ransohoff DF, Feinstein AR.
Problems of spectrum and bias in evaluating
the efficacy of diagnostic tests. N Engl J Med 1978;299(17):926-30.
- Choi BCK.
Sensitivity and specificity of a single diagnostic test in
the presence of work-up bias. J Clin Epidemiol 1992;45:581-6.
- Klassen TP, JadadAR, Moher D.
Guides for reading and interpreting
systematic reviews. Arch Pediatr Adolesc Med 1998;152:700-4.
- Davidoff F, Haynes B, Sackett D, Smith R.
Evidence-based medicine. BMJ 1995;310:1085-8.
- Choi BCK, McQueen DV, Rootman I. Bridging the gap between scientists
and decision makers. J Epidemiol Community Health 2003;57:918.
- Egger M, Smith GD.
Meta-analysis. Potentials and promise. BMJ 1997;315:1371-4.
- Delaney BC, Hyde CJ, McManus RJ, Wilson S, Fitzmaurice DA, Jowett S,
Systematic review of near patient test evaluations in primary care.
- Morrison DS, Petticrew M, Thomson H.
What are the most effective ways of improving population health through
transport interventions? Evidence from systematic reviews. J Epidemiol
Community Health 2003;57:327-33.
Egger M, Smith GD, Phillips AN.
Meta-analysis: principles and procedures. BMJ 1997;315:1533-7.
- Froese R, Pauly D, editors. FishBase. World Wide Web electronic
publication. Glossary Searched Term: Metadatabase [cited 2004 Oct 22]. Available
from: URL: http://www.fishbase.org/Glossary/ Glossary.cfm?TermEnglish=metadatabase*.
- F/P/T Committee of Officials for the Ministers Responsible for Seniors.
An inventory of Canadian programs for the prevention of falls among seniors
living in the community. Ottawa, Canada: Health Canada; 2001 [cited
2004 Oct 22]. Available
from: URL: http://www.hc-sc.gc.ca/seniors-aines/pubs/ inventory/pdf/Inventory_e.pdf*.
- Hemmings J, Wilkinson J.
What is a public health observatory? J Epidemiol Community Health
- Liverpool Public Health Observatory. Liverpool, England: The
University of Liverpool; 2004 [cited 2004 Oct 22]. Available
from: URL: http://www.liv.ac.uk/PublicHealth/obs/LPHO.htm*.
- Campbell NRC, Burgess E, Choi BCK, Taylor G, Wilson E, Cléroux J,
et al. Lifestyle modifications to prevent and control
Med Assoc J 1999;160(9 suppl):S1-6.
- Choi BCK, Pak AWP.
Lessons for surveillance in the 21st century:
historical perspective from the past 5 millennia. Soz Praventivmed 2001;46:361-8.
- Choi BCK, Orlova A, Marsh M, Issa N, Morrison H.
dissemination approaches for public health decision-makers: encyclopaedia and
fire alarm. J Epidemiol Community Health 2004;58:634.
- Choi BCK, Pak AWP, Ottoson JM.
Understanding the basic concepts of public health surveillance. J
Epidemiol Community Health 2002;56:402.
- Choi BCK.
Future challenges for diagnostic research: striking a balance between
simplicity and complexity. J Epidemiol Community Health 2002;56:334-5.
- Choi BCK, Pak AWP.
A simple approximate mathematical model to predict the number of severe
acute respiratory syndrome cases and deaths. J Epidemiol Community
- Choi BCK, Pak AWP, Choi JCL, Choi ECL.
Health proverbs. J Epidemiol
Community Health 2004;58:1010.
- Simpson J. The concise oxford dictionary of proverbs. New York (NY): Oxford
University Press; 1992.
- Choi BCK.
Public health practitioners can learn from the weather forecasters. J
Epidemiol Community Health 2004;58:450.
- Choi BCK.
Innovative ideas needed for timely and effective information
dissemination. J Epidemiol Community Health 2005;59:259.
- Centre for Chronic Disease Prevention and Control, Public Health
Agency of Canada. Chronic disease clock. Ottawa, Ontario: Centre for
Chronic Disease Prevention and Control [cited 2004 Nov 16]. Available from:
- World Health Organization. The World Health Report 2002. Geneva,
Switzerland: World Health Organization; 2002 [cited 2004 Oct 22]. Available from: URL: http://www.who.int/whr/2002/en/*.
- Choi BCK.
Modulated release of health risk information to the general
public with the use of mnemonics. J Epidemiol Community Health
- National Post. Sowing PARC’s next revolution: Xerox squandered
the 1970s innovations of its Palo Alto Research Center, but hopes to make
amends. Ontario, Canada: CanWest Interactive Inc; 1999 Apr 21. p. C13. Available from: URL: http://financialpost.informart.ca/ar/ar_form.php*.
- Choi BCK, Eijkemans GJM, Tennassee LM.
Prioritization of occupational
sentinel health events for workplace health and hazard surveillance: the Pan
American Health Organization experience. J Occup Environ Med
- Choi BCK.
Understanding the basic principles of knowledge translation. J
Epidemiol Community Health 2005;59:93.
- Canny AM, Goodrich TW.
Smoking makes you ugly – an innovative
approach to smoking cessation. AORN J 2001;74:722-5.
- Koh JS, Kang H, Choi SW, Kim HO.
Cigarette smoking associated with
premature facial wrinkling: image analysis of facial skin replicas. Int J
- Goldstein N. Tobacco isolated as a cause of skin aging still another
reason to quit smoking! Hawaii Med J 2002;61:272.
- Trueb RM.
Association between smoking and hair loss: another
opportunity for health education against smoking? Dermatology
- Insurance Toronto. Life Expectancy Calculator. [Internet] Ontario,
Canada: Insurance Toronto [cited 2004
Oct 20]. Available from: URL: http://www.insurancetoronto.com/calculators/
- Boyer EL. Scholarship reconsidered: priorities of the professorate.
Princeton (NJ): The Carnegie Foundation for the Advancement of Teaching; 1990.
Back to top
*URLs for nonfederal organizations are provided solely as a
service to our users. URLs do not constitute an endorsement of any organization
by CDC or the federal government, and none should be inferred. CDC is
not responsible for the content of Web pages found at these URLs.