Diabetes Research and Clinical Practice - Volume 120 - Supplement 1 - October 2016

227 / 244

Ewha Womans University School of Medicine, 2Clinical Trial Center,
Ewha Womans University Medical Center, Korea
The relationship between bodymass index (BMI) andmortality
is still controversial. Furthermore, the association between
BMI and cardiovascular events is not conclusive, and may
differ by ethnicity. This study aimed to estimate the associ-
ation between BMI and mortality and cardiovascular disease
including ischemic heart disease and stroke in Korean general
population.
This study was based on the sample cohort database released
from the Korean National Health Insurance Service (NHIS),
which is consisted of 1,025,340 subjects. We analyzed the
adults over 30 years who had taken the national health
examination at least once from 2002 to 2012. Hazard ratios of
death and cardiovascular event were calculated using Cox ’s
proportional hazards models with adjusting for age, smoking,
alcohol drinking status, physical activity level, income level,
and family history of cardiovascular disease.
During follow-up, 7,257 men and 3,801 women died, and
51,593 cardiovascular events (men: 26,684, women: 24,909)
occurred. Subjects with a BMI ≥ 30 kg/m2 and <25 kg/m2
showed a raised risk of death from overall mortality, and
those with a BMI < 25 kg/m 2 had an increased risk of cardio-
vascular mortality after adjustment for multiple variables.
The lowest risk of mortality was appeared in subjects with a
BMI of 25 –27.4 kg/m2. The cardiovascular events were lowest
in subjects with a BMI 20–22.4 kg/m2 and rose as the BMI was
increased.
BMI showed a U-shaped association with overall mortality
with lowest in subjects with slight obesity, whereas cardio-
vascular event exhibited a J-shaped pattern with lowest in
normal weight subjects in Korean general population.
PJ-56
Fifteen-year trends in lifestyle modification in patients with
diabetes mellitus: From KNHANES 1998 –2013
Hyun Min KIM 1*, Sun Ok SONG2, Jaetaek KIM1. 1Department of
Internal medicine, Chung-Ang University College of Medicine,
2 Department of Endocrinology and Metabolism, National Health
Insurance Service Ilsan Hospital, Korea
Healthy diet and adequate physical activity the initial, and
often the primary, component in the management of diabetes
mellitus. Furthermore, health-related behaviors such as
alcohol consumption and smoking have been known to be
related with risk of diabetic complication or cardiovascular
disease. In this study, we investigated trends in the achieve-
ment of goal of lifestyle modification in the patients with
diabetes mellitus in a representative Korean population.
From the Korea National Health and Nutrition Examination
Survey (KNHANES) conducted in 1998, 2005, 2009, and 2013,
daily intakes of total energy, carbohydrate, protein, fat, and
sodium were calculated based on the food items consumed.
Physical activity, alcohol consumption, smoking status, and
sleep duration were assessed from the questionnaire.
A total of 4,625 patients with diabetes mellitus were
analyzed. Patients from 2013 were more obese, however,
age, gender, and glycemic control status were not different.
From 1998 to 2013, diet in men had not changed. Only 30%
consumed adequate amount of carbohydrate, and though
group with moderate salt consumption slightly increased,
most patients still exceeded the recommended intake. In
women, consumption of carbohydrate was far much higher
and especially in even non-obese women, only 13% showed
adequate carbohydrate consumption. Most drastic change
appeared in physical activity. With exception of 1998,
analysis from 2005 to 2013 showed decrease in proportion
of sustaining moderate physical activity and vigorous phys-
ical activity, especially in obese men. Though men having
regular strength exercise increased a little, it decreased in
obese group. In women, moderate physical activity decreased
markedly regardless of obesity, which showed reflection
of sedentary lifestyle. Despite of decrease in current
cigarette smokers for 15 years, 40% of men still were
smokers. Alcohol consumption showed decrease on the
whole, but in men, one third still showed binge drinking
and the fraction was higher in obese group. As for sleeping
duration, men showed no significant difference, but in
women, it decreased a little.
Despite the importance of diet and physical activity in
diabetic patients is well known, actual dietary habit in
diabetes patients did not improve significantly for 15 years
and as for physical activity, they showed alarmingly insuf-
ficient level. There were some improvement in drinking and
smoking habit, however, still many patients showed current
smoking or binge drinking. Adoption and maintenance of a
healthy lifestyle should be emphasized in people with
diabetes.
PJ-58
Patterns of search queries of diabetes-related terms: An
infodemiological study using Google trends
Malcolm KOO1, Kun-Wei TSAI2, Shih-Chun LIN2*. 1Department
of Medical Research, Dalin Tzu Chi Hospital, Buddhist Tzu Chi
Medical Foundation, 2Division of Geriatrics, Dalin Tzu Chi Hospital,
Buddhist Tzu Chi Medical Foundation, Taiwan
Purpose: Google Trends has been used to explore the searching
trends of various health outcomes and behaviors such as
influenza, allergic diseases, dementia, multiple sclerosis,
epilepsy, cancer screening, smoking cessation, and behavior
change. The aim of this study was to investigate patterns of
search queries of diabetes-related Chinese terms using Google
Trends.
Methods: Google Trends(http://www.google.com.tw/trends/),
which allows Internet users to examine trends of certain
query terms by time, geographic location, and category was
used to search for the following query terms in Chinese,
“diabetes ”, “complications of diabetes”, “prevention of dia-
betes”, and “blood glucose”. The observation period was
limited to 10 years, containing 522 weeks from January 1,
2006 to December 31, 2015. The geographical region of the
search trend was limited to Taiwan. The results were
normalized to a range of 0 to 100, with 100 equals to the
peak value over the study time period with respect to the
search term. Cycle plots, including sequential plots and
seasonal subseries plots, were used to visualize cyclical
patterns in the data.
Results: Seasonality patterns were observed in the search
query terms. For the query term “diabetes ”, two peaks
appeared in March and May whereas one dip appeared in
August. Similarly pattern appeared when the English term,
with geographical location limited to Taiwan, was used. On the
other hand, for the query “complications in diabetes”, higher
volume was observed around March and April with two dips in
January, August, and October. For “ prevention of diabetes”, a
peak appeared in May and a dip appeared in November. For
“blood glucose”, the search volume was lower in January and
February.
Conclusions: Different seasonality patterns were observed
depending on the diabetes-related query terms. Certain peaks
can possibly be explained by the activities generated by
health professionals and students in relation to the time of
examination and major academic health conferences. While
Google Trends holds potential for easy access to aggregated
results of big data, there are limitations to its applicability
to derive meaningful insights about population health
behavior.
Poster Presentations / Diabetes Research and Clinical Practice 120S1 (2016) S65–S211 S209