Articles
Cardiovascular mortality, all-cause mortality, and diabetes
Lancet Diabetes Endocrinol
2014; 2: 474–80
Published Online
April 3, 2014
http://dx.doi.org/10.1016/
S2213-8587(14)70057-9
See Comment page 441
Department of Endocrinology,
China-Japan Friendship
Hospital, Beijing, China
(Prof G Li MD, W Yang MD,
B Zhang MD,
Y Shuai BD, J Hong MD);
Center of Endocrinology
and Cardiovascular Disease,
National Center of Cardiology
& Fuwai Hospital, Beijing,
China (G Li, Y An MD,
Q Gong MD); Division of
Diabetes Translation
(P Zhang PhD, E W Gregg PhD),
Center for Global Health
(M M Engelgau MD), Centers for
Disease Control and
Prevention, Atlanta, GA, USA;
Department of Cardiology,
Da Qing First Hospital, Da Qing,
China (J Wang MD, H Li MD,
Y Hu MD); Department of
Management of
Noncommunicable Diseases,
World Health Organization,
Geneva, Switzerland
(G Roglic MD); and Phoenix
Epidemiology and Clinical
Research Branch, National
Institute of Diabetes and
Digestive and Kidney Diseases,
Phoenix, AZ, USA
(P H Bennett FRCP)
Correspondence to:
Prof Guangwei Li, Department
of Endocrinology, 167 Beilishi
Road, Xicheng District, Beijing,
100037, China
guangwei_li@medmail.com.cn
Or
Dr Ping Zhang, Division of
Diabetes Translation, Centers for
Disease Control and Prevention,
4770 Buford HWY, NE, F-75
Atlanta, GA 30341, USA
pzhang@cdc.gov
474
Downloaded for Anonymous User (n/a) at University of Queensland Library from ClinicalKey.com.au by Elsevier on September 24, 2017.
incidence after lifestyle intervention for people with
impaired glucose tolerance in the Da Qing Diabetes
Prevention Study: a 23-year follow-up study
Guangwei Li, Ping Zhang, Jinping Wang, Yali An, Qiuhong Gong, Edward W Gregg, Wenying Yang, Bo Zhang, Ying Shuai, Jing Hong,
Michael M Engelgau, Hui Li, Gojka Roglic, Yinghua Hu, Peter H Bennett
Summary
Background Lifestyle interventions among people with impaired glucose tolerance reduce the incidence of diabetes,
but their effect on all-cause and cardiovascular disease mortality is unclear. We assessed the long-term effect of
lifestyle intervention on long-term outcomes among adults with impaired glucose tolerance who participated in the
Da Qing Diabetes Prevention Study.
Methods The study was a cluster randomised trial in which 33 clinics in Da Qing, China—serving 577 adults with
impaired glucose tolerance—were randomised (1:1:1:1) to a control group or lifestyle intervention groups (diet or
exercise or both). Patients were enrolled in 1986 and the intervention phase lasted for 6 years. In 2009, we followed up
participants to assess the primary outcomes of cardiovascular mortality, all-cause mortality, and incidence of diabetes
in the intention-to-treat population.
Findings Of the 577 patients, 439 were assigned to the intervention group and 138 were assigned to the control group
(one refused baseline examination). 542 (94%) of 576 participants had complete data for mortality and 568 (99%)
contributed data to the analysis. 174 participants died during the 23 years of follow-up (121 in the intervention group vs 53
in the control group). Cumulative incidence of cardiovascular disease mortality was 11·9% (95% CI 8·8–15·0) in the
intervention group versus 19·6% (12·9–26·3) in the control group (hazard ratio [HR] 0·59, 95% CI 0·36–0·96; p=0·033).
All-cause mortality was 28·1% (95% CI 23·9–32·4) versus 38·4% (30·3–46·5; HR 0·71, 95% CI 0·51–0·99; p=0·049).
Incidence of diabetes was 72·6% (68·4–76·8) versus 89·9% (84·9–94·9; HR 0·55, 95% CI 0·40–0·76; p=0·001).
Interpretation A 6-year lifestyle intervention programme for Chinese people with impaired glucose tolerance can
reduce incidence of cardiovascular and all-cause mortality and diabetes. These findings emphasise the long-term
clinical benefits of lifestyle intervention for patients with impaired glucose tolerance and provide further justification
for adoption of lifestyle interventions as public health measures to control the consequences of diabetes.
Funding Centers for Disease Control and Prevention, WHO, the China-Japan Friendship Hospital, Da Qing First Hospital.
Introduction implications. We assessed such effects in people who
Lifestyle interventions for people with impaired glucose participated in the Da Qing Diabetes Prevention Study
tolerance can delay or prevent the development of over a 23-year period.
diabetes and lead to improvements for other
cardiovascular risk factors.1–11 However, whether these Methods
changes reduce the incidence of long-term complications, Study design and participants
and the excess cardiovascular disease and all-cause The design of the Da Qing Diabetes Prevention Study and
mortality that accompany diabetes is uncertain.12,13 In our the 20-year follow-up study have been reported pre-
previous analysis7 of 20-year follow-up data from the viously.1,7,14 In 1986, 33 primary care clinics in Da Qing,
Da Qing Diabetes Prevention Study, we reported a China serving 577 people with impaired glucose tolerance
statistically insignificant 17% reduction in cardiovascular were enrolled into a cluster randomised trial. All patients
disease mortality and a significantly lower incidence of treated at the clinics were eligible if they had impaired
severe diabetic retinopathy that seems to have been glucose tolerance in 1985. We did a 20-year follow-up
mediated primarily by delaying the onset of diabetes.14 study up to the end of 20067 and here report results after
Because cardiovascular disease is the major cause of an additional 3 years. The institutional review boards of
excess mortality in people with impaired glucose WHO and the China-Japan Friendship Hospital approved
tolerance, more definitive information about the effect of the protocol. All study participants, or their proxies who
lifestyle intervention on cardiovascular disease and all- provided information about deceased participants, gave
cause mortality in such people has crucial public health written informed consent.
www.thelancet.com/diabetes-endocrinology Vol 2 June 2014
For personal use only. No other uses without permission. Copyright ©2017. Elsevier Inc. All rights reserved.
 Articles

Randomisation and masking any diabetes-related macrovascular or microvascular

Clinics were cluster randomised (1:1:1:1) to one of three
lifestyle interventions (25 clinics with a total of
439 people) or a control group (eight clinics with
138 people). Of the 25 intervention clinics, nine
(148 participants) were assigned to a diet only
intervention, nine clinics (155 participants) to an
exercise only intervention, and seven clinics
(136 participants) to a diet plus exercise intervention.
The diet intervention was designed to produce weight
loss in those who were overweight or obese, and to
reduce simple carbohydrate and alcohol intake in people
of normal weight. The exercise intervention was
designed to increase the leisure time spent doing
physical activity.
All participants (including those at clinics in the control
group) received examinations at baseline, 2 years, 4 years,
and 6 years after randomisation and at 20-year and
23-year follow-up. Participants in the control group
received standard medical care. The interventions were
delivered for 6 years (1986–92), after which all participants
were informed of the results and asked to continue with
usual medical care.
Investigators who assessed the outcomes at follow-up
were masked to treatment allocation. Patients and other
investigators were not masked.
complications. This report includes data for only the first
three because no new data for diabetes-related
macrovascular and microvascular complications were
collected at the 23-year follow-up.
Statistical analysis
Power calculations were done for the original 6-year
intervention trial. For the present study we estimated
minimal detectable differences. With an α of 0·05, we
estimated that there was an 80% chance of detecting a
43% reduction in all-cause mortality and a 63% reduction
in cardiovascular disease mortality when comparing the
control group with the combined lifestyle intervention
groups.
The observational period for mortality was from date of
randomisation to date of death, loss to follow-up, or Dec 31,
2009, whichever came first. For diabetes incidence, the
observational period was from the date of randomisation
to the date of diagnosis, or Dec 31, 2009, whichever came
first. As the incidence of diabetes at both the end of the
6-year intervention period and the 20-year follow-up period
was similar in each of the three intervention groups, we
combined the intervention groups a priori to increase the
577 patients enrolled in randomly assigned clinic

Procedures 1 refused baseline examination

We tried to follow up all the original study participants to
establish their vital status. For deceased participants, we
collected date and cause of death from death certificates, 438 in intervention group 138 in control group
reviews of medical records, and interviews with proxy
informants. We asked proxy informants about the date, 31 lost to 3 lost to
place, and circumstances of death along with information follow-up follow-up
about hospitals or physicians from whom the participant
had received care around the time of death. We obtained 407 assessed 135 assessed
medical records and death certificates and, together with
the informant interviews, they were reviewed and Figure 1: Trial profile
adjudicated independently by two doctors (JW and YA) to
establish the underlying cause of death. Disagreements
were resolved by a third senior physician (HL or YH). Intervention Control
group group
Causes of death were classified into two categories: (n=430) (n=138)
cardiovascular disease death (defined as death attributed
Sex
to coronary heart disease, stroke, or sudden death), and
Men 230 (53%) 79 (57%)
non-cardiovascular disease death (death from all other
Women 200 (47%) 59 (43%)
causes).
Age (years) 44·7 (9·3) 46·6 (9·3)
Diabetes was defined by 1985 WHO criteria with the
BMI (kg/m2) 25·7 (3·8) 26·2 (3·8)
results of oral glucose tolerance tests done every 2 years
Systolic blood pressure (mm Hg) 132·4 (23·5) 134·4 (23·4)
between 1986 and 1992, and at the 20-year and 23-year
Diastolic blood pressure (mm Hg) 87·3 (14·5) 88·5 (13·5)
follow-up examinations,15 or from a report of physician-
diagnosed diabetes with evidence in the medical record FPG (mmol/L) 5·6 (0·8) 5·5 (0·8)

of high glucose concentrations, or use of hypoglycaemic 2hPG (mmol/L) 9·0 (0·9) 9·0 (16·1)

drugs, as described previously.7 Current smoker (%) 165 (38%) 69 (50%)

Outcomes
The four primary outcomes were all-cause mortality,
cardiovascular disease mortality, diabetes incidence, and
Data are n (%) or mean (SD). FPG=fasting venous plasma glucose. 2hPG=venous
plasma glucose concentration 2 h after intake of 75 g oral glucose.
Table 1: Baseline characteristics

www.thelancet.com/diabetes-endocrinology Vol 2 June 2014 475

Downloaded for Anonymous User (n/a) at University of Queensland Library from ClinicalKey.com.au by Elsevier on September 24, 2017.
For personal use only. No other uses without permission. Copyright ©2017. Elsevier Inc. All rights reserved.
 Articles
statistical power to detect a mortality difference attributable
to the intervention. Mortality and diabetes incidence were
calculated as the number of events divided by total person-
years since randomisation.
We assessed cumulative all-cause and cardiovascular
mortality with the Kaplan-Meier method. We fitted a
Weibull proportional hazards model16 with the SAS
NLMIXED procedure for the primary analysis. The time
to all-cause and cardiovascular disease death or date of
diagnosis of diabetes was treated as the dependent
variable. Treatment group was included as a fixed effect
and clinic was included as a random effect. No other
covariates were adjusted for because the original study
was a cluster randomised trial. We estimated hazard
ratios (HRs) and 95% CIs from the NLMIXED model.
Differences were considered statistically significant if
p<0·05 in two-sided tests. All primary analyses were for
the intention-to-treat population.
Mortality of people with and without diabetes varies
greatly by sex and age.17–19 Therefore, we did post-hoc
secondary analyses to assess the effect of intervention on
the primary outcomes of all-cause mortality and
cardiovascular disease mortality separately in women
and men. Because of an imbalance in the age distribution
and smoking rates between men and women in the
intervention and control groups, we adjusted for these
differences—as well as for the effect of delay in onset of
diabetes—in post-hoc multivariable analyses with the
SAS NLMIXED procedure. We did the statistical analyses
with SAS (version 9.1).
Role of the funding source
Employees of the study sponsors were involved in the
study design; collection, analysis, interpretation of the
data; and the writing of the report. The corresponding
authors had full access to all data in the study and had
See Online for appendix the final responsibility for the decision to submit for
publication.
Intervention group
(n=430)
All-cause mortality
Deaths 121 (28%)
Deaths per 1000 person-years (95% CI) 14·3 (11·8–16·9)
Cumulative incidence (%; 95% CI) 28·1% (23·9–32·4)
Cardiovascular disease mortality
Deaths 51 (12%)
Deaths per 1000 person-years (95% CI) 6·0 (4·4–7·7)
Cumulative incidence (%; 95% CI) 11·9% (8·8–15·0)
Diabetes incidence
Cases of diabetes 312 (73%)
Cases per 1000 person-years (95% CI) 73·2 (65·1–81·3)
Cumulative incidence (%; 95% CI) 72·6% (68·4–76·8)
Data are n (%) unless stated otherwise. Hazard ratios adjusted by clinic.
Table 2: Effect of lifestyle intervention on all-cause mortality, cardiovascular disease mortality, and diabetes (1986–2009)
476
Downloaded for Anonymous User (n/a) at University of Queensland Library from ClinicalKey.com.au by Elsevier on September 24, 2017.
For personal use only. No other uses without permission. Copyright ©2017. Elsevier Inc. All rights reserved.
Results
Of the original 577 study participants, one declined the
baseline examination, eight had only a baseline
examination, and 20 participants were lost to follow-up
during the 6-year intervention period, mainly because of
job relocation. During the subsequent 17 years after the
intervention, only six participants were lost to follow-up.
Thus, 542 participants (94%) had complete data for
mortality and 568 (99%) contributed data to the analysis
(figure 1). 563 participants (98%) had complete data for
diabetes incidence. Table 1 shows the baseline
characteristics.
174 participants died during the 23 years of follow-up
(121 in the intervention group and 53 in the control
group). The cumulative incidence of all-cause death was
28·1% (95% CI 23·9–32·4) in the intervention group
and 38·4% (95% CI 30·3–46·5) in the control group,
with an HR (adjusted for cluster randomisation) of 0·71
(95% CI 0·51–0·99; p=0·049; table 2, figure 2). Few
patients died during the first 10 years of the study, but
the rates rose progressively thereafter (table 2, figure 2).
78 participants died as a result of cardiovascular disease;
51 in the intervention group versus 27 in the control
group. Cumulative incidences were 11·9% (95% CI
8·8–15·0) versus 19·6% (95% CI 12·9–26·3; HR 0·59,
95% CI 0·36–0·96; p=0·03; table 2, figure 2). Significant
differences between groups in the incidence of
diabetes—present at the end of the 6-year intervention
period and at 20-year follow-up1,7—persisted, with a
cumulative incidence of 72·6% (95% CI 68·4–76·8) in
the intervention group and 89·9% (95% CI 84·9–94·9)
in the control group (HR 0·55, 95% CI 0·40–0·76;
p=0·001; table 2, figure 2).
Secondary post-hoc analyses stratified by sex and age
showed substantial differences between women and
men for the effect of intervention on all-cause and
cardiovascular disease mortality (table 3, appendix).
Among women, the cumulative incidence of all-cause
Control group Hazard ratio p value
(n=138) (95% CI)
53 (38%) ·· ··
19·9 (14·5–25·2) ·· ··
38·4% (30·3–46·5) 0·71 (0·51–0·99) 0·049
27 (20%) ·· ··
10·1 (6·3–14·0) ·· ··
19·6% (12·9–26·3) 0·59 (0·36–0·96) 0·033
124 (90%) ·· ··
122·9 (101·3–144·5) ·· ··
89·9% (84·9–94·9) 0·55 (0·40–0·76) 0·001
www.thelancet.com/diabetes-endocrinology Vol 2 June 2014
 Articles

mortality in the control group was almost double that in

the intervention group (table 3) and cardiovascular A
45 Control group
disease mortality showed a similar pattern (table 3). Intervention group
40
Conversely, for men, the intervention had no significant

Proportion of participants (%)

effect on either all-cause mortality or cardiovascular 35
disease mortality (table 3), although the reduction in 30
diabetes incidence was much the same in men and 25
women (table 3).
20
A greater proportion of men than women were smokers
(61% vs 16%) and mean age in the control group was 15

higher for men than for women (appendix). A higher 10

percentage of women were smokers in the control 5
HR 0·71, 95% CI (0·51–0·99)
compared with the intervention group (appendix). We 0
did a post-hoc multivariable analysis to test whether such 0 2 4 6 8 10 12 14 16 18 20 22 23
differences in baseline characteristics could account for Number at risk
the overall reduction of mortality and differences in the Control group 138 137 135 130 127 118 113 108 103 99 95 84 82
Intervention 438 426 412 395 384 373 364 355 338 321 305 290 286
effectiveness of the intervention between the sexes group
(appendix). However, after adjustment for age, sex,
smoking and the interaction between sex and B
25
intervention , the effect of lifestyle intervention remained
significant for both all-cause and cardiovascular disease
Proportion of participants (%)

20
mortality, suggesting that differences in baseline
characteristics did not account for the differences.
15
Because lifestyle intervention delays the onset of
diabetes and reduces the incidence of diabetes, we
10
assessed how mortality was affected by the time free
from diabetes, which was defined as the time between
randomisation and onset of diabetes (appendix). 5

Increased delay in the onset of diabetes was associated HR 0·59, 95% CI (0·36–0·96)
with significantly lower all-cause and cardiovascular 0
0 2 4 6 8 10 12 14 16 18 20 22 23
disease mortality. After inclusion of time to onset of
diabetes in the multivariable models the intervention Number at risk
Control group 138 137 135 130 127 118 113 108 103 99 95 84 82
variable was no longer statistically significant, suggesting Intervention 438 426 412 395 384 373 364 355 338 321 305 290 286
that the reduction in mortality associated with the group
intervention is mediated by its effect in delaying the C
onset of diabetes. 100

90
Discussion 80
This study is the first randomised clinical trial to show
Proportion of participants (%)

70
that lifestyle intervention in people with impaired
glucose tolerance reduces all-cause and cardiovascular 60
disease mortality (panel). These findings seem to mainly 50
be a result of lower mortality among women in the
40
intervention group. The Malmo study13 showed lower
mortality over a 12-year period in men with impaired 30
glucose tolerance treated with lifestyle intervention 20
compared with a group who received routine treatment,
10
but the study was not randomised. Other lifestyle HR 0·55, 95% CI (0·40–0·76)
studies—such as the Finnish Diabetes Prevention Study2 0
0 2 4 6 8 10 12 14 16 18 20 22 23
and the Diabetes Prevention Program3—consistently
showed that a lifestyle intervention could reduce the Follow-up (years)
Number at risk
incidence of diabetes and decrease other cardiovascular Control group 138 105 69 48 40 37 34 27 27 23 14 13 13
disease risk factors8,10,11,13 but have not reported a reduction Intervention 438 387 314 250 230 206 192 161 147 136 114 114 114
group
in the incidence of all-cause and cardiovascular disease
mortality. Moreover, in the Look AHEAD study,23 a Figure 2: Cumulative incidence of (A) all-cause mortality, (B) cardiovascular disease mortality, and (C)
lifestyle intervention designed to produce weight loss in diabetes incidence over the 23-year follow-up (1986–2009)
obese or overweight people with diabetes did not HR=hazard ratio.

www.thelancet.com/diabetes-endocrinology Vol 2 June 2014 477

Downloaded for Anonymous User (n/a) at University of Queensland Library from ClinicalKey.com.au by Elsevier on September 24, 2017.
For personal use only. No other uses without permission. Copyright ©2017. Elsevier Inc. All rights reserved.
 Articles

Women Men
Control group Intervention group Hazard ratio p value Control group Intervention group Hazard ratio p value
(n=59) (n=200) (95% CI)* (n=79) (n=230) (95% CI)*
All-cause mortality
Number of deaths 17 30 ·· ·· 36 91 ·· ··
Deaths per 1000 person-years (95% CI) 14·1 (7·4–20·8) 7·4 (4·7–10·0) ·· ·· 24·7 (16·6–32·7) 20·8 (16·6–25·1) ·· ··
Cumulative incidence (%; 95% CI) 28·8% (17·2–40·4) 15·0% (10·1–19·9) 0·46 (0·24–0·87) 0·02 45·6% (34·6–56·6) 39·6% (33·3–45·9) 0·97 (0·65–1·46) 0·9
Cardiovascular disease mortality
Number of deaths 10 12 ·· ·· 17 39 ·· ··
Deaths per 1000 person-years (95% CI) 8·3 (3·2–13·4) 3·0 (1·3–4·6) ·· ·· 11·6 (6·1–17·2) 8·9 (6·1–11·7) ·· ··
Cumulative incidence (%; 95% CI) 17·0 (7·4–26·6) 6·0 (2·7–9·3) 0·28 (0·11–0·71) 0·01 21·5 (12·5–30·5) 17·0 (12·2–21·8) 0·91 (0·50–1·65) 0·7
Diabetes incidence
Number with diabetes 55 148 ·· ·· 69 164 ·· ··
Cases per 1000 person-years (95% CI) 128·5 (94·5–162·5) 74·9 (62·8–86·9) ·· ·· 118·8 (90·7–146·8) 71·8 (60·8–82·8) ·· ··
Cumulative incidence (%; 95% CI) 93·2 (86·7–99·7) 74·0 (67·9–80·1) 0·55 (0·35–0·87) 0·006 87·3 (80·0–94·6) 71·3 (65·4–77·2) 0·56 (0·39–0·81) 0·004

*Adjusted for randomisation by clinic and age.

Table 3: Effect of lifestyle intervention on all-cause mortality, cardiovascular disease mortality, and diabetes in women and men (1986–2009)

study, a difference in cardiovascular disease mortality

Panel: Research in context between the intervention and control group started to
emerge 12 years after the study began, slowly increased
Systematic review
to a 17% difference by the 20-year follow-up,7 but became
Three systematic reviews20–22 of randomised controlled clinical trials have assessed the
statistically significant only after 23 years (figure 2).
effects of lifestyle intervention on prevention of type 2 diabetes for people with impaired
The reduction in mortality mainly occurred in women;
glucose tolerance. They show that dietary or physical activity interventions reduce the
the intervention seemed to have little effect in men,
incidence of diabetes, not only during the period of active intervention,20 but also for
despite similar reductions in the incidence of diabetes.
many years afterwards: for 13 years or more in the Finnish Diabetes Prevention study,11 at
The prevalence of smoking at baseline was much higher
least 10 years in the US Diabetes Prevention Program,8 and 20 years in the Da Qing
for men than for women, and epidemiological analysis of
Diabetes Prevention Study.7 Lifestyle interventions for people with impaired glucose
the whole study population showed that smoking
tolerance are also associated with amelioration of some cardiovascular risk factors such as
increased the risk of all-cause death by almost 50% (data
blood pressure and concentrations of HDL and LDL cholesterol, and triglycerides.10 The risk
not shown). Nevertheless, adjustment for imbalances in
of death for people with type 2 diabetes is about twice that of people of a similar age who
smoking and age did not account for the apparent sex
do not have diabetes. A key question is whether or not lifestyle intervention to prevent or
differences in mortality, nor did this adjustment eliminate
delay the onset of diabetes can lower this excess mortality risk in people with impaired
the overall beneficial effect of the intervention. Poorer
glucose tolerance. However, to date, no randomised clinical trials have reported a
compliance with lifestyle intervention by men than by
significant reduction for either cardiovascular disease mortality or all-cause mortality.21,22
women, or differences in unmeasured confounders,
Interpretation might also have contributed to more favourable outcomes
Our study shows that a 6-year period of lifestyle intervention in Chinese people with for women. We have been unable to establish any
impaired glucose tolerance reduced the incidence of diabetes over a protracted period, definitive reasons for differences in the effect of the
and was ultimately associated with a significant reduction in total and cardiovascular intervention on mortality between men and women.
disease mortality. This reduction in mortality seems to be partly mediated by a delay in Our study has several strengths. First, individuals with
onset of diabetes. These findings provide yet further justification to implement lifestyle impaired glucose tolerance were recruited by screening a
interventions for people with impaired glucose tolerance as clinical and public health well-defined population and most people who were
measures to control the long-term consequences of diabetes. screened were enrolled in the trial.1 Consequently, the
findings are probably generalisable to most people in
significantly reduce cardiovascular disease events or China with impaired glucose tolerance. Second,
cardiovascular disease mortality. A key difference randomisation by clinic rather than by the individual
between these studies and the Da Qing Diabetes minimises the likelihood of contamination between the
Prevention Study is the length of follow-up; in previous intervention and control groups. Our primary analyses
studies, the length of follow-up might have been were adjusted to control for the cluster randomisation.
insufficient to detect an effect of intervention on Third, follow-up for mortality is very high (94%), reducing
mortality. Although the association between duration of non-response bias. Of the 35 participants who were lost to
diabetes and mortality is well established, serious chronic follow-up, 29 were lost during the initial active
complications and excess mortality typically only occur intervention phase of the study mainly because of
after at least 10 years of having diabetes.24–28 In the present relocation for work and migration from Da Qing. Fourth,

478 www.thelancet.com/diabetes-endocrinology Vol 2 June 2014

Downloaded for Anonymous User (n/a) at University of Queensland Library from ClinicalKey.com.au by Elsevier on September 24, 2017.
For personal use only. No other uses without permission. Copyright ©2017. Elsevier Inc. All rights reserved.

all participants received usual care from the primary care
clinics in a single health-care system in Da Qing, which
was provided by the state-run oil company. This reduces
the likelihood that variations in medical care could have
biased the results. Fifth, diabetes incidence, mortality,
and causes of death were assessed systematically.
Diabetes was assessed by oral glucose tolerance test every
2 years during the 6-year intervention period and at the
20-year and 23-year follow-up examinations, and then
systematically measured, albeit by conventional clinical
methods, for the remainder of the follow-up period.
Finally, and perhaps most importantly, the follow-up
period lasted for 23 years, providing enough time for the
development of the chronic complications of diabetes and
a sufficient number of deaths to enable us to detect
differences in mortality between the control and
intervention groups.
The study has some important limitations. First,
because of the small size of the original trial, we combined
the three intervention groups to provide sufficient power.
Second, different methods of follow-up were used during
the 6 years of active intervention compared with the
subsequent years. Nevertheless, the baseline and follow-
up assessments were systematic and identical in the
control and intervention groups. Third, not all deaths
were documented by death certificate. However, death
was ascertained from hospital records for more than 90%
of those who died. This method has been used
successfully in other studies29,30 of mortality in low-income
and middle-income countries. Fourth, the study was
originally designed and powered to establish whether a
lifestyle intervention could reduce the incidence of
diabetes over a 6-year period. Only later, after this
hypothesis had been proven, was the follow-up phase
designed. Consequently, we lack systematic information
about changes in behaviour and cardiovascular risk
factors such as smoking habits, blood pressure, serum
cholesterol concentration, and drug use beyond the 6-year
active intervention phase. This shortcoming limits our
ability to investigate possible explanatory variables.
However, the primary analyses based on the initial cluster
randomisation of participants are robust and only used
information for mortality outcomes that can be obtained
reliably in retrospect from death certificates, medical
records, and informant interviews.
This study is the first to show that lifestyle intervention
in people with impaired glucose tolerance can both
reduce the incidence of diabetes and decrease mortality,
and so it has important implications for public health
policy about diabetes prevention in China and worldwide.
In China, as elsewhere, the health and economic burdens
of diabetes are large and growing. In 2008, an estimated
92 million adults in China (9·7% of the adult population)
had diabetes and 148 million (15·5%) had prediabetes.31
A more recent study reports an even larger estimate of
113·9 million (11·6% of adults) with diabetes in 2010.32
Worldwide, diabetes was estimated to be responsible for
www.thelancet.com/diabetes-endocrinology Vol 2 June 2014
Downloaded for Anonymous User (n/a) at University of Queensland Library from ClinicalKey.com.au by Elsevier on September 24, 2017.
For personal use only. No other uses without permission. Copyright ©2017. Elsevier Inc. All rights reserved.
Articles
5·1 million excess deaths and at least US$548 billion in
health-care expenditure in 2013.33 Our findings imply
that implementation of lifestyle intervention for people
with impaired glucose tolerance—especially women—
who are at high risk for diabetes in China, or elsewhere,
would reduce the incidence of diabetes and associated
health-care expenditure, eventually resulting in a lower
number of diabetes-related deaths.
Contributors
GL coordinated and designed the study, acquired funding, collected data,
did the statistical analysis, and wrote the report. PZ coordinated and
designed the study, acquired funding, did the statistical analysis, and
wrote the report. JW and QG designed the study, collected data, and did
the statistical analysis. YA collected data and did the statistical analysis.
EWG and MME designed the study, acquired funding, did the statistical
analysis, and wrote the report. WY designed the study. BZ, YS, JH, and
HL collected data. GR designed the study and acquired funding. YH
designed the study and collected data. PHB designed the study, did the
statistical analysis, and wrote the report.
Declaration of interests
We declare that we have no competing interests.
Acknowledgments
Supported by CDC/WHO Cooperative Agreement No. U58/
CCU424123-01-02 and the China-Japan Friendship Hospital. We thank
the participants in the original Da Qing Diabetes Prevention Study and
their proxies who contributed to the follow-up study. We also thank
Lingzhi Kong, China Ministry of Health, the leadership at the
China-Japan Friendship Hospital, Da Qing First Hospital, the Da Qing
City Health Bureau, and the Beijing and West Pacific Regional Office of
WHO for their general support. We thank Xilin Yang (Tianjin Medical
University), Yang Yang (National Center of Cardiology and Fuwai
Hospital, China), and Ted Thompson (Centers for Disease Control and
Prevention, USA) for statistical advice. Our special thanks go to the late
Prof Xiaoren Pan as this study would not have been possible without
his leadership in the design and implementation of the original
Da Qing Diabetes Prevention Study. The contents of this report are
solely the responsibility of the authors and do not necessarily represent
the official positions of the Centers for Disease Control and Prevention,
the National Institute of Diabetes and Digestive and Kidney Diseases,
or WHO.
References
1 Pan XR, Li GW, Hu YH, et al. Effects of diet and exercise in
preventing NIDDM in people with impaired glucose tolerance. The
Da Qing IGT and Diabetes Study. Diabetes Care 1997; 20: 537–44.
2 Tuomilehto J, Lindstrom J, Eriksson JG, et al. Prevention of type 2
diabetes mellitus by changes in lifestyle among subjects with
impaired glucose tolerance. N Engl J Med 2001; 344: 1343–50.
3 Knowler WC, Barrett-Connor E, Fowler SE, et al. Reduction in the
incidence of type 2 diabetes with lifestyle intervention or
metformin. N Engl J Med 2002; 346: 393–403.
4 Ratner R, Goldberg R, Haffner S, et al. Impact of intensive lifestyle
and metformin therapy on cardiovascular disease risk factors in the
diabetes prevention program. Diabetes Care 2005; 28: 888–94.
5 Kosaka K, Noda M, Kuzuya T. Prevention of type 2 diabetes by
lifestyle intervention: a Japanese trial in IGT males.
Diabetes Res Clin Pract 2005; 67: 152–62.
6 Ramachandran A, Snehalatha C, Mary S, Mukesh B, Bhaskar AD,
Vijay V. The Indian Diabetes Prevention Programme shows that
lifestyle modification and metformin prevent type 2 diabetes in
Asian Indian subjects with impaired glucose tolerance (IDPP-1).
Diabetologia 2006; 49: 289–97.
7 Li G, Zhang P, Wang J, et al. The long-term effect of lifestyle
interventions to prevent diabetes in the China Da Qing Diabetes
Prevention Study: a 20-year follow-up study. Lancet 2008;
371: 1783–89.
8 Knowler WC, Fowler SE, Hamman RF, et al. 10-year follow-up of
diabetes incidence and weight loss in the Diabetes Prevention
Program Outcomes Study. Lancet 2009; 374: 1677–86.
479
 Articles
9 Saito T, Watanabe M, Nishida J, et al. Lifestyle modification and
prevention of type 2 diabetes in overweight Japanese with impaired
fasting glucose levels: a randomized controlled trial.
Arch Intern Med 2011; 171: 1352–60.
10 Orchard TJ, Temprosa M, Barrett-Connor E, et al. Long-term effects
of the Diabetes Prevention Program interventions on cardiovascular
risk factors: a report from the DPP Outcomes Study. Diabet Med
2013; 30: 46–55.
11 Lindstrom J, Peltonen M, Eriksson JG, et al. Improved lifestyle and
decreased diabetes risk over 13 years: long-term follow-up of the
randomised Finnish Diabetes Prevention Study (DPS). Diabetologia
2013; 56: 284–93.
12 Uusitupa M, Peltonen M, Lindstrom J, et al. Ten-year mortality and
cardiovascular morbidity in the Finnish Diabetes Prevention
Study--secondary analysis of the randomized trial. PLoS One 2009;
4: e5656.
13 Eriksson KF, Lindgarde F. No excess 12-year mortality in men
with impaired glucose tolerance who participated in the Malmo
Preventive Trial with diet and exercise. Diabetologia 1998;
41: 1010–16.
14 Gong Q, Gregg EW, Wang J, et al. Long-term effects of a
randomised trial of a 6-year lifestyle intervention in impaired
glucose tolerance on diabetes-related microvascular complications:
the China Da Qing Diabetes Prevention Outcome Study.
Diabetologia 2011; 54: 300–07.
15 WHO Study Group. Diabetes Mellitus. Report of a WHO Study
Group. 727, 1–113; Geneva: World Health Organization, 1985.
16 Collett D. Modeling survival data in medical research. London:
Chapman & Hall, 1994.
17 Barnett KN, Ogston SA, McMurdo ME, Morris AD, Evans JM.
A 12-year follow-up study of all-cause and cardiovascular mortality
among 10,532 people newly diagnosed with Type 2 diabetes in
Tayside, Scotland. Diabet Med 2010; 27: 1124–29.
18 Seshasai SR, Kaptoge S, Thompson A, et al. Diabetes mellitus,
fasting glucose, and risk of cause-specific death. N Engl J Med 2011;
364: 829–41.
19 Sievers ML, Nelson RG, Knowler WC, Bennett PH. Impact of
NIDDM on mortality and causes of death in Pima Indians.
Diabetes Care 1992; 15: 1541–49.
480
Downloaded for Anonymous User (n/a) at University of Queensland Library from ClinicalKey.com.au by Elsevier on September 24, 2017.
For personal use only. No other uses without permission. Copyright ©2017. Elsevier Inc. All rights reserved.
20 Yamaoka K, Tango T. Efficacy of lifestyle education to prevent type 2
diabetes: a meta-analysis of randomized controlled trials.
Diabetes Care 2005; 28: 2780–86.
21 Hopper I, Billah B, Skiba M, Krum H. Prevention of diabetes and
reduction in major cardiovascular events in studies of subjects with
prediabetes: meta-analysis of randomised controlled clinical trials.
Eur J Cardiovasc Prev Rehabil 2011; 18: 813–23.
22 Yoon U, Kwok LL, Magkidis A. Efficacy of lifestyle interventions
in reducing diabetes incidence in patients with impaired glucose
tolerance: a systematic review of randomized controlled trials.
Metabolism 2013; 62: 303–14.
23 Wing RR, Reboussin D, Lewis CE. Intensive lifestyle intervention
in type 2 diabetes. N Engl J Med 2013; 369: 2358–59.
24 Knowler WC, Sartor G, Melander A, Schersten B. Glucose tolerance
and mortality, including a substudy of tolbutamide treatment.
Diabetologia 1997; 40: 680–86.
25 Kim NH, Pavkov ME, Looker HC, et al. Plasma glucose regulation
and mortality in pima Indians. Diabetes Care 2008; 31: 488–92.
26 Brun E, Nelson RG, Bennett PH, et al. Diabetes duration and
cause-specific mortality in the Verona Diabetes Study. Diabetes Care
2000; 23: 1119–23.
27 Holman RR, Paul SK, Bethel MA, Matthews DR, Neil HA. 10-year
follow-up of intensive glucose control in type 2 diabetes.
N Engl J Med 2008; 359: 1577–89.
28 Hu FB, Stampfer MJ, Solomon CG, et al. The impact of diabetes
mellitus on mortality from all causes and coronary heart disease in
women: 20 years of follow-up. Arch Intern Med 2001; 161: 1717–23.
29 Gajalakshmi V, Peto R. Commentary: verbal autopsy procedure for
adult deaths. Int J Epidemiol 2006; 35: 748–50.
30 Yang G, Rao C, Ma J, et al. Validation of verbal autopsy procedures
for adult deaths in China. Int J Epidemiol 2006; 35: 741–48.
31 Yang W, Lu J, Weng J, et al. Prevalence of diabetes among men and
women in China. N Engl J Med 2010; 362: 1090–101.
32 Xu Y, Wang L, He J, et al. Prevalence and control of diabetes in
Chinese adults. JAMA 2013; 310: 948–59.
33 International Diabetes Federation. Diabetes Atlas, 5th edn.
Brussels: International Diabetes Federation, 2011.
www.thelancet.com/diabetes-endocrinology Vol 2 June 2014