Combined Impact of Lifestyle-Related Factors on Total
and Cause-Specific Mortality among Chinese Women:
Prospective Cohort Study
Sarah J. Nechuta1, Xiao-Ou Shu1, Hong-Lan Li2, Gong Yang1, Yong-Bing Xiang2, Hui Cai1, Wong-Ho
Chow3, Butian Ji3, Xianglan Zhang1, Wanqing Wen1, Yu-Tang Gao2, Wei Zheng1*
1Division of Epidemiology, Department of Medicine, Vanderbilt University School of Medicine, Nashville, Tennessee, United States of America, 2Department of
Epidemiology, Shanghai Cancer Institute, Shanghai, China, 3Division of Cancer Epidemiology and Genetics, National Cancer Institute, Rockville, Maryland, United States of
Background: Although cigarette smoking, excessive alcohol drinking, obesity, and several other well-studied unhealthy
lifestyle-related factors each have been linked to the risk of multiple chronic diseases and premature death, little is known
about the combined impact on mortality outcomes, in particular among Chinese and other non-Western populations. The
objective of this study was to quantify the overall impact of lifestyle-related factors beyond that of active cigarette smoking
and alcohol consumption on all-cause and cause-specific mortality in Chinese women.
Methods and Findings: We used data from the Shanghai Women’s Health Study, an ongoing population-based prospective
cohort study in China. Participants included 71,243 women aged 40 to 70 years enrolled during 1996–2000 who never
smoked or drank alcohol regularly. A healthy lifestyle score was created on the basis of five lifestyle-related factors shown to
be independently associated with mortality outcomes (normal weight, lower waist-hip ratio, daily exercise, never exposed
to spouse’s smoking, higher daily fruit and vegetable intake). The score ranged from zero (least healthy) to five (most
healthy) points. During an average follow-up of 9 years, 2,860 deaths occurred, including 775 from cardiovascular disease
(CVD) and 1,351 from cancer. Adjusted hazard ratios for mortality decreased progressively with an increasing number of
healthy lifestyle factors. Compared to women with a score of zero, hazard ratios (95% confidence intervals) for women with
four to five factors were 0.57 (0.44–0.74) for total mortality, 0.29 (0.16–0.54) for CVD mortality, and 0.76 (0.54–1.06) for cancer
mortality. The inverse association between the healthy lifestyle score and mortality was seen consistently regardless of
chronic disease status at baseline. The population attributable risks for not having 4–5 healthy lifestyle factors were 33% for
total deaths, 59% for CVD deaths, and 19% for cancer deaths.
Conclusions: In this first study, to our knowledge, to quantify the combined impact of lifestyle-related factors on mortality
outcomes in Chinese women, a healthier lifestyle pattern—including being of normal weight, lower central adiposity,
participation in physical activity, nonexposure to spousal smoking, and higher fruit and vegetable intake—was associated
with reductions in total and cause-specific mortality among lifetime nonsmoking and nondrinking women, supporting the
importance of overall lifestyle modification in disease prevention.
Please see later in the article for the Editors’ Summary.
Citation: Nechuta SJ, Shu X-O, Li H-L, Yang G, Xiang Y-B, et al. (2010) Combined Impact of Lifestyle-Related Factors on Total and Cause-Specific Mortality among
Chinese Women: Prospective Cohort Study. PLoS Med 7(9): e1000339. doi:10.1371/journal.pmed.1000339
Academic Editor: Kay-Tee Khaw, University of Cambridge, United Kingdom
Received March 15, 2010; Accepted August 3, 2010; Published September 14, 2010
Copyright: ? 2010 Nechuta et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: Supported by National Institutes of Health grant R37 CA070867. The funders had no role in study design, data collection and analysis, decision to
publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
Abbreviations: BMI, body mass index; CI, confidence interval; CVD, cardiovascular disease; HR, hazard ratio; MET, metabolic equivalent; PAR, population
attributable risk; SES, socioeconomic; SWHS, Shanghai Women’s Health Study; WHR, waist-hip ratio
* E-mail: firstname.lastname@example.org
PLoS Medicine | www.plosmedicine.org1 September 2010 | Volume 7 | Issue 9 | e1000339
Lifestyle-related factors—such as high adiposity, low or no
exercise participation, unhealthy dietary habits, and environmen-
tal tobacco smoke—each have been linked to an increased risk of
multiple chronic diseases and premature death [1–11]. However,
few studies have investigated the combined impact of these
lifestyle-related factors and mortality outcomes [2,12–15]. Re-
search to quantify the overall impact of lifestyle-related factors on
mortality outcomes will provide important information valuable
for disease prevention. A recent prospective cohort study of 77,782
participants of the Nurse’s Health Study (NHS) found a more than
4-fold increase in risk of all-cause mortality among women aged
34–59 y who reported ever smoking, a body mass index (BMI)
$25 kg/m2, ,30 min per day of physical activity, an unhealthy
diet score, and heavy or no alcohol drinking, compared to women
with none of these risk factors . Another prospective cohort
study among 20,244 British men and women aged 45–79 y
similarly reported a 4-fold increase in risk of all-cause mortality for
participants with no health behaviors compared to participants
who had four health behaviors (nonsmoker, plasma vitamin C
levels indicative of $5 daily servings of fruits and vegetables,
moderate alcohol intake, and physically active) .
Most studies of combinations of established lifestyle factors and
mortality have been conducted in the United States and countries
in Western Europe. Data are limited for other populations,
including Chinese women, whose lifestyles differ considerably
from their European counterparts [16,17]. Further, active smoking
and alcohol consumption, which are two well-studied predictors of
mortality [2,13], have been included in the previous studies.
However, many women, and in particular Asian women , do
not actively smoke or drink regularly, and thus it is important to
study practical disease prevention measures for these women.
Little is known, however, at present about the combined impact of
lifestyle factors beyond that of active smoking and alcohol drinking
In the Shanghai Women’s Health Study, a population-based
cohort study of approximately 75,000 middle-aged and older
Chinese women, less than 3% of cohort members reported ever
smoking and drinking alcohol regularly, providing a unique
opportunity to quantify the overall impact of lifestyle factors other
than active smoking and alcohol consumption on total and cause-
specific mortality. Well-studied lifestyle-related factors relevant for
this population were selected on the basis of prior knowledge of
lifestyle factors in relation to mortality and with consideration of
practical public health recommendations [1,3–9,18–24]. Specifi-
cally, the lifestyle factors selected included: (1) BMI, (2) waist-hip
ratio (WHR), (3) exercise participation, (4) environmental tobacco
smoke (assessed as exposure to spousal smoking), and (5) fruit and
vegetable daily intake.
Participants of this analysis are individuals in the Shanghai
Women’s Health Study (SWHS), an ongoing prospective cohort
study of Chinese women. The study methods and rationale have
been reported in detail elsewhere . Briefly, participants were
recruited from seven urban counties in Shanghai, China. A total of
74,942 women aged 40–70 y were recruited from December 1996
through May 2000 with a participation rate of 92.7%. The
baseline survey included an in-person interview, self-administered
questionnaire, and anthropometric measurements taken by
trained interviewers using standardized protocols. Information
was collected on demographics, lifestyle habits (e.g., diet, physical
activity, alcohol, smoking), menstrual and reproductive history,
medical history, occupational history, and select information from
each participant’s spouse (e.g., disease history, smoking and
alcohol habits). Both the food frequency and physical activity
questionnaires have been validated and reported elsewhere
[25,26]. All participants provided written informed consent, and
human participant Institutional Review Board (IRB) approval was
obtained by the appropriate IRBs in China and the United States.
Follow-up for participants has included in-person interviews
every 2–3 y to collect interim health history. Response rates were
99.8%, 98.7%, and 96.7% for the first, second, and third follow-up
surveys, respectively. Data on vital status and cancer diagnoses
also have been obtained by annual linkage to the population-based
Shanghai cancer and vital statistics registries. Outcome data for
the present analysis were censored at December 31, 2007.
Data from the baseline interview were used to assess the lifestyle
factors of interest. We were interested in lifestyle-related factors that
to mortality. BMI, a measure of general adiposity, was calculated as
measured weightin kilograms divided by measured height in meters
squared and categorized using the World Health Organization
(WHO) classifications : underweight (,18.5 kg/m2), normal
weight (18.5–24.99 kg/m2), overweight (25–29.99 kg/m2), obese
($30 kg/m2). Waist and hip circumference measurements were
used to calculate the WHR (waist divided by hip circumference), a
measure of central adiposity, and classified into three categories
according to tertiles. During the baseline interview, participants
were asked about regular exercise in the past 5 y (‘‘regular’’ was
defined as at least once per week, for more than 3 mo continuously).
Information was also collected on type, intensity, and duration for
up to three activities. We categorized exercise using standard
metabolic equivalents (METs) as MET-hours/day (no exercise
participation, .0 to 1.99 MET-h/d, and $2.0 MET-h/d) [26,28].
One MET-hour/day is approximately equivalent to about 15 min
of participation in moderate-intensity activities [20,29]. Exposure to
environmental tobacco smoke was defined as ever exposed to
spousal smoking or never exposed. Grams per day of fruit and
vegetable intake were assessed via a food frequency questionnaire
for intake over the past 12 mo and categorized into tertiles.
The primary study outcome was deaths from all causes. Cause
of death information was collected from death certificates and
coded according to the International Classification of Diseases, 9th
Revision (ICD-9). Cause-specific deaths examined included deaths
due to cardiovascular disease (CVD) (ICD-9 codes: 390–459) and
cancer (ICD-9 codes: 140–208).
Among 74,942 women who completed the baseline assessment,
2,113 reported ever smoking (2.8%) and 1,678 reported ever
drinking (2.2%); these women were excluded from the analyses
(n=3,513). We also excluded women with missing data on the
lifestyle factors (anthropometric measures ([n=59] and FFQ items
for main foods of interest [n=11]), with extreme daily energy
intake (defined as ,500 or $3,500 kcal per day) (n=108), and
who were lost to follow-up shortly after the baseline recruitment
(n=8). In addition, women who did not have information on
exposure to spousal smoking (n=7,452) were excluded from
analyses of environmental tobacco smoke and the combined effect
of lifestyle factors on mortality. We compared select characteristics
for women included in the current analyses to women in the entire
SWHS cohort (Table S1). With the exception of age, other
Lifestyle-Related Factors and Mortality
PLoS Medicine | www.plosmedicine.org2 September 2010 | Volume 7 | Issue 9 | e1000339
characteristics were comparable across the three groups: (1) the
entire cohort, (2) the cohort after excluding those who met any of
the exclusion criteria stated above except exposure to spousal
smoking, and (3) the cohort after further excluding women with
missing data for spousal smoking. Due to the large sample size, the
tests for several characteristics across these three groups were
statistically significant. The cohorts included in the current analysis
were somewhat younger than the entire cohort, particularly
because of the exclusion of those who had missing data on
exposure to spousal smoking, which was primarily due to a
Two healthy lifestyle scores were created on the basis of previous
research and public health recommendations [5,9,18,20,21,27], as
well as consideration of adequate sample sizes for the five lifestyle
factors. As shown in Table 1, a point was assigned to each category
for the lifestyle factors BMI, WHR, exercise, and daily fruit and
vegetable intake (zero [least healthy] to two [most healthy]), while
for spousal smoking status a binary variable was used (ever, zero
points and never, one point). Healthy lifestyle score 1 was assigned
to each woman by summing the points for the five factors, with a
possible range of 0–13. Healthy lifestyle score 2 (Table 1) was
created by assigning points to simple binary indicators for each of
the five factors with one point for having the healthy factor and
again summing the points for the five factors to assign a score to
each woman (range of 0–5 points). A higher score indicated a
healthier lifestyle, and we hypothesized that mortality would
decrease as number of healthy lifestyle factors increased.
Cox proportional hazards regression models were used to
evaluate the associations of mortality with each lifestyle factor
individually and then the healthy lifestyle scores. Adjusted hazard
ratios (HRs) and their corresponding 95% confidence intervals
(CIs) were derived from Cox models after adjusting for potential
confounders. Age was used as the time-scale for all models ,
with entry time defined as age at baseline interview and exit time
defined as age at death, last follow-up, or December 31, 2007,
whichever came first. We first examined associations for each
lifestyle factor with mortality adjusted for age and socioeconomic
(SES) indicators (occupation [manual and agricultural workers/
unknown, clerical, professional], education [#elementary, junior
high school, high school, .high school], and income/person [low,
#5,000 CNY; middle, 5,000–9999 CNY; high $10,000 CNY]).
Next, we additionally adjusted for the other lifestyle-factors. Both
BMI and WHR remained associated with mortality outcomes after
adjustment for each other and the other lifestyle factors; hence,
both measures were included in the lifestyle scores. Linear trends
were evaluated using the Wald test, treating the lifestyle score as a
continuous variable. We examined the proportional hazards
assumption, both graphically and by testing the significance of
interaction terms for the two lifestyle scores and years of follow-up,
and found no evidence for apparent departure from the
assumption of proportional hazards.
For healthy lifestyle score 2, we calculated the total population
attributable risk (PAR), via summing the exposure-category
specific PARs, which estimates the proportion of deaths associated
with not having the highest score (i.e., four to five healthy lifestyle-
related factors) [31,32]. We used the following formula to estimate
total PARs (percentage), which is appropriate for multicategory
exposures and uses adjusted relative risks :
where pdi=proportion of cases in the ith exposure level;
RRi=relative risks for comparing women with no healthy factors
(i=1),1 healthy factor (i=2),two healthy factors(i=3),or 3 healthy
factors (i=4), to women with four to five healthy lifestyle factors.
PAR estimates are based on the assumption that the observed
associations between the lifestyle factors and mortality are causal
. All analyses were performed using SAS version 9.2. Tests of
statistical significance were based on two-sided probability, and p-
values,0.05 were considered statistically significant.
After an average of 9.1 y of follow-up (648,096 person-years),
2,860 deaths were identified among the 71,243 women who
reported never smoking or drinking alcohol regularly, including
1,351 from cancer and 775 from CVDs. Compared to women
who survived during follow-up, a higher percentage of deceased
participants were classified as underweight, overweight or obese,
had a higher WHR, reported not participating in exercise
regularly, were exposed to spousal smoking, and had a lower
daily intake of fruits and vegetables (Table 2).
Table 3 shows the HRs for each of the five lifestyle factors with
total and cause-specific mortality. In age and SES-adjusted
analyses, compared to obese women, those who were normal or
overweight had significantly decreased HRs for total mortality, but
women who were underweight had a significantly increased HR
(Table 3). The association with underweight was no longer
significant after excluding deaths in the first 3 y (HR=1.19; 95%
CI 0.93–1.52), suggesting an effect of reverse causation due to
weight loss caused by preexisting chronic conditions. HRs for all-
cause mortality were significantly decreased for women who had a
lower WHR, were physically active, never exposed to spousal
smoking, or had higher daily fruit and vegetable intake. Additional
adjustment for all the other lifestyle factors did not appreciably
change these results, although the associations were attenuated for
normal weight, and the HR for spousal smoking status became
marginally significant (p=0.061) as shown in Table 3. Similar
patterns of associations with WHR were observed for cancer and
cardiovascular deaths; findings were less consistent for BMI
(Table 3). The patterns of associations with exercise participation,
spousal smoking, and fruit and vegetable consumption were
comparable to total mortality for deaths from CVD and generally
weak or absent for cancer mortality (Table 3).
We also considered waist circumference as a measure of central
adiposity, however, as compared to WHR, waist circumference
was not as strongly associated with mortality outcomes. Compared
with the highest waist circumference tertile ($81 cm), the HRs for
total mortality for the lowest tertile (,73 cm) and the middle
tertile (73 to ,81 cm) were 0.78 (95% CI 0.68–0.89) and 0.89
(95% CI 0.80–0.99), respectively, adjusting for SES indicators,
BMI, exercise participation, spouse smoking status, and fruit and
vegetable intake. Similar HRs were found for cardiovascular and
cancer mortality, although the HRs for the middle tertile of waist
circumference were not statistically significant (unpublished data).
A higher healthy lifestyle score 1 was significantly associated
with a reduced risk of mortality from all-causes (ptrend,0.01), and
from CVD (ptrend,0.01) and cancer (ptrend=0.022) (Table 4). For
example, women with 7–9 points (most healthy), had a 47%
reduction in risk of all-cause mortality (HR=0.53; 95% CI 0.43–
0.63), compared to women with 0–2 points (least healthy).
Reductions in mortality associated with a higher lifestyle score
were the strongest for deaths due to CVD. Similar patterns were
generally seen for healthy lifestyle score 2 and total and cause-
specific mortality (Table 4); hence, score 2 was used in subsequent
analyses as it is simpler and easier to interpret than score 1. Not
having four to five healthy lifestyle factors was associated with total
Lifestyle-Related Factors and Mortality
PLoS Medicine | www.plosmedicine.org3 September 2010 | Volume 7 | Issue 9 | e1000339
PARs of 33% for total mortality, 59% for CVD mortality, and
19% for cancer mortality (Table 4).
We examined the relation between all-cause mortality and
healthy lifestyle score 2 among three subgroups of women classified
by their chronic disease history at baseline: (1) women with
potentially fatal chronic conditions including cancer, stroke, or
coronary heart disease (CHD) (n=6,009); (2) women with only less
seriousconditions, includinghypertension and diabetes(n=12,209);
and (3) healthy women with no history of above-mentioned
conditions (n=45,573) (Table 5). Results for women in these three
groups were fairly similar to overall findings, with significant trends
for increasing number of healthy lifestyle factors in each subgroup.
We also examined the association of all-cause mortality and healthy
lifestyle score 2 by age (,55 y and $55 y). Results in these two age
groups were similar to overall findings (unpublished data).
Sensitivity analyses were conducted to investigate the potential
for bias due to the existence of subclinical diseases by excluding
deaths occurring in the first 3 y of follow-up. Results from these
analyses were similar to those shown in Table 4 for total mortality
and mortality due to CVD and cancer. HRs for women with four
to five healthy lifestyle factors compared to zero factors were 0.60
(95% CI 0.45–0.80; ptrend,0.01) for total mortality, 0.31 (95% CI
0.16–0.60; ptrend,0.01) for CVD mortality, and 0.75 (95% CI
0.51–1.11; ptrend=0.11) for cancer mortality.
Figure 1 displays cumulative mortality estimates from the Cox
proportional hazards regression model with age as the time-scale
for score 2, adjusting for education, occupation, and income. The
cumulative mortality for the healthy lifestyle score 2 by age at
study exit was lowest for women with four to five healthy lifestyle
factors and highest for women with zero factors (Figure 1).
In this population-based prospective cohort study of Chinese
women aged 40–70 y, we found that healthier lifestyle-related
factors—including normal weight, lower WHR, participation in
exercise, never being exposed to spousal smoking, and higher daily
fruit and vegetable intake—were significantly and independently
associated with lower risk of total and cause-specific mortality.
Healthy lifestyle scores, composite measures of these five factors,
were significantly associated with decreasing mortality as a
number of healthy factors increased. The associations persisted
for all women regardless of their baseline comorbidities. To our
knowledge, this is the first large prospective cohort study
specifically designed to quantify the combined impact of lifestyle-
related factors on mortality outcomes among lifetime nonsmokers
and nonalcohol drinkers. Results show that lifestyle factors other
than active smoking and alcohol drinking have a major combined
impact on mortality on a scale comparable to the effect of smoking
as the leading cause of death in most populations [11,13,14].
In general, the literature is limited in regard to the study of
combinations of lifestyle factors and mortality [2,12–14,33–38].
Further, most such studies have included alcohol and/or smoking
[2,12,14,33–37], and little is known about the combined impact of
lifestyle factors other than active smoking and drinking in relation
to mortality. The answer to this question is of particular
importance as there are a substantial number of people worldwide
who are nonsmokers and do not drink excessively [14,39]. In an
attempt to address this question, in a subgroup analysis among
never-smokers in the Nurse’s Health Study, van Dam and
colleagues reported a 2-fold excess risk of all-cause mortality
among women who had a high BMI, low physical activity, and
unhealthy diet . That study, however, did not consider
environmental tobacco smoke or measures of central adiposity
such as WHR.
Another limitation of previously published studies is that most
studies have been conducted in the United States or Western
Europe, and few studies have examined the combined impact of
lifestyle factors in relation to mortality among Asian populations.
We did, however, identify three reports from Japan, two
Table 1. Combined healthy lifestyle scores in the Shanghai Women’s Health Study.
Lifestyle Factors Assessed at Baseline ClassificationScoring Classification
Lifestyle Score 1 Lifestyle Score 2
18.5–24.99, normal weight21
WHRTertile 3, $0.83000
Tertile 2, 0.786 to ,0.83010
Tertile 1, ,0.78621
Exercise participation (MET h/d) No activitya
.0 to 1.99b
Spouse smokeEver exposed to spouse’s smoking00
Fruit and vegetable daily intake (g) Tertile 1, ,404.3 g/d00
Tertile 2, 404.3 to ,626.5 g/d10
Tertile 3, $626.5 g/d21
aNo exercise participation.
b,,30 min of moderate-intensity activity per day.
c,$30 min of moderate-intensity activity per day.
Lifestyle-Related Factors and Mortality
PLoS Medicine | www.plosmedicine.org4 September 2010 | Volume 7 | Issue 9 | e1000339
conducted in rural northern Japan [34,37] and one among
individuals of the Japan Collaborative Cohort Study . Each
of these reports demonstrated that healthier lifestyles based on
several lifestyle-related factors were associated with substantial
reductions in death among Asian men and women. None of
these reports, however, focused on evaluating the impact of
To our knowledge, this is the first investigation of combinations
of lifestyle factors and risk of mortality among Chinese women.
We selected five factors that are easy to assess and interpret, based
both on prior knowledge of lifestyle factors in relation to mortality
and public health recommendations [1,4–9,18–24]. BMI, exercise
participation, and fruit and vegetable intake have been well-
studied in relation to mortality [1,5–9,24]. WHR and environ-
among nonsmokers and
mental tobacco smoke have not been studied as much, but
evidence is accumulating for these two factors as important
predictors of total mortality [3,4,23,40,41], and both were shown
to be associated with mortality among SWHS participants [18,21].
Several large prospective cohort studies among women have
shown WHR to be an important predictor of mortality
independent of BMI [3,4,18,41], and in some populations,
WHR may be an even stronger predictor of mortality [4,41].
Hence, on the basis of previous studies that both BMI and WHR
may be independent measures of adiposity among women and our
findings for independent effects of BMI and WHR after
adjustment for each other and additional potential confounders,
we included both BMI and WHR in the lifestyle scores. In
addition, environmental tobacco smoke is a particularly important
exposure for women living in China and other Asian countries
Table 2. Age-adjusted baseline characteristics by survival status in the Shanghai Women’s Health Study (n=71,243).
CharacteristicsPercent Survived (n=68,383) Percent Deceased (n=2,860)p-Valuea
Age at baseline (y)
40–49 50.7 14.6
Junior high school 37.341.3
High school 28.422.9
.High school14.1 9.5
Manual and agricultural workers/unknown50.056.6
0.786 to ,0.83033.4 30.7
Exercise participation (MET, h/d)
.0 to ,1.99 24.323.5
Fruit and vegetable intake tertiles (g/d)
Among women who never smoked cigarettes or drank alcohol regularly. All values (except age) were directly standardized to the age distribution of the cohort.
ap-Value from chi-square test for general association.
bExcluded from the analysis were women without information on exposure to spousal smoking (n=7,452).
Lifestyle-Related Factors and Mortality
PLoS Medicine | www.plosmedicine.org5 September 2010 | Volume 7 | Issue 9 | e1000339
Table 3. Adjusted HRs for lifestyle-related factors and risk of all-cause, cardiovascular, and cancer mortality among nonsmoking and nondrinking women aged 40–70 y at baseline
(n=71,243), Shanghai Women’s Health Study, 1996–2007.
All-Cause (n=2,860 deaths)
HR (95% CI)
HR (95% CI)
HR (95% CI)
HR (95% CI)
HR (95% CI)
HR (95% CI)
0.786 to ,0.830
Exercise participation (MET, h/d)
Fruit and vegetable intake (g/d)
404.3 to ,626.5
aHRs are estimated from Cox proportional hazards regression models using age as the time-scale and adjusted for education, occupation, and income.
bAdditionally adjusted for other lifestyle factors in the table.
cExcludes women without information on exposure to spousal smoking (n=7,452).
Lifestyle-Related Factors and Mortality
PLoS Medicine | www.plosmedicine.org6September 2010 | Volume 7 | Issue 9 | e1000339
given the high smoking prevalence among Asian men [21,42]. No
previous study included either WHR or environmental tobacco
smoke in the assessment of the combined impact of lifestyle factors
This study has several strengths, including a population-based
prospective cohort study design and large overall sample size.
Selection bias was minimized due to the exceptionally high
response rates at recruitment (92.7%) and in the follow-up surveys
(96.7%–99.8%). Baseline assessments were conducted by trained
interviewers using standardized protocols, and anthropometric
data were based on measurements instead of self-report.
Limitations of this study should be considered for interpretation
of results. One concern is the potential for information or reverse
causation bias due to the presence of subclinical disease or
prevalent clinical disease. To address this concern, we analyzed
the association of mortality with the lifestyle score among women
without prevalent CVD, cancer, stroke, diabetes, or hypertension
and also after excluding deaths in the first 3 y of follow-up.
Findings for these subgroups were not appreciably different from
the overall results. Women without information on exposure to
spousal smoking were excluded from the lifestyle score and
mortality analyses. Exclusion of these women, however, is unlikely
Table 4. Healthy lifestyle scores and risk of all-cause, cardiovascular, and cancer mortality among nonsmoking and nondrinking
women aged 40–70 y at baseline (n=63,791), Shanghai Women’s Health Study, 1996–2007.
Lifestyle Score PercentAll-Cause (n=2,302 deaths)CVD (n=605)Cancer (n=1,113)
n DeathsHR(95% CI)n Deaths HR(95% CI)n Deaths HR(95% CI)
0–2 13.2507 1.00 (Reference)163 1.00 (Reference)204 1.00 (Reference)
316.6465 0.84(0.74–0.95)1250.72(0.57–0.92) 2050.89 (0.73–1.08)
4 22.44970.78 (0.69–0.89) 1320.70 (0.56–0.89)2410.88(0.73–1.06)
522.1 4150.70 (0.62–0.80) 1050.61 (0.48–0.78) 2140.82(0.68–1.00)
6 15.8 2700.66(0.57–0.77) 550.47(0.34–0.64) 1540.85 (0.68–1.05)
7–910.1 1480.53 (0.43–0.63) 250.31(0.20–0.48) 95 0.76 (0.59–0.97)
0 10.83631.00 (Reference)1231.00 (Reference)1531.00(Reference)
1 29.9801 0.93 (0.82–1.05)2190.77(0.62–0.96) 3761.00(0.83–1.20)
2 34.6 7230.84(0.74–0.95)180 0.66 (0.52–0.83)353 0.90(0.74–1.09)
3 19.53430.75 (0.64–0.87)72 0.51(0.38–0.68) 1860.87 (0.70–1.09)
4–5 5.2720.57 (0.44–0.74)11 0.29(0.16–0.54)450.76 (0.54–1.06)
All HRs are estimated from Cox proportional hazards regression models with age as the time-scale and are adjusted for education, occupation, and income. Range for
score 1, 0–13 possible points; range for score 2, 0–5 possible points.
aEstimated by summing exposure-category specific PARs from score 0 to 3 using the group with score 4–5 as the reference.
Table 5. Healthy lifestyle score two and risk of all-cause mortality among nonsmoking and nondrinking women aged 40–70 y by
chronic disease status at baseline, Shanghai Women’s Health Study, 1996–2007.
Women with Potentially Fatal
Diseases at Baselinea(n=6,009)
Women with Diabetes and Hyperten-
sion only at Baseline (n=12,209)Healthy Women at Baselineb(n=45,573)
n DeathsHR(95% CI) n Deaths HR(95% CI)n DeathsHR (95% CI)
0 1131.00(Reference)123 1.00(Reference)127 1.00 (Reference)
1 2411.02(0.81–1.28) 2501.02 (0.82–1.27)3100.87 (0.71–1.07)
2 1860.86 (0.67–1.08)1860.92 (0.73–1.16)3510.87 (0.71–1.07)
3 960.82 (0.62–1.09) 690.77 (0.57–1.04)1780.79 (0.62–0.99)
4–5 170.53 (0.32–0.89) 160.69(0.41–1.16) 390.61(0.43–0.88)
All HRs are estimated from Cox proportional hazards regression models with age as the time-scale and are adjusted for education, occupation, and income.
aPotentially fatal chronic diseases included cancer, stroke, and coronary heart disease.
bNo history of cancer, stroke, coronary heart disease, diabetes, or hypertension.
Lifestyle-Related Factors and Mortality
PLoS Medicine | www.plosmedicine.org7September 2010 | Volume 7 | Issue 9 | e1000339
to materially affect our findings, though the sample size was
reduced slightly. Measurement error, particularly for self-reported
data on diet and exercise, is another potential concern. However,
we have previously shown good validity and reliability for diet and
physical activity data from the SWHS [25,26]. Furthermore,
nondifferential errors tend to attenuate the observed associations,
and thus the true association between lifestyle factors and mortality
may be stronger than that estimated in this study. We did not
adjust for potential mediators such as blood lipid levels and
hypertension in the analysis since the primary purpose of the study
was to quantify the overall impact of lifestyle on mortality
outcomes. Adjustment for mediators in the causal pathway
between lifestyle factors and mortality would affect the quantifi-
cation of the overall impact of these lifestyle factors on mortality
For ease of interpretation, healthy lifestyle scores were created
in the analysis assuming an equal weight for each of the factors
included. A weighted approach based on the effect size of each
variable could improve the estimate of the overall impact of
lifestyle factors on mortality. However, as demonstrated in our
study, the estimates using score 1 (semi-weighted) and score 2
(nonweighted) are similar, suggesting that a weighted approach
may not improve the estimates substantially. Despite an overall
large sample size, the sample sizes for some cause-specific analyses
were relatively small, which may affect the precision of the point
estimates. In addition, the observed associations between lifestyle
factors and mortality outcomes in our study may be underesti-
mated because of the use of baseline covariate measurements only
. Extended follow-up of this cohort will provide the
opportunity to further evaluate the impact of these lifestyle-related
factors on mortality outcomes in the future.
Most of the lifestyle-related factors studied here may be
improved by individual motivation to change unhealthy behaviors.
For example, changes in physical activity levels and energy
Figure 1. Mortality and healthy lifestyle score 2, adjusted for education, occupation, and income, Shanghai Women’s Health Study
Lifestyle-Related Factors and Mortality
PLoS Medicine | www.plosmedicine.org8 September 2010 | Volume 7 | Issue 9 | e1000339
expenditure to reduce adiposity can be made by increasing activity
levels through walking daily or participating in group exercise
classes. Increased fruit and vegetable intake is fairly straightfor-
ward for the majority of Chinese women in urban communities,
given that many varieties of fruit and vegetables are readily
available at the markets. However, both the physical and social
environments also are important contributors to sustained lifestyle
changes, and may be more significant than individual motivation
for some lifestyle factors, which is particularly true for exposure to
spousal smoking. Change in exposure to spousal smoking may
start with increased awareness by both women and their husbands
about the detrimental health effects of smoking, but also will
require community-based interventions and change in the social
environment (e.g., promotion of home smoking bans in commu-
In conclusion, in this first study to quantify the combined impact
of lifestyle-related factors on mortality outcomes among Chinese
women, we found that a higher healthy lifestyle score based on five
factors was associated with substantial reductions in total and
cause-specific mortality among lifetime nonsmoking and non-
drinking women. Reductions in premature deaths associated with
higher healthy lifestyle scores were seen among women with and
without preexisting comorbidities. Our study suggests that a
combined healthy lifestyle—including being of normal weight,
lower central adiposity, participation in physical activity, non-
exposure to spousal smoking, and higher fruit and vegetable
intake—can result in lower mortality among middle-aged and
older women, including women with a history of severe disease.
Research is needed to design appropriate interventions to increase
these healthy lifestyle factors among Asian women.
and after exclusions, Shanghai Women’s Health Study.
Found at: doi:10.1371/journal.pmed.1000339.s001 (0.08 MB
Baseline characteristics for study participants before
We thank the participants and the research staff of the Shanghai Women’s
Health Study for making this study possible.
ICMJE criteria for authorship read and met: SJN XOS HLL GY YBX HC
WHC BJ XZ WW YTG WZ. Agree with the manuscript’s results and
conclusions: SJN XOS HLL GY YBX HC WHC BJ XZ WW YTG WZ.
Designed the experiments/the study: XOS YBX HC BJ XZ YTG WZ.
Analyzed the data: SJN WW WZ. Collected data/did experiments for the
study: HLL GY YBX YTG WZ. Enrolled patients: GY YBX BJ YTG.
Wrote the first draft of the paper: SJN. Contributed to the writing of the
paper: SJN XOS GY HC XZ YTG WZ. Directed the overall study
operation: XOS. Contributed to study design and data collection: WHC.
Contributed to data interpretation and manuscript preparation: WHC.
Principal investigator of the Shanghai Women’s Health Study: WZ.
1. Whitlock G, Lewington S, Sherliker P, Clarke R, Emberson J, et al. (2009) Body-
mass index and cause-specific mortality in 900 000 adults: collaborative analyses
of 57 prospective studies. Lancet 373: 1083–1096.
2. Knoops KT, de Groot LC, Kromhout D, Perrin AE, Moreiras-Varela O,
et al. (2004) Mediterranean diet, lifestyle factors, and 10-year mortality in
elderly European men and women: the HALE project. JAMA 292:
3. Pischon T, Boeing H, Hoffmann K, Bergmann M, Schulze MB, et al. (2008)
General and abdominal adiposity and risk of death in Europe. N Engl J Med
4. Folsom AR, Kushi LH, Anderson KE, Mink PJ, Olson JE, et al. (2000)
Associations of general and abdominal obesity with multiple health outcomes in
older women: the Iowa Women’s Health Study. Arch Intern Med 160:
5. Oguma Y, Sesso HD, Paffenbarger RS, Jr., Lee IM (2002) Physical activity and
all cause mortality in women: a review of the evidence. Br J Sports Med 36:
6. Lollgen H, Bockenhoff A, Knapp G (2009) Physical activity and all-cause
Med 30: 213–224.
7. Steffen LM, Jacobs DR, Jr., Stevens J, Shahar E, Carithers T, et al. (2003)
Associations of whole-grain, refined-grain, and fruit and vegetable consumption
with risks of all-cause mortality and incident coronary artery disease and
ischemic stroke: the Atherosclerosis Risk in Communities (ARIC) Study.
Am J Clin Nutr 78: 383–390.
8. Genkinger JM, Platz EA, Hoffman SC, Comstock GW, Helzlsouer KJ (2004)
Fruit, vegetable, and antioxidant intake and all-cause, cancer, and cardiovas-
cular disease mortality in a community-dwelling population in Washington
County, Maryland. Am J Epidemiol 160: 1223–1233.
9. World Health Organization/Food and Agricultural Organization (2003) Expert
report on diet, nutrition, and the prevention of chronic diseases, technical report.
Series 916. Geneva: World Health Organization.
10. International Agency for Research on Cancer (2004) Tobacco smoking and
involuntary smoking. Lyon: IARC (IARC monographs on the evaluation of
carcinogenic risks to humans Vol. 83).
11. Danaei G, Ding EL, Mozaffarian D, Taylor B, Rehm J, et al. (2009) The
preventable causes of death in the United States: comparative risk assessment of
dietary, lifestyle, and metabolic risk factors. PLoS Med 6: e1000058.
12. Tamakoshi A, Tamakoshi K, Lin Y, Yagyu K, Kikuchi S (2009) Healthy lifestyle
and preventable death: findings from the Japan Collaborative Cohort (JACC)
Study. Prev Med 48: 486–492.
13. van Dam RM, Li T, Spiegelman D, Franco OH, Hu FB (2008) Combined
impact of lifestyle factors on mortality: prospective cohort study in US women.
BMJ DOI: 10.1136/bmj.a1440.
14. Khaw KT, Wareham N, Bingham S, Welch A, Luben R, et al. (2008) Combined
impact of health behaviours and mortality in men and women: the EPIC-
Norfolk prospective population study. PLoS Med 5: e12. doi:10.1371/
15. Kvaavik E, Batty GD, Ursin G, Huxley R, Gale CR (2010) Influence of
individual and combined health behaviors on total and cause-specific mortality
in men and women: the United Kingdom health and lifestyle survey. Arch
Intern Med 170: 711–718.
16. Kim S, Popkin BM, Siega-Riz AM, Haines PS, Arab L (2004) A cross-national
comparison of lifestyle between China and the United States, using a
comprehensive cross-national measurement tool of the healthfulness of lifestyles:
the Lifestyle Index. Prev Med 38: 160–171.
17. Zheng W, Chow WH, Yang G, Jin F, Rothman N, et al. (2005) The Shanghai
Women’s Health Study: rationale, study design, and baseline characteristics.
Am J Epidemiol 162: 1123–1131.
18. Zhang X, Shu XO, Yang G, Li H, Cai H, et al. (2007) Abdominal adiposity and
mortality in Chinese women. Arch Intern Med 167: 886–892.
19. Zhang X, Shu XO, Chow WH, Yang G, Li H, et al. (2008) Body mass index at
various ages and mortality in Chinese women: impact of potential methodo-
logical biases. Int J Obes (Lond) 32: 1130–1136.
20. Matthews CE, Jurj AL, Shu XO, Li HL, Yang G, et al. (2007) Influence of
exercise, walking, cycling, and overall nonexercise physical activity on mortality
in Chinese women. Am J Epidemiol 165: 1343–1350.
21. Wen W, Shu XO, Gao YT, Yang G, Li Q, et al. (2006) Environmental tobacco
smoke and mortality in Chinese women who have never smoked: prospective
cohort study. BMJ 333: 376–380.
22. Cai H, Shu XO, Gao YT, Li H, Yang G, et al. (2007) A prospective study of
dietary patterns and mortality in Chinese women. Epidemiology 18: 393–401.
23. McGhee SM, Ho SY, Schooling M, Ho LM, Thomas GN, et al. (2005)
Mortality associated with passive smoking in Hong Kong. BMJ 330: 287–288.
24. Key TJ, Thorogood M, Appleby PN, Burr ML (1996) Dietary habits and
mortality in 11,000 vegetarians and health conscious people: results of a 17 year
follow up. BMJ 313: 775–779.
25. Shu XO, Yang G, Jin F, Liu D, Kushi L, et al. (2004) Validity and
reproducibility of the food frequency questionnaire used in the Shanghai
Women’s Health Study. Eur J Clin Nutr 58: 17–23.
26. Matthews CE, Shu XO, Yang G, Jin F, Ainsworth BE, et al. (2003)
Reproducibility and validity of the Shanghai Women’s Health Study physical
activity questionnaire. Am J Epidemiol 158: 1114–1122.
27. WHO Expert Consultation (2004) Appropriate body-mass index for Asian
populations and its implications for policy and intervention strategies. Lancet
28. Ainsworth BE, Haskell WL, Whitt MC, Irwin ML, Swartz AM, et al. (2000)
Compendium of physical activities: an update of activity codes and MET
intensities. Med Sci Sports Exerc 32: S498–504.
Lifestyle-Related Factors and Mortality
PLoS Medicine | www.plosmedicine.org9September 2010 | Volume 7 | Issue 9 | e1000339
29. Ainsworth BE, Haskell WL, Leon AS, Jacobs DR, Jr., Montoye HJ, et al. (1993)
Compendium of physical activities: classification of energy costs of human
physical activities. Med Sci Sports Exerc 25: 71–80.
30. Korn EL, Graubard BI, Midthune D (1997) Time-to-event analysis of
longitudinal follow-up of a survey: choice of the time-scale. Am J Epidemiol
31. Rockhill B, Newman B, Weinberg C (1998) Use and misuse of population
attributable fractions. Am J Public Health 88: 15–19.
32. Wacholder S, Benichou J, Heineman EF, Hartge P, Hoover RN (1994)
Attributable risk: advantages of a broad definition of exposure. Am J Epidemiol
33. Spencer CA, Jamrozik K, Norman PE, Lawrence-Brown M (2005) A simple
lifestyle score predicts survival in healthy elderly men. Prev Med 40: 712–717.
34. Tsubono Y, Koizumi Y, Nakaya N, Fujita K, Takahashi H, et al. (2004) Health
practices and mortality in Japan: combined effects of smoking, drinking, walking
and body mass index in the Miyagi Cohort Study. J Epidemiol 14 (Suppl 1):
35. Haveman-Nies A, de Groot LP, Burema J, Cruz JA, Osler M, et al. (2002)
Dietary quality and lifestyle factors in relation to 10-year mortality in older
Europeans: the SENECA study. Am J Epidemiol 156: 962–968.
36. Meng L, Maskarinec G, Lee J, Kolonel LN (1999) Lifestyle factors and chronic
diseases: application of a composite risk index. Prev Med 29: 296–304.
37. Tsubono Y, Fukao A, Hisamichi S (1993) Health practices and mortality in a
rural Japanese population. Tohoku J Exp Med 171: 339–348.
38. Breslow L, Enstrom JE (1980) Persistence of health habits and their relationship
to mortality. Prev Med 9: 469–483.
39. (2009) Cigarette smoking among adults and trends in smoking cessation - United
States, 2008. MMWR Morb Mortal Wkly Rep 58: 1227–1232.
40. Humble C, Croft J, Gerber A, Casper M, Hames CG, et al. (1990) Passive
smoking and 20-year cardiovascular disease mortality among nonsmoking wives,
Evans County, Georgia. Am J Public Health 80: 599–601.
41. Lahmann PH, Lissner L, Gullberg B, Berglund G (2002) A prospective study of
adiposity and all-cause mortality: the Malmo Diet and Cancer Study. Obes Res
42. Kubo M, Hata J, Doi Y, Tanizaki Y, Iida M, et al. (2008) Secular trends in the
incidence of and risk factors for ischemic stroke and its subtypes in Japanese
population. Circulation 118: 2672–2678.
43. Emberson JR, Whincup PH, Morris RW, Wannamethee SG, Shaper AG (2005)
Lifestyle and cardiovascular disease in middle-aged British men: the effect of
adjusting for within-person variation. Eur Heart J 26: 1774–1782.
44. Ji M, Ding D, Hovell MF, Xia X, Zheng P, et al. (2009) Home smoking bans in
an urbanizing community in China. Am J Prev Med 37: 132–136.
Lifestyle-Related Factors and Mortality
PLoS Medicine | www.plosmedicine.org 10 September 2010 | Volume 7 | Issue 9 | e1000339
Background. It is well established that lifestyle-related
factors, such as limited physical activity, unhealthy diets,
excessive alcohol consumption, and exposure to tobacco
smoke are linked to an increased risk of many chronic
diseases and premature death. However, few studies have
investigated the combined impact of lifestyle-related factors
and mortality outcomes, and most of such studies of
combinations of established lifestyle factors and mortality
have been conducted in the US and Western Europe. In
addition, little is currently known about the combined
impact on mortality of lifestyle factors beyond that of active
smoking and alcohol consumption.
Why Was This Study Done? Lifestyles in regions of the
world can vary considerably. For example, many women in
Asia do not actively smoke or regularly drink alcohol, which
are important facts to note when considering practical
disease prevention measures for these women. Therefore, it
is important to study the combination of lifestyle factors
appropriate to this population.
What Did the Researchers Do and Find? The researchers
used the Shanghai Women’s Health Study, an ongoing
prospective cohort study of almost 75,000 Chinese women
aged 40–70 years, as the basis for their analysis. The
baseline data on anthropometric measurements, lifestyle
habits (including the responses to validated food frequency
occupational history, and select information from each
participant’s spouse, such as smoking history and alcohol
consumption. This information was used by the researchers
to create a healthy lifestyle score on the basis of five lifestyle-
related factors shown to be independently associated with
mortality outcomes in this population: normal weight, lower
waist-hip ratio, daily exercise, never being exposed to
spouse’s smoking, and higher daily fruit and vegetable
intake. The score ranged from zero (least healthy) to five
(most healthy) points. The researchers found that higher
healthy lifestyle scores were significantly associated with
decreasing mortality and that this association persisted for
all women regardless of their baseline comorbidities. So in
effect, healthier lifestyle-related factors, including normal
weight, lower waist-hip ratio, participation in exercise, never
being exposed to spousal smoking, and higher daily fruit and
vegetable intake, were significantly and independently
associated with lower risk of total, and cause-specific,
What Do These Findings Mean? This large prospective
cohort study conducted among lifetime nonsmokers and
nonalcohol drinkers shows that lifestyle factors, other than
active smoking and alcohol consumption, have a major
combined impact on total mortality on a scale comparable
to the effect of smoking—the leading cause of death in most
populations. However, the sample sizes for some cause-
specific analyses were relatively small (despite the overall
large sample size), and extended follow-up of this cohort will
provide the opportunity to further evaluate the impact of
these lifestyle-related factors on mortality outcomes in the
The findings of this study highlight the importance of overall
lifestyle modification in disease prevention, especially as
most of the lifestyle-related factors studied here may be
improved by individual motivation to change unhealthy
behaviors. Further research is needed to design appropriate
interventions to increase these healthy lifestyle factors
among Asian women.
Additional Information. Please access these Web sites via
the online version of this summary at http://dx.doi.org/10.
N The Vanderbilt Epidemiology Center has more information on
the Shanghai Women’s Health Study
N The World Health Organization provides information on
health in China
N The document Health policy and systems research in China
contains information about health policy and health
systems research in China
N TheChineseMinistryofHealth alsoprovideshealth information
Lifestyle-Related Factors and Mortality
PLoS Medicine | www.plosmedicine.org11 September 2010 | Volume 7 | Issue 9 | e1000339