International Journal of Obesity (2017) 41, 1074–1079

© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved 0307-0565/17
www.nature.com/ijo

ORIGINAL ARTICLE
Chilli consumption and the incidence of overweight and
obesity in a Chinese adult population
Z Shi1, M Riley2, AW Taylor1 and A Page1

BACKGROUND: The frequency of spicy food intake has recently been associated with a reduced risk of mortality in the Chinese
population. This study aimed to prospectively examine the association between chilli intake and the incidence of overweight/
obesity in a Chinese adult population.
METHODS: Adults aged 20–75 years in the China Health and Nutrition Survey were followed between 1991 and 2011. Dietary data
were collected during home visits using a 3-day food record in 1991, 1993, 1997, 2000, 2004, 2006, 2009 and 2011. Cox regression
was used in the analysis. Overweight/obesity was defined as body mass index ⩾ 25 kg m−2.
RESULTS: A total 12 970 adults were followed for a median of 9 years. During 126 884 person-years of follow-up, 3203 subjects
developed overweight/obesity. The absolute incidence rate of overweight/obesity was 26.4, 22.3, 24.4 and 20.5 per 1000 person-
years among those who consumed no chilli or 1–20, 20.1–50, ⩾ 50.1 g per day, respectively. Chilli consumption was therefore
inversely associated with the incidence of overweight/obesity. After adjusting for age, gender, energy and fat intake, smoking,
alcohol drinking and physical activity, those whose cumulative average chilli intake was 0, 1–20, 20.1–50 and ⩾ 50.1 g per day had a
hazard ratio for overweight/obesity of 1.00, 0.81 (95% confidence interval = 0.73–0.89), 0.77 (0.69–0.86) and 0.73 (0.63–0.84);
P for trend o0.001, respectively. There was no interaction between chilli intake and gender, income, education and residence
(urban/rural) in relation to the risk of overweight/obesity.
CONCLUSIONS: Chilli intake is inversely associated with the risk of becoming overweight/obese in Chinese adults.
International Journal of Obesity (2017) 41, 1074–1079; doi:10.1038/ijo.2017.88

INTRODUCTION
Worldwide, chilli is one of the most commonly used spices.1 There
is a large geographic and culture variation in chilli consumption.
For example, chilli intake is higher in Asian than European
countries.2 In China, more than 30% of adults consume spicy food,
including chilli, daily.3
The beneficial effects of chilli and its active component
capsaicin have been reported including reducing mortality,3
obesity4 and rhinitis,5 and increasing muscle strength.6 However,
a case–control study in Mexico found chilli consumption is
positively associated with stomach cancer.7
Over the past several decades, a substantial amount of research
has been conducted on the effects of chilli and the major pungent
constituent of chilli, capsaicin (8-methyl-N-vanillyl-6-nonenamide),
on energy balance.4,8 Existing evidence from animal studies and
small clinical trials suggest that chilli consumption may reduce
energy intake, increase energy expenditure and enhances fat
oxidation,4,8–11 especially among non-regular chilli consumers.12
Thus chilli consumption may have the potential to assist in obesity
prevention. However, the association between chilli consumption
and obesity has not been well studied in large population
studies. Only one cross-sectional study has assessed the associa-
tion between the frequency of spicy food consumption and
obesity.13 Contrary to the evidence from animal studies and
clinical trials, in the China Kadoorie Biobank (CKB) study, the
frequency of spicy food consumption was shown to be positively
associated with overweight/obesity among half a million Chinese
adults.13 However, the study measured frequency of intake and
other details of ‘hot spicy food’ rather than collecting detailed
food intake information, in particular of chilli intake. An inverse
association between spicy food consumption and mortality was
found in the CKB cohort3 independent of body mass index.
Another cross-sectional study from China showed that preference
for very hot chilli was inversely associated with insulin resistance
but this study did not adjust for body fatness.14
Using data from the China Health and Nutrition Survey (CHNS),
the aim of the current study was to prospectively examine the
association between chilli consumption and incident of over-
weight/obesity in a Chinese adult population.
MATERIALS AND METHODS
Study sample
The CHNS study is an ongoing open prospective household-based cohort
study that includes nine provinces in China (representing 553 million
people). Nine waves of data collection (that is, 1989, 1991, 1993, 1997,
2000, 2004, 2006, 2009 and 2011) have been conducted.15,16 The survey
uses a multistage random-cluster sampling process to select samples in
both urban and rural areas. All the members in the selected household
were invited to participate in the study, however, dietary intake in 1989
only involved middle-aged adults. Since 1997, new households in the same
community joined the survey as replenishment samples to those lost to
follow-up owing to rural migration and city construction. The response rate
based on those who participated in 1989 and remained in the 2006 survey
was above 60%. The survey was approved by the institutional review
committees of the University of North Carolina (USA) and the National
Institute of Nutrition and Food Safety (China). Informed consent was
obtained from all the participants. Between the 1991 and 2011 survey,

1
School of Medicine, University of Adelaide, Adelaide, SA, Australia and 2Commonwealth Scientific and Industrial Research Organisation (CSIRO), PO Box 10041, Adelaide, SA,
Australia. Correspondence: Associate Professor Z Shi, School of Medicine, University of Adelaide, Level 7, SAHMRI, North Terrace, Adelaide, SA 5005, Australia.
E-mail: Zumin.shi@adelaide.edu.au
Received 24 November 2016; revised 9 March 2017; accepted 19 March 2017; accepted article preview online 31 March 2017; advance online publication, 25 April 2017

there were 29 220 participants aged 20 years and older. We excluded
those without dietary intake (n = 13 793), had implausible energy intake
(men: 46000 kcal or o800 kcal; women: 44000 or o600 kcal) (n = 651),
pregnant women or breastfeeding (n = 798) or implausible body mass
index ( o14 or 445 kg m − 2, n = 38), and age 475 years (n = 2202) during
a survey year (otherwise included during other survey years). In total,
18 611 participated in at least two waves of data collection including
information on chilli intake. After excluding those who were overweight/
obese at baseline (n = 5641), 12 970 participants were included in the final
analysis.
Outcome variable: overweight/obesity
Height and weight were measured at each wave. Overweight/obesity was
defined as body mass index ⩾ 25 kg m − 2.
Exposure variables: chilli intake
At each wave, all foods and condiments in the home inventory, purchased
from markets or picked from gardens, and food waste, were weighed and
recorded by interviewers at the beginning and end of the 3-day food
consumption survey. Individual dietary intake data were collected by a
trained investigator conducting a 24 h dietary recall on each of the 3
consecutive days. The dietary recall was supported by dietary records kept
by the individual with the final dietary data including the type and amount
of food, the type of meal and the place of consumption. Cooking oil and
condiments consumption for each individual in the household was
estimated using household intake weighted by individual energy intake.
Detailed description of the dietary measurement has been published
previously.15 The dietary assessment method has been validated for
energy intake. The correlation coefficient between the reported energy
intake and total energy expenditure determined by the doubly labelled
water method was 0.56 (Po0.01) for men and 0.60 (Po 0.01) for women.17
Food consumption data were converted to nutrient intake using the
Chinese Food Composition Table.
Chilli in China is consumed as fresh food as well as dried product. We
calculated a cumulative average intake of chilli (sweet capsicum is not
included) for each individual at each time period to reduce variation within
individuals and to represent long-term habitual intake.18 For example,
the 1991 intake was used for the follow-up between 1991 and 1993, the
average of the 1991 and 1993 intake was used for the follow-up between
1997 and 2000 and so on.
Habitual intake of spicy food was only asked in 2009 by the questions
‘Do you like to eat hot pepper or spicy food? (i) No, (ii) sometimes (⩾2
times per week), (iii) often (three to four times per week), (iv) usually (⩾5
times per week), (v) unknown’; and ‘What kind of spicy food do you like?
(i) A little bit hot, (ii) moderate hot, (iii) very hot, (iv) unknown’.
Covariates
Detailed information on sociodemographic and lifestyle factors were
collected in each wave using a structured questionnaire. We included the
following constructed variables to reflect socioeconomic status: education
(low: illiterate/primary school; medium: junior middle school; and high:
high middle school or higher), per capita annual family income (recoded
into tertiles as low, medium and high), urbanisation levels15 (recoded into
tertiles as low, medium and high).
Physical activity level (metabolic equivalent of task) was estimated on
the basis of self-reported activities (including occupational, domestic,
transportation and leisure time physical activity) and duration using a
Compendium of Physical Activities. Smoking status was categorised as
non-smokers, ex-smokers and current smokers.
Blood sample measurement (in 2009 only) included fasting plasma
glucose and glycated haemoglobin.19 C-reactive protein was measured in
the blood via the immunoturbidimetric method with Denka Seiken, Japan
reagents. We defined diabetes as fasting plasma glucose 47.0 mmol l − 1,
glycated haemoglobin 46.5 or having known diabetes (self-reported
doctor diagnosed). Hypertension was defined as systolic blood pressure
above 140 mm Hg and/or diastolic blood pressure above 90 mm Hg or
having known hypertension.
Statistical analysis
Chilli intake was recoded into four levels: non-consumers, 1–20, 20.1–50,
⩾ 50.1 g per day.
© 2017 Macmillan Publishers Limited, part of Springer Nature.
Chilli intake and overweight/obesity among adults
Z Shi et al
1075
The chi-square test was used to compare differences between groups
for categorical variables and analysis of variance for continuous variables.
A random effect model using xtreg command in Stata was used to
assess the association between chilli intake and physical activity adjusting
for age, gender, energy intake and year of survey. We used Cox
proportional hazards models with time-varying cumulative chilli consump-
tion and covariates to compute hazard ratios. A set of models were used:
model 1 adjusted for age, gender and energy intake; model 2 further
adjusted for intake of fat, smoking, alcohol drinking, income, urbanisation,
education and physical activity. Model 3 further adjusted for two dietary
patterns (traditional south pattern and modern pattern, determined using
factor analysis based on our previous publication20). Traditional south
pattern is characterised by high intake of rice, pork and vegetables, and
low intake of wheat; a modern dietary pattern had high intake of fruit, soy
milk, egg, milk and deep fried products.
Cox proportional hazards assumptions were investigated by visual
inspection of log–log plots generated by stphplot syntax in Stata, showing
no deviations.
In sensitivity analyses, we also used baseline chilli intake, or most recent
chilli intake as the exposure variable. To assess the association between
cumulative chilli intake and weight change between 1991 and 2011
among all participants (including those who were overweight/obese at
baseline), we used mixed-effect linear regression adjusting for covariates
the same as model 3 mentioned above. All the analyses were performed
using STATA 14.1 (Stata Corporation, College Station, TX, USA). Statistical
significance was considered when Po0.05 (two-sided).
RESULTS
At baseline, participants with high chilli consumption had low
income, were more physically active and less likely to have
hypertension compared with non-consumers (Table 1). Energy
intake increased with the increase in chilli consumption. The age-
and gender-adjusted mean intake of chilli was around 15 g
per day between 1991 and 2011 (Supplementary Figure 1). The
median portion size for chilli intake was 50 g. In each survey,
around 30% of the participants consumed chilli during the 3 days.
Among the chilli consumers, about 30% had chilli intake above
50 g per day (data not shown). Compared with non-consumers,
moderate consumers ( o20 g per day) had a lower intake of
energy but high consumers had a higher intake of energy
(Supplementary Figure 2). Among chilli consumers, there was a
positive association between chilli intake and energy intake.
Cumulative chilli intake was positively associated with self-
reported frequency intake and the preference for hot chilli in
2009 (Supplementary Figure 3). Chilli consumption was positively
associated with the traditional dietary pattern (correlation coeffi-
cient 0.14, P o0.001) but inversely associated with the modern
dietary pattern (correlation coefficient − 0.06, P o0.001; data not
shown). Compared with non-consumers, moderate and high
chilli consumption was associated with 3.32 (0.13, 6.51) and 4.27
(0.80, 7.73) metabolic equivalent of task/week higher of physical
activity level after adjusting for age, gender and year of survey
(Supplementary Table 1).
Overall, 12 970 adults were followed for a median of 9 years.
During 126 884 person-years of follow-up, 3203 participants
developed overweight/obesity. Chilli consumption was inversely
associated with the incidence of overweight/obesity. The absolute
incidence rate of overweight/obesity was 26.4, 22.3, 24.4 and 20.5
per 1000 person-years across levels of cumulative average chilli
consumption of 0, 1–20, 20–50, 450 g per day, respectively
(Table 2). After adjusting for age, gender, energy and fat intake,
smoking, alcohol drinking and physical activity, those who ate
chilli 0, 1–20, 20–50 and 450 g per day had a hazard ratio for
overweight/obesity of 1.00, 0.81 (95% confidence interval:
0.73–0.89), 0.77 (0.69–0.86) and 0.73 (0.63–0.84) (P for trend
o0.001), respectively. After further adjusting for overall dietary
patterns, the association slightly attenuated but remained
statistically significant. Adjusting for hypertension did not change
the above association (data not shown). There was no interaction
International Journal of Obesity (2017) 1074 – 1079
 Chilli intake and overweight/obesity among adults
Z Shi et al
1076
Table 1. Baseline sample characteristics by baseline chilli intake categories

Factor None 1–20 g 20.1–50 g ⩾ 50.1 g P-value

per day per day per day

N 9221 1056 1336 1357

Energy intake (kcal per day), mean (s.d.) 2417.9 (697.0) 2378.3 (706.6) 2486.7 (706.6) 2748.0 (754.6) o0.001
Fat intake (g per day), mean (s.d.) 64.2 (36.2) 66.6 (35.9) 69.3 (40.2) 67.7 (43.3) o0.001
Protein intake (g per day), mean (s.d.) 72.1 (23.3) 71.9 (24.2) 72.9 (23.2) 79.1 (26.3) o0.001
Carbohydrate intake (g per day), mean (s.d.) 383.4 (138.6) 366.1 (139.3) 386.0 (133.8) 450.2 (151.6) o0.001
Traditional southern dietary pattern score, mean (s.d.) − 0.1 (1.1) 0.1 (1.0) 0.2 (0.9) 0.3 (1.1) o0.001
Modern dietary pattern score, mean (s.d.) − 0.2 (0.9) − 0.1 (1.0) − 0.3 (0.8) − 0.6 (0.8) o0.001
Age (years), mean (s.d.) 40.5 (15.4) 39.0 (15.0) 39.4 (14.9) 39.1 (14.4) o0.001
BMI (kg m − 2), mean (s.d.) 21.2 (2.0) 21.1 (2.0) 21.0 (2.0) 21.1 (2.0) 0.014

BMI status 0.90

Underweight (BMI o18.5 kg m − 2) 9.9% 10.6% 10.3% 10.2%
Normal (18.5 ⩾ BMI and o 25 kg m − 2) 90.1% 89.4% 89.7% 89.8%

Sex 0.005
Men 49.3% 49.2% 51.1% 54.3%
Women 50.7% 50.8% 48.9% 45.7%

Income o0.001
Low 30.1% 29.4% 29.7% 37.4%
Medium 32.9% 31.2% 31.9% 34.9%
High 37.0% 39.4% 38.4% 27.7%

Education o0.001
Low 46.4% 42.9% 44.0% 53.9%
Medium 32.9% 29.0% 33.0% 28.5%
High 20.7% 28.1% 23.0% 17.6%

Hypertension 10.2% 10.1% 8.0% 7.5% 0.004

Diabetes (baseline in 2009 only, n = 672; 458, 83, 74, 57) 7.4% 7.2% 8.1% 12.3% 0.64

Urbanisation o0.001
Low 42.6% 33.6% 37.9% 51.1%
Medium 28.3% 33.0% 32.3% 29.4%
High 29.1% 33.4% 29.7% 19.5%

Smoking 0.004
Non-smoker 65.8% 64.7% 61.4% 63.4%
Ex-smokers 1.9% 1.6% 1.3% 2.7%
Current smokers 32.4% 33.7% 37.2% 33.9%

Sleep duration (hours per day; baseline in 2004 onwards, n = 2425; 1553, 0.410
242, 246, 204)
7–9 79.7% 81.0% 83.3% 81.9%
⩽6 7.6% 7.0% 6.1% 5.9%
⩾ 10 12.7% 12.0% 10.6% 12.3%

Physical activity (MET, hours per week), mean (s.d.) 204.3 (171.4) 198.7 (168.4) 212.1 (190.3) 214.6 (174.7) 0.061
Abbreviations: BMI, body mass index; MET, metabolic equivalent of task.

between cumulative average chilli intake and gender, income, DISCUSSION

education and residence (urban/rural) in relation to the risk of
overweight/obesity. The above association was also found if we
use baseline chilli consumption and most recent chilli consump-
tion as exposure variables.
In the cross-sectional analysis of the 2009 data, 70% of the
participants reported consuming spicy food. Compared with non-
consumers of spicy food intake, consumers of spicy food was
positively associated with overweight/obesity (Supplementary
Figure 4). However, there was no dose–response relationship
between frequency/hotness of spicy food intake and overweight/
obesity.
High cumulative chilli intake was inversely associated with body
weight gain during follow-up using a fully adjusted mixed-effect
linear regression model (Supplementary Figure 5).
In this large prospective cohort study, we found that high intake
of chilli was inversely associated with the risk of overweight/
obesity independent of overall dietary pattern, energy intake and
lifestyle factors. High chilli intake was positively associated with
energy intake. To the best of our knowledge, this is the first
prospective study of chilli intake and the development of
overweight/obesity using population data.
Comparison with other studies
The inverse association between chilli intake and overweight/
obesity in our study is consistent with most of the animal and
small sample human studies10,11,21 but is inconsistent with the
findings of the CKB study.13 In the CKB study, the strength and
frequency of spicy food consumption was positively associated

International Journal of Obesity (2017) 1074 – 1079 © 2017 Macmillan Publishers Limited, part of Springer Nature.
 Chilli intake and overweight/obesity among adults
Z Shi et al
1077
Table 2. Hazard ratio (95% confidence interval) for incident overweight/obesity by chilli intake levels

Chilli intake (g per day)

None 1–20 20.1–50 ⩾50.1 P for trend

Cumulative average chilli intake

Cases 1842 596 494 271
Person-years 69 697 23 519 20 539 13 129
Incident rate (per 1000) 26.4 25.3 24.1 20.6
Model 1 1.00 0.78 (0.71–0.86) 0.76 (0.68–0.84) 0.69 (0.61–0.79) o0.001
Model 2 1.00 0.81 (0.73–0.89) 0.77 (0.69–0.86) 0.73 (0.63–0.84) o0.001
Model 3 1.00 0.81 (0.73–0.90) 0.81 (0.73–0.91) 0.82 (0.71–0.94) 0.001

Baseline chilli intake

Cases 2297 248 316 342
Person-years (1000) 8871 983 1318 1517
Incident rate (per 1000) 25.9 25.2 24.0 22.6
Model 1 1.00 0.96 (0.84–1.10) 0.90 (0.80–1.01) 0.82 (0.73–0.92) o0.001
Model 2 1.00 0.96 (0.83–1.11) 0.94 (0.82–1.06) 0.87 (0.77–0.99) 0.003
Model 3 1.00 0.96 (0.84–1.11) 0.98 (0.86–1.11) 0.94 (0.83–1.06) 0.098

Most recent chilli intake

Cases 2296 270 357 280
Person-years (1000) 88 132 10 419 14 192 14 142
Incident rate (per 1000) 26.1 25.9 25.2 19.8
Model 1 1.00 0.94 (0.83–1.07) 0.90 (0.80–1.01) 0.71 (0.63–0.80) o0.001
Model 2 1.00 0.95 (0.83–1.09) 0.94 (0.83–1.06) 0.71 (0.62–0.82) o0.001
Model 3 1.00 0.97 (0.84–1.11) 0.99 (0.88–1.12) 0.79 (0.69–0.91) 0.008
Model 1 adjusted for age, gender and energy intake. Model 2 further adjusted for intake of fat, smoking, alcohol drinking, income, urban, education and
physical activity. Model 3 further adjusted for dietary patterns20 (traditional south pattern is characterised by high intake of rice, pork and vegetables, and low
intake of wheat; a modern dietary pattern had high intake of fruit, soy milk, egg, milk and deep fried products). All the adjusted variables are treated as time-
varying covariates.

with adiposity. However, the findings may be confounded by the
cross-sectional study design and the lack of actual chilli intake
data. A Google search using ‘chilli’ and ‘weight loss’ in Chinese
yielded over 400 000 results suggesting a public perception that
eating chilli assists with losing weight. Similar to the findings of
the CKB study,13 using 2009 cross-sectional data, we also found a
positive association between spicy food intake and overweight/
obesity. In a recent study from CKB, it had been shown that
spicy food consumption was inversely associated with the risk of
mortality.3 Consistent with our findings, a previous study using
data from CHNS showed that preference for hot chilli was
inversely associated with insulin resistance.14 Thus, a reverse
causation mechanism seems likely in a cross-sectional study of
chilli intake and obesity: people with overweight/obesity may
increase their intake of chilli to manage their body weight. In our
cross-sectional analyses, no association was found between
preference for hot food and being overweight or obese. This
suggests that preference for spicy food may not be a good
indicator to study the association between chilli consumption and
obesity.
Potential explanations and mechanisms of findings
In our study, chilli consumption was positively associated with
the traditional dietary pattern but inversely associated with the
modern dietary pattern. Adjusting for overall dietary patterns
attenuated the association between chilli intake and the risk of
overweight/obesity but the association remained strong and
significant. It suggests that the association between chilli intake
and overweight/obesity is independent of overall dietary patterns.
Adding chilli to a high carbohydrate breakfast decreased the
desire to eat and hunger before lunch in 13 Japanese women,11
resulting in reduced energy intake. The effect of chilli on energy
intake was associated with an increase in the ratio of sympathetic:
parasympathetic nervous system activity in 10 Caucasian men.11
In another study of eight Caucasian adult males, the acute effect
of red pepper (in combination with caffeine) on reducing energy
intake was much greater than its concurrent effect on increasing
energy expenditure (3690 vs 320 kJ per day).21
In our study, the association of chilli intake and overweight/
obesity is independent of energy intake. In fact, there was a
positive association between high chilli intake and energy intake.
Compared with non-consumers, mean energy intake was greater
by more than 200 kcal per day in individuals who consumed
more than 50 g per day of chilli. Although, physical activity
was higher in the high chilli consumers compared with non-
consumers, the increased levels of physical activity could not fully
explain the energy intake difference. This association may be more
than the often observed increase of all dietary components with
increasing energy intake. Furthermore, vegetables are often stir
fried with vegetable oil in China. A positive association between
vegetable-rich food pattern and vegetable oil consumption has
been reported.22
In the current study, individuals with chilli consumption below
20 g per day and above 50 g per day had reduced and increased
energy intake, respectively compared with non-consumers.
In humans, it has been demonstrated that dietary capsaicin
enhances anorexigenic sensations such as satiety and fullness.23,24
Although the exact mechanism of action is unknown, there are
several physiological processes that could be involved. For
example, gastrointestinal vagal afferents relay information on
the quantity and type of nutrients consumed, and have an
important role in the short-term control of food intake and
meal size.25 These vagal afferents express the capsaicin-sensitive
transient receptor potential vanilloid channel 1 and the sensitivity
of these afferents is enhanced by capsaicin or capsaicin
analogues.26,27 However, capsaicin can also de-sensitise transient
receptor potential vanilloid channel 1.28 Therefore, we postulate
that at low levels of capsaicin intake, there is sensitisation of vagal

© 2017 Macmillan Publishers Limited, part of Springer Nature. International Journal of Obesity (2017) 1074 – 1079
 Chilli intake and overweight/obesity among adults
Z Shi et al
1078
afferents leading to increased satiety signalling and reduced food
intake. Further, above a certain threshold, capsaicin desensitises
vagal afferents leading to reduced satiety signalling and an
increase in food intake. However, this is highly speculative and
requires further investigation. Ludy and Mattes12 found that
adding 1 g of chilli to a meal reduced energy intake only in non-
regular chilli users but not in regular chilli users as compared with
controls.
It could be that energy expenditure induced by chilli intake has
a major role in the inverse association between chilli consumption
and overweight/obesity. Increased energy expenditure associated
with capsaicin intake has been replicated in several studies
involving different populations. Numerous studies have shown
that capsaicin has a potential thermogenic effect.4,11,21,29–31
Pain relief resulting from the consumption of chilli may improve
sleep and prevent weight gain.4 Sleep was assessed in the 2004,
2006 and 2009 surveys in CHNS. We did not find any association
between most recent chilli intake and short sleep. However, high
cumulative mean chilli consumption was positively associated
with long sleep duration (49 h per day) in the rural area (data not
shown). There may also be a potential ‘healthy participant effect’.
Participants feeling healthier (that is, those with lower prevalence
of obesity, hypertension and so on) may keep eating more chilli.
Finally, the antimicrobial properties of chilli may affect gut
microbiota and lead to a lower risk of developing overweight/
obesity32 by mechanisms not yet fully understood.
Strengths and potential limitations
Several strengths of the study should be mentioned. One of the
strengths of the study is the longitudinal study design and the use
of 3-day food records in combination with a household food
inventory. The use of multiple measures of chilli intake provides a
more robust estimate of long-term intake. Height and weight were
measured using standard methods by trained health workers
during the survey. The mean number of years of follow-up is
9 years. The sample size and the large number of incident
overweight/obese subjects provide sufficient statistical power. The
finding is generalisable in China.
A limitation to the study is that chilli intake reported in the
3-day survey period may not represent long-term intake. We were
not able to adjust for seasonal variation of chilli intake in the study
as the surveys were conducted mainly between September
and November. In each wave of the survey, only about 30%
reported chilli intake. Based on the self-reported frequency of hot
pepper and spicy food intake in 2009, 70% of the participants
consumed spicy food. The frequency question did not measure
chilli specifically and referred to preference, which may not reflect
behaviour. For this reason, we used cumulative average intake
of chilli. However, the findings were consistent when we used
different chilli consumption variables. There was a significant asso-
ciation between income and region and chilli intake. Although we
have adjusted for income, urbanisation and dietary pattern in the
multivariable analyses, residual confounding by these factors is
still possible. When we excluded participants from regions where
chilli consumption was most prevalent or least prevalent, the
association between chill intake and obesity remained (data not
shown). The association between chilli consumption and over-
weight/obesity in our study may be underestimated, as a high
proportion of chilli consumers were classified as non-consumers
based on the 3-day food intake. The use of cumulative chilli intake
proved the exposure that best defined usual consumption but was
nevertheless based on a relatively small number of days during
the observation period.
In conclusion, chilli intake is inversely associated with incidence
of overweight/obesity among Chinese adults. As the majority of
the population already eat chilli, encouraging more regular and
higher chilli consumption may provide a low cost and simple
International Journal of Obesity (2017) 1074 – 1079
strategy to reduce the incidence of overweight and obesity
in China.
CONFLICT OF INTEREST
The authors declare no conflict of interest.
ACKNOWLEDGEMENTS
This research uses data from China Health and Nutrition Survey (CHNS). We thank the
National Institute for Nutrition and Health, China Center for Disease Control and
Prevention, Carolina Population Center (P2C HD050924, T32 HD007168), the
University of North Carolina at Chapel Hill, the NIH (R01-HD30880, DK056350, R24
HD050924 and R01-HD38700) and the NIH Fogarty International Center (D43
TW009077, D43 TW007709) for financial support, for the CHNS data collection and
analysis files from 1989 to 2015 and future surveys, and the China-Japan Friendship
Hospital, Ministry of Health for support for CHNS 2009, Chinese National Human
Genome Center at Shanghai since 2009 and Beijing Municipal Center for Disease
Prevention and Control since 2011.
AUTHOR CONTRIBUTIONS
ZS contributed to the conception, analysis and interpretation of data; drafting
of the report; and have given approval of the final version for publication. AWT,
MR and AP contributed to the analysis and interpretation of the data,
commented on the report, revising the manuscript and approving the final
version for publication.
REFERENCES
1 Rozin P, Schiller D. The nature and acquisition of a preference for chili pepper
by humans. Motiv Emot 1980; 4: 77–101.
2 Astrup A, Kristensen M, Gregersen NT, Belza A, Lorenzen JK, Due A et al. Can
bioactive foods affect obesity? Ann N Y Acad Sci 2010; 1190: 25–41.
3 Lv J, Qi L, Yu C, Yang L, Guo Y, Chen Y et al. Consumption of spicy foods and total
and cause specific mortality: population based cohort study. BMJ 2015; 351:
h3942.
4 Tremblay A, Arguin H, Panahi S. Capsaicinoids: a spicy solution to the manage-
ment of obesity? Int J Obes 2016; 40: 1198–1204.
5 Fokkens W, Hellings P, Segboer C. Capsaicin for rhinitis. Curr Allergy Asthma Rep
2016; 16: 1–5.
6 Wu H, Wei M, Zhang Q, Du H, Xia Y, Liu L et al. Consumption of chilies, but not
sweet peppers, is positively related to handgrip strength in an adult population.
J Nutr Health Aging 2016; 20: 546–552.
7 Lopez-Carrillo L, Hernandez Avila M, Dubrow R. Chili pepper consumption and
gastric cancer in Mexico: a case-control study. Am J Epidemiol 1994; 139: 263–271.
8 Whiting S, Derbyshire EJ, Tiwari B. Could capsaicinoids help to support weight
management? A systematic review and meta-analysis of energy intake data.
Appetite 2014; 73: 183–188.
9 Janssens PL, Hursel R, Martens EA, Westerterp-Plantenga MS. Acute effects of
capsaicin on energy expenditure and fat oxidation in negative energy balance.
PLoS ONE 2013; 8: e67786.
10 Ludy MJ, Moore GE, Mattes RD. The effects of capsaicin and capsiate on energy
balance: critical review and meta-analyses of studies in humans. Chem Senses
2012; 37: 103–121.
11 Yoshioka M, St-Pierre S, Drapeau V, Dionne I, Doucet E, Suzuki M et al. Effects of
red pepper on appetite and energy intake. Br J Nutr 1999; 82: 115–123.
12 Ludy MJ, Mattes RD. The effects of hedonically acceptable red pepper doses on
thermogenesis and appetite. Physiol Behav 2011; 102: 251–258.
13 Sun D, Lv J, Chen W, Li S, Guo Y, Bian Z et al. Spicy food consumption is associated
with adiposity measures among half a million Chinese people: the China Kadoorie
Biobank study. BMC Public Health 2014; 14: 1293.
14 Li J, Wang R, Xiao C. Association between chilli food habits with iron status and
insulin resistance in a Chinese population. J Med Food 2014; 17: 472–478.
15 Zhai FY, Du SF, Wang ZH, Zhang JG, Du WW, Popkin BM. Dynamics of the Chinese
diet and the role of urbanicity, 1991-2011. Obes Rev 2014; 15(Suppl 1): 16–26.
16 Popkin BM, Du S, Zhai F, Zhang B. Cohort Profile: The China Health and Nutrition
Survey—monitoring and understanding socio-economic and health change in
China, 1989-2011. Int J Epidemiol 2010; 39: 1435–1440.
17 Yao M, McCrory MA, Ma G, Tucker KL, Gao S, Fuss P et al. Relative influence of diet
and physical activity on body composition in urban Chinese adults. Am J Clin Nutr
2003; 77: 1409–1416.
© 2017 Macmillan Publishers Limited, part of Springer Nature.

18 Hu FB, Stampfer MJ, Rimm E, Ascherio A, Rosner BA, Spiegelman D et al. Dietary
fat and coronary heart disease: a comparison of approaches for adjusting for total
energy intake and modeling repeated dietary measurements. Am J Epidemiol
1999; 149: 531–540.
19 Yan S, Li J, Li S, Zhang B, Du S, Gordon-Larsen P et al. The expanding burden of
cardiometabolic risk in China: the China Health and Nutrition Survey. Obes Rev
2012; 13: 810–821.
20 Shi Z, Taylor AW, Riley M, Byles J, Liu J, Noakes M. Association between dietary
patterns, cadmium intake and chronic kidney disease among adults. Clin Nutr
2017; S0261-5614: 31366–31368.
21 Yoshioka M, Doucet E, Drapeau V, Dionne I, Tremblay A. Combined effects of red
pepper and caffeine consumption on 24 h energy balance in subjects given free
access to foods. Br J Nutr 2001; 85: 203–211.
22 Shi Z, Hu X, Yuan B, Hu G, Pan X, Dai Y et al. Vegetable-rich food pattern is related
to obesity in China. Int J Obes (Lond) 2008; 32: 975–984.
23 Westerterp-Plantenga MS, Smeets A, Lejeune MP. Sensory and gastrointestinal
satiety effects of capsaicin on food intake. Int J Obes 2005; 29: 682–688.
24 Smeets AJ, Janssens PL, Westerterp-Plantenga MS. Addition of capsaicin and
exchange of carbohydrate with protein counteract energy intake restriction
effects on fullness and energy expenditure. J Nutr 2013; 143: 442–447.
25 Kentish SJ, Page AJ. Plasticity of gastro-intestinal vagal afferent endings. Physiol
Behav 2014; 136: 170–178.
Supplementary Information accompanies this paper on International Journal of Obesity website (http://www.nature.com/ijo)
© 2017 Macmillan Publishers Limited, part of Springer Nature.
Chilli intake and overweight/obesity among adults
Z Shi et al
1079
26 Kentish SJ, Frisby CL, Kritas S, Li H, Hatzinikolas G, O'Donnell TA et al. TRPV1
channels and gastric vagal afferent signalling in lean and high fat diet induced
obese mice. PLoS ONE 2015; 10: e0135892.
27 Wang X, Miyares RL, Ahern GP. Oleoylethanolamide excites vagal sensory neu-
rones, induces visceral pain and reduces short-term food intake in mice via
capsaicin receptor TRPV1. J Physiol 2005; 564 (Pt 2): 541–547.
28 Vyklicky L, Novakova-Tousova K, Benedikt J, Samad A, Touska F, Vlachova V.
Calcium-dependent desensitization of vanilloid receptor TRPV1: a mechanism
possibly involved in analgesia induced by topical application of capsaicin. Physiol
Res 2008; 57 (Suppl 3): S59–S68.
29 Matsumoto T, Miyawaki C, Ue H, Yuasa T, Miyatsuji A, Moritani T. Effects of
capsaicin-containing yellow curry sauce on sympathetic nervous system activity
and diet-induced thermogenesis in lean and obese young women. J Nutr Sci
Vitaminol (Tokyo) 2000; 46: 309–315.
30 Clegg ME, Golsorkhi M, Henry CJ. Combined medium-chain triglyceride and chilli
feeding increases diet-induced thermogenesis in normal-weight humans. Eur J
Nutr 2013; 52: 1579–1585.
31 Yoshioka M, Lim K, Kikuzato S, Kiyonaga A, Tanaka H, Shindo M et al. Effects of
red-pepper diet on the energy metabolism in men. J Nutr Sci Vitaminol (Tokyo)
1995; 41: 647–656.
32 Cichewicz RH, Thorpe PA. The antimicrobial properties of chile peppers (Capsicum
species) and their uses in Mayan medicine. J Ethnopharmacol 1996; 52: 61–70.
International Journal of Obesity (2017) 1074 – 1079