Aging Clinical and Experimental Research (2019) 31:575–593

https://doi.org/10.1007/s40520-018-1012-z

REVIEW

Effectiveness of high-intensity interval training on glycemic control

and cardiorespiratory fitness in patients with type 2 diabetes:
a systematic review and meta-analysis
Jing‑xin Liu1 · Lin Zhu1,2 · Pei‑jun Li1 · Ning Li1 · Yan‑bing Xu3

Received: 4 May 2018 / Accepted: 14 July 2018 / Published online: 30 July 2018
© The Author(s) 2018

Abstract
Aims The aim of this systematic review and meta-analysis was to quantify the effect of high-intensity interval training (HIIT)
on glycemic control and cardiorespiratory fitness compared with moderate-intensity training (MICT) and no training at all
in patients with type 2 diabetes (T2D).
Methods Relevant articles were sourced from PubMed, Embase, the Web of Science, EBSCO, and the Cochrane Library.
Randomized-controlled trials were included based upon the following criteria: participants were clinically diagnosed with
T2D, outcomes that included glycemic control (e.g., hemoglobin A1c); body composition (e.g., body weight); cardiorespira-
tory fitness (e.g., ­VO2peak) are measured at baseline and post-intervention and compared with either a MICT or control group.
Results Thirteen trials involving 345 patients were finally identified. HIIT elicited a significant reduction in BMI, body
fat, HbA1c, fasting insulin, and V ­ O2peak in patients with type 2 diabetes. Regarding changes in the body composition of
patients, HIIT showed a great improvement in body weight (mean difference: − 1.22 kg, 95% confidence interval [CI] − 2.23
to − 0.18, P = 0.02) and body mass index (mean difference: − 0.40 kg/m2, 95% CI − 0.78 to − 0.02, P = 0.04) than MICT
did. Similar results were also found with respect to HbA1c (mean difference: − 0.37, 95% CI − 0.55 to − 0.19, P < 0.0001);
relative ­VO2peak (mean difference: 3.37 ml/kg/min, 95% CI 1.88 to 4.87, P < 0.0001); absolute ­VO2peak (mean difference:
0.37 L/min, 95% CI 0.28 to 0.45, P < 0.00001).
Conclusions HIIT may induce more positive effects in cardiopulmonary fitness than MICT in T2D patients.

Keywords High-intensity interval training · Glycemic control · Cardiorespiratory fitness · Type 2 diabetes

Introduction in 2014 [1], with a predicted prevalence of 552 million by

2030 [2]. Because of the growing economic and social bur-
Type 2 diabetes (T2D) is a metabolic disease characterized dens associated with T2D treatment, effective and accessi-
by hyperglycemia resulting from a resistance to insulin ble lifestyle interventions for people with T2D have never
or a relative insulin insufficiency that can induce cardio- been more important. Exercise intervention is recognized
vascular disease and lead to cardiovascular deterioration. as an integral concept for lifestyle intervention in T2D
According to epidemiological survey results, more than patients [3, 4], and it has been recommended by both the
422 million people worldwide were living with diabetes American Diabetes Association and the American College
of Sports Medicine that patients should perform at least
150 min/week of moderate-to-vigorous aerobic exercise
* Lin Zhu [5, 6]. Abundant evidence from randomized-controlled
40848567@qq.com
trials (RCTs) shows the benefits of aerobic exercise in
1
School of Kinesiology, Shanghai University of Sport, glycemic control; for example, it reduces fasting glucose
Shanghai 200438, China and improves insulin sensitivity, both of which help to
2
Guangzhou Sport University, Guangzhou 510500, China alleviate the development of diabetes complications and
3 mortality [7–10]. Furthermore, a recent meta-analysis
Department of Child Health Care and Rehabilitation,
Lanzhou University Second Hospital, Lanzhou 730030, demonstrated that aerobic exercise training is associated
China with a decrease in HbA1c, insulin resistance, and fasting

13
Vol.:(0123456789)
576 Aging Clinical and Experimental Research (2019) 31:575–593

glucose, and suggested that high-intensity aerobic exer- Patients and methods
cise is superior to lower intensity exercise in improving
cardiorespiratory fitness in T2D patients [11]. However, Search strategy
the majority of patients do not typically achieve the rec-
ommended level of physical activity, despite the fact that The databases which we searched included PubMed,
increases in physical activity level can improve glycemic the Web of Science, EBSCO, Embase, and the Cochrane
control and cardiorespiratory fitness in T2D patients. In Library. All of the databases were searched from their date
addition, a lack of time has been identified as one of the of inception until April 2018. We included only studies writ-
key barriers preventing patients from performing sufficient ten in English. We used combined key phrases and Medical
physical activity, which means that patients must partici- Subject Heading (MeSH) terms as follows: “type 2 diabetes
pate in more time-efficient training programs to achieve mellitus,” “diabetes mellitus, type II,” “type 2 diabetes,”
optimized outcomes. “T2D,” “T2DM,” “high-intensity interval training,” “high-
High-intensity interval training (HIIT), therefore, appears intensity aerobic interval exercise,” “high-intensity interval
to be a feasible and time-efficient alternative exercise pro- training,” “aerobic interval training,” “high-intensity inter-
tocol to aerobic exercise: it involves alternating, repetitive mittent exercise,” “HIT,” and “HIIT.” Supporting informa-
short bouts of high-intensity exercise interspersed with less tion appendix in S1 gives a detailed description of the search
active or passive recovery periods. Numerous recent studies strategy. In addition, the reference lists of included studies
have shown HIIT to be superior in improving health ben- and reviews were also examined for additional potentially
efits compared with lower intensity aerobic exercise in a eligible studies.
variety of populations [12–14]. Støa et al. [15] found that
people with T2D who performed a supervised HIIT pro- Inclusion and exclusion criteria
gram at an intensity of 85–95% of their maximal heart rate
with 52% ­VO2peak interval experienced a significant increase Type of study
in ­VO2peak and a reduction in hemoglobin A1c (HbA1c),
body weight, and body mass index (BMI) compared with This review included studies with randomized-controlled tri-
those who performed moderate-intensity continuous train- als. We excluded matched controlled trial designs, uncon-
ing (MICT), though no significant changes in insulin resist- trolled trials, observational studies, and animal studies.
ance or blood lipid levels were found. Karstoft et al. [16]
compared the efficacy of HIIT with energy expenditure-
matched continuous-walking training in people with T2D Type of participant
and observed greater improvements in ­VO2peak, body weight,
fat mass, and glycemic control with the former. Mitranun The study participants were clinically diagnosed with type
et al. [17] also found that HIIT improved HbA1c, maximal 2 diabetes. Patients with type 1 diabetes and gestational dia-
aerobic capacity, and other cardiovascular risk factors in betes were excluded. There was no limitation on the age,
T2D patients, even if the total exercise time was reduced gender, or ethnicity of the study participants.
to half of that recommended. Similar to the current study, a
recent meta-analysis by Jelleyman et al. demonstrated that Intervention variables and outcome measures
HIIT is more effective than MICT for improving insulin sen-
sitivity and cardiorespiratory fitness in healthy individuals The studies included here were required to report at least one
[18]. However, this study did not determine the suitability outcome measure, measured at baseline and post-interven-
of HIIT in individuals with T2D. Indeed, although a few tion, and compared to either a moderate-intensity exercise
RCTs have demonstrated the efficiency of HIIT in the pre- intervention or control group. The HIIT program had to be
vention and treatment of T2D patients, no consensus has yet prescribed at least two times per week for 4 weeks, with
been reached that HIIT is a superior training protocol for moderate-intensity continuous training or another treatment
the improvement of glycemic control, body composition, (e.g., drug therapy) as the control group.
and cardiorespiratory fitness compared with moderate-inten-
sity continuous aerobic training among patients with T2D. Primary outcomes
Therefore, we performed a meta-analysis to determine the
impact of HIIT on body composition, glycemic control, and Outcome measures included glycemic control (e.g., HbA1c,
cardiorespiratory fitness, and to compare it to that of MICT fasting glucose, and fasting insulin); body composition [e.g.,
and that of no intervention in randomized-controlled trials in body weight, BMI, body fat (%), and waist circumference];
T2D patients, which we hope can provide clinical evidence cardiorespiratory fitness (e.g., ­VO2peak). The criteria which
to enable patients to achieve optimal outcomes.

13
Aging Clinical and Experimental Research (2019) 31:575–593 577

we used complied with the PICO concept (patient/problem/ included studies was assessed using the chi-squared test and
population; intervention; comparison/control/compara- I2 test. A threshold of P < 0.10 was considered to be statisti-
tor; outcome). For articles reported in more than two pub- cally significant and an I2 value > 50% was indicative of high
lications, only one full copy was used for meta-analysis. heterogeneity. We used the weighted mean difference (MD)
Abstracts presented at academic conferences, case reports, or standardized MD (SMD) with 95% confidence intervals
observational studies, examples of animal research, and (CIs) for summary statistics and derived such for the com-
studies of which the full text could not be obtained were parison of HIIT with MICT or other treatment methods. MD
excluded. was used when all studies reported the same outcome using
the same scale, while SMD was used when studies reported
Evaluation of bias and quality assessment different units or scales for the outcome. If heterogeneity did
not exist between studies, we incorporated a fixed-effects
The risk of bias and methodological quality of the included model approach to combined outcome measures. A random-
trials were assessed independently by two reviewers (Liu effects model was used when there was a large degree of
and Li), who used the Cochrane Collaboration’s tool [19] heterogeneity between studies. To account for within-group
to check for concealed allocation, allocation concealment, intervention effect sizes, we used fixed-effects modeling to
blinding, incomplete outcome data, selective reporting, and estimate the change from baseline. Potential heterogeneity
other biases. Each reviewer was required to award one of sources were identified by sensitivity analyses conducted by
three grades (either unclear, low risk, or high risk) to each omitting one study successively and comparing the influence
item. The Grading of Recommendations Assessment, Devel- of each study on the overall pooled estimate if I2 > 50%.
opment, and Evaluation (GRADE) system [20] was used Data were analyzed using the change from baseline for
to assess the quality of the evidence from very low to high both groups. If the study did not contain change data, we
based on risk of bias, inconsistency, indirectness, impreci- used the following two equations for conversion:
sion, and publication bias. A third reviewer was consulted
if any disagreement occurred.
M = |M1 − M2 |, (1)
where M is the effect mean, M1 is the mean of the baseline,
Data extraction and M2 is the end value mean;
S2 = S12 + S22 − 2 × R × S1 × S2 , (2)
The two investigators assessed each article’s title or abstract
where S is the standard deviation of the effect, S1 is the
for eligibility. When a disagreement happened, a third inves-
standard deviation of the baseline value, S2 is the final stand-
tigator participated in a discussion to reach a final consensus.
ard deviation, and R is constant (0.4 or 0.5).
For studies that met the inclusion criteria, full papers were
obtained for further analysis. The two authors independently
extracted data from the published works using standard data
extraction forms. If there were any inconsistencies in the Results
process of data extraction, the two authors would check the
original text and reach an agreement through discussion or Search results
through verification by a third author. Information on trial
design, characteristics of the patients, HIIT protocol, and The initial database searches returned a total of 484 articles
relevant results was noted according to a redesigned form. (PubMed, n = 84; EMBASE, n = 30; The Cochrane Library,
We recorded the name of the first author and the year of n = 63; EBSCO, n = 30; the Web of Science, n = 277) that
publication; the number of patients/participants and their were each screened and evaluated for eligibility based on
ages, gender, and BMIs; the duration of diagnosis; and the their respective titles only. Following removal of duplicates,
experimental and control interventions (e.g., exercise inten- 421 articles underwent further identification and screen-
sity and duration, interval intensity and duration, session ing. In total, 378 non-relevant articles were excluded after
time, and duration in weeks). When data were insufficient or screening the titles and abstracts. Of the remaining articles,
inapplicable, we attempted to contact the authors by e-mail 43 were selected to be read in full. At this point, 30 addi-
or used an equation to reveal all available data. tional articles were excluded for varying reasons (e.g., the
study was not randomized, there were reduplicative partici-
Data analysis pants, the study was observational in nature, the research
was performed on animals, the study was presented at an
The Review Manager software (RevMan 5.3; Cochrane, academic conference, and/or the study had no required data),
London, UK) was used to conduct the meta-analysis. The rendering a final sample of 13 papers. Figure 1 describes the
statistical heterogeneity of the treatment effect among the study selection flow.

13
578 Aging Clinical and Experimental Research (2019) 31:575–593

Fig. 1  Flowchart of the study

selection process Records identified though Additional records identified
database searching (n = 484) through other sources (n = 5)

Records after duplicates

removed (n = 421)

Records excluded
Records screened (n =421) for irrelevant studies
(n =378)

Full-text articles assessed 30 full-text articles were excluded

for eligibility (n = 43) for the following reasons:
Study design (20)
Reduplicative participants (2)
Included animal researches (2)
Studies included in Participation in academic
qualitative synthesis (n = 13) conferences (2)
Required data not available (4)

Studies included in
quantitative synthesis (n = 13)

Characteristics of included trials session ranged from 30 s to 4 min, and interval duration
ranged from 30 s to 3 min.
A total of 345 participants were included in the analysis,
of which 163 (47.2%) participants underwent a HIIT inter- Risk of bias among the selected articles
vention. The characteristics of the study participants, the
HIIT training protocols used, and the main results from the The 13 studies were assessed for risk of bias; the evalu-
included studies are described in Table 1. The countries or ation results are shown in Table 2. Among the included
regions of publication were mainly the United Kingdom studies, the method of randomization was only clearly
(n = 2), Norway (n = 2), the Republic of Korea (n = 1), stated in four studies [21, 25, 26, 28], while three reported
Chile (n = 1), Denmark (n = 2), France (n = 1), Thailand allocation concealment [25, 26, 28], five blinded partici-
(n = 1), Australia (n = 1), Italy (n = 1), and Canada (n = 1). pants or personnel [15, 16, 21, 23, 28], and three did not
The main HIIT intervention ranged in duration from 11 to employ assessor blinding [22, 24, 27]. Only one study did
16 weeks (16 weeks in 4 studies, 12 weeks in 8 studies, not clearly state complete outcomes data and employed
and 11 weeks in 1 study) and occurred two-to-five times selective reporting [22]; no other bias in all studies. The
weekly (median: three times). Total training duration per evaluation of the overall quality of evidence and results is
shown in Table 4, and the level of evidence for RCTs is

13
 Table 1  Characteristics of the included trials
Article, year Country Main characteristics of the HIIT MICT or CON No. of
subjects patient
Exercise intensity and Frequency and duration Exercise intensity Frequency and duration dropouts
interval

Alvarez Chile HIIT: mean age was Train: running at 90–100% 2.2–37.5 min/time, three CON: non-exercise 5
2016 [21] 45.6 ± 3.1 years, mean ­HRmax intensity for times per week for 16
duration of diagnosis was 30–120 s; Interval: low- weeks
3.14 ± 1.1 years, mean intensity walking for
BMI was 30.8 ± 1.0 kg/m2, 30–120 s
n = 13
CON: mean age was
43.1 ± 1.5 years, mean
duration of diagnosis was
3.6 ± 1.1 years, mean BMI
was 30.4 ± 0.4 kg/m2,
n = 10
Hollekin Norway Total of 47 patients Protocol was 4 × 4 min 40 min /bout; three times MICT: Moderate-intensity 210 min per week for 12 10
Aging Clinical and Experimental Research (2019) 31:575–593

2014 [22] 55.9 ± 6.0 years; 36% exercise at 90–95% H ­ Rmax per week for 12 weeks aerobics training weeks
female; mean dura- with 3 min of low-inten-
tion of diagnosis was sity exercise at 70% H­ Rmax
3.6 ± 2.5 years
HIIT: mean BMI was
30.2 ± 2.8 kg/m2, 5% in
mild stage, 90% in moder-
ate stage, 5% in severe
stage, n = 24
MICT: mean BMI was
29.7 ± 3.7 kg/m2, 29.4%
in mild stage, 70.6% in
moderate stage, n = 23

13
579
 Table 1  (continued)
580

Article, year Country Main characteristics of the HIIT MICT or CON No. of
subjects patient

13
Exercise intensity and Frequency and duration Exercise intensity Frequency and duration dropouts
interval

Karstoft Denmark HIIT: mean age was Alternating 3 min intervals Five times per week, MICT: walking at ≥ 55% Five times per week, for 4 0
2013 [16] 57.5 ± 2.4 years, 41.7% of fast (≥ 70% of V
­ O2peak) 60 min/time for 4 months ­VO2peak 60 min/session months
female, mean BMI was and slow (40% of V ­ O2peak) CON: non-exercise
29.0 ± 1.3 kg/m2, mean walking
duration of diagnosis was
3.5 ± 0.7 years, n = 12
MICT: mean age was
60.8 ± 2.3 years, 33.3%
female, mean BMI was
29.9 ± 1.6 kg/m2, mean
duration of diagnosis was
6.2 ± 1.5 years, n = 12
CON: mean age was
60.8 ± 2.3 years, 37.5%
females, mean BMI was
29.7 ± 1.9 kg/m2, mean
duration of diagnosis was
4.5 ± 1.5 years, n = 8
Lee 2015 [23] Korea Mean age was 15.3 ± 2.2 Exercise at ≥ 80% HRR, 400 Kcal/session, three MICT: Exercise at ≤ 40% Six sessions per week for 0
years, mean BMI was train program includ- sessions per week for 12 HRR, 200 kcal/session 12 weeks
24.0 ± 3.8 kg/m2, mean ing 30-s sprint and 30- s weeks
duration of diagnosis was recovery
4.0 ± 2.2 years, n = 20
Maillard France Included 16 postmenopausal Repeated cycles of sprinting Two times per week for 16 MICT: Exercise at 55–60% Two times/week, 16 weeks 1
2016 [24] women with T2D, mean for 8 s (at around 80% weeks HRR for 40 min
age was 69 ± 1 years, mean ­HRmax) followed by pedal-
BMI was 31 ± 1 kg/m2 ing slowly (20–30 rpm)
HIIT: mean age was for 12 s (maximum of 60
68.2 ± 1.9 years, mean cycles per 20-min session)
BMI was 32.6 ± 1.7 kg/
m2, N = 8
MICT: mean age was
70.1 ± 2.4 years, mean
BMI was 29.7 ± 1.2 kg/
m2, n = 8
Aging Clinical and Experimental Research (2019) 31:575–593
 Table 1  (continued)
Article, year Country Main characteristics of the HIIT MICT or CON No. of
subjects patient
Exercise intensity and Frequency and duration Exercise intensity Frequency and duration dropouts
interval

Mitranun Thailand Total of 43 adults with T2D Protocol was 1-min high- 20 min /session, three MICT: exercise intensity Three times/week, 12 weeks 2
2014 [17] (64.4% females), HIIT: intensity exercise at sessions per week for 12 at 50–60% V­ O2peak for CON: non-exercise
mean age was 61.2 ± 2.8 50–85% V ­ O2peak with weeks 25–30 min
years, mean BMI was 4-min low-intensity at
29.6 ± 0.5 kg/m2, mean 50–60% ­VO2peak interval
duration of diagnosis was
19.5 ± 1.5 years; n = 14.
MICT: mean age was
61.7 ± 2.7 years, mean
BMI was 29.4 ± 0.7 kg/m2,
mean duration of diagnosis
was 20.5 ± 1.5 years;
n = 14. CON: mean age
Aging Clinical and Experimental Research (2019) 31:575–593

was 60.9 ± 2.4 years, mean

BMI was 29.7 ± 0.4 kg/m2,
mean duration of diagnosis
was 21.1 ± 2.32 years;
n = 15
Ramos Australia HIIT: mean weight was Protocol was 4*4 min bouts 38 min per session, 3 times MICT: 30 min at 60–70% 5 times/week, 16 weeks 0
2016 [25] 99 ± 18 kg, n = 9 at 85–95% H ­ Rmax, interval per week for 16 weeks ­HRpeak
MICT: mean weight was with 3 min of active recov-
98 ± 17 kg, n = 6 ery at 50–70% ­HRmax
Støa Norway HIIT: mean age was 59 ± 11 Protocol was 4*4 min Three times per week for 12 MICT: 60-min walking at Three times per week for 12 5
2017 [15] years, mean BMI was of walking or uphill at weeks 70–75% of H
­ Rpeak weeks
32.0 ± 4.7 kg/m2, mean 85–95% of H ­ Rmax
duration of diagnosis was
9 ± 7 years, n = 19
MICT: mean age was
59 ± 10 years, mean BMI
was 31.1 ± 4.5 kg/m2,
mean duration of diagnosis
was 6 ± 5 years, n = 19

13
581
 Table 1  (continued)
582

Article, year Country Main characteristics of the HIIT MICT or CON No. of
subjects patient

13
Exercise intensity and Frequency and duration Exercise intensity Frequency and duration dropouts
interval

Terada Canada HIIT: mean age was 62 ± 3 HIIT protocol involved 30–60 min/day, 5 days per MICT: continuous exer- 5 days per week for 12 0
2013 [26] years, mean BMI was alternating 1-min intervals week for 12 weeks cise at 40% ­VO2peak, weeks
28.4 ± 4.1 kg/m2, 50% 100% ­VO2peak with 3-min 30–60 min/day
female, mean duration of recovery intervals at 20%
diagnosis was 6 ± 4 years, ­VO2peak
n=8
MICT: mean age was 63 ± 5
years, mean BMI was
33.1 ± 4.5 kg/m2, 42.9%
female, mean duration of
diagnosis was 8 ± 74 years,
n=7
Backx UK Involved 15 males and 4 Protocol was 1–2 min at 60 min/day, 3 days per week MICT: Exercise at moder- Five times/week, 12 weeks 2
2011 [27] females; total median age 40–50% HRR and 1, 2, or for 12 weeks ate-to-high-intensity for
was 59.6 (44.0–69.0) years 3 min at 80–90% HRR 30 min
HIIT: median BMI was 30.0
(25.3–40.1) kg/m2, n = 10
MICT: median BMI was
32.3 (26.4–40.5) kg/m2,
n=9
Cassidy UK HIIT: mean age was 61 ± 9 Training: Pedal Three sessions per week for Non-exercise 5
2016 [28] years, mean BMI was cadence > 80 rev/min, 12 weeks
31 ± 5 kg/m2, mean dura- ranching a RPE 16–17
tion of diagnosis was 5 ± 3 (very hard); interval:
years, n = 12 3-min recovery cycle
MICT: mean age was 59 ± 9
years, mean BMI was
32 ± 6 kg/m2, mean dura-
tion of diagnosis was 4 ± 2
years, n = 11
Bellia 2017 [29] Italy HIIT: mean age was Protocol involved a 4-min Two–three times per week MICT: protocol was 10,000 steps per day or 70,000 steps 7
58.8 ± 7.9, mean BMI was walk at 75–80% ­HRmax to for 12 weeks per week for 12 weeks
27.7 ± 2.8 kg/m2, mean be repeated two-to-four
duration of diagnosis was times, interval with 3-min
5.9 ± 4.4 years, n = 11 active recovery at 45–50%
CON: mean age was ­HRmax
56.3 ± 6.4, mean BMI was
29.9 ± 3.4 kg/m2, mean
duration of diagnosis was
3.4 ± 3.7 years, n = 11
Aging Clinical and Experimental Research (2019) 31:575–593
 Aging Clinical and Experimental Research (2019) 31:575–593 583

downgraded due to inconsistency and imprecision in most

RCT​ randomized-controlled trial, MCT Matched controlled trial designs, HRmax maximal heart rate, HIIT high-intensity interval training, MICT moderate-intensity continuous training, CON
of the studies.
dropouts
patient
No. of

Effects of HIIT on body composition

3
The included studies assessed body weight (11/13; 84.6%);
Frequency and duration

3 days/week, 11 weeks

BMI (11/13; 84.6%); body fat (6/13; 46.2%); waist circum-

ference (7/13; 53.8%) as outcomes. Of these, 8/11 (72.7%
of body weight studies); 8/11 (72.7% of BMI studies); 5/6
(83.3% of body fat studies); 6/7 (85.7% of waist circumfer-
ence studies) compared HIIT to MICT. The meta-analyses
showed (Table 3) a significant reduction in body weight of
1.22 kg [95% CI − 2.23 to − 0.18, P = 0.02] for patients in
MICT: 40 min of cycling at

the HIIT group as compared with those in the MICT group.

Furthermore, in comparison with baseline, there was a
reduction in BMI of 0.85 kg/m2 (95% CI − 1.57 to − 0.12,
Exercise intensity

50% of Wpeak
MICT or CON

P = 0.02) (Table 3), and, as compared with the MICT group,

the reduction was 0.40 kg/m2 (95% CI − 0.78 to − 0.02,
P = 0.04) (Table 3). In addition, as compared with baseline,
there was a reduction in body fat of 1.86% (95% CI − 3.68 to
− 0.04, P = 0.02) (Table 3), but the reduction was not statisti-
cally significant as compared with that in the MICT group.
Frequency and duration

In addition, there was no significant difference in the waist

3 days per week for 11

circumference reduction following HIIT versus MICT or at

baseline (Table 3).
weeks

Effects of HIIT on glycemic control

Ten studies with 220 patients assessed HbA1c. Of these,

20 min of cycling consist-

nine studies compared changes in HbA1c in HIIT groups to

of active recovery (20%
at 95% ­Wpeak and 1 min
Training: initiated with a

those in MICT groups, while only three studies compared

­Wpeak) was performed
ing of cycles of 1 min
Exercise intensity and

such to changes in CON groups. Relative to baseline, there

was a significant reduction in HbA1c (SMD: − 0.29, 95%
CI − 0.55 to − 0.04, P = 0.02) (Fig. 2a; Table 3). Compared
with MICT, the reduction was 0.37% (95% CI − 0.55 to
interval

− 0.19, P < 0.0001, Fig. 2a). However, in comparison with

HIIT

control, HRR heart rate reserve, RPE rating of perceived exertion

a control intervention, a non-significant change in HbA1c

28 ± 3.5 kg/m2, mean dura-

of -0.39% (95% CI − 0.81 to 0.02, P < 0.06, Fig. 2b) was

tion of diagnosis was 7 ± 5
MICT: mean age was 58 ± 8
duration of diagnosis was

duration of diagnosis was

Main characteristics of the

found. As compared with baseline, there was a significant

28.1 ± 3.5 kg/m2, mean

27.4 ± 3.1 kg/m2, mean

54 ± 6, mean BMI was

57 ± 7, mean BMI was

years, mean BMI was

reduction in fasting insulin (SMD: − 0.46, 95% CI − 0.81 to

CON: mean age was
Denmark HIIT: mean age was

8 ± 4 years, n = 13

6 ± 4 years, n = 12

− 0.11, P = 0.01, Table 3). However, this reduction was not

significantly different as compared with that in the control
years, n = 7

intervention or MICT groups (Table 3). No significant dif-

subjects

ference in the fasting glucose or HOMA-IR (homeostatic

model assessment of insulin resistance) was found for par-
ticipants in the HIIT group as compared with those in the
Country

MICT group (Table 3). We further used sensitivity analysis

in HOMA-IR because of the larger heterogeneity (I2 = 73%)
Table 1  (continued)

within the group. The results of sensitivity analysis showed

that the heterogeneity (I2 = 0%) was significantly reduced
Article, year

after exclusion of Lee 2015, but there was no significant

2018 [30]
Winding

change in results.

13
584 Aging Clinical and Experimental Research (2019) 31:575–593

Table 2  Risk-of-bias assessment for the included studies

Study Random sequence Allocation Blinding Incomplete Selective reporting Other bias
generation concealment outcome data
Participants or Outcome
personnel assessment

Alvarez [21] Low Unclear Low Low Low Low Low

Hollekin [22] Unclear Unclear Unclear Unclear Unclear Unclear Low
Karstoft [16] Unclear Unclear Low Low Low Low Low
Lee [23] Unclear Unclear Unclear Low Low Low Low
Maillard [24] Unclear Unclear Unclear Unclear Low Low Low
Mitranun [17] Unclear Unclear Unclear Low Low Low Low
Ramos [25] Low Low Low Low Low Low Low
Støa [15] High High Low Low Low Low Low
Terada [26] Low Low High Low Low Low Low
Backx [27] High Unclear Unclear Unclear Low Low Low
Cassidy [28] Low Low Low Low Low Low Low
Bellia [29] Unclear Unclear Unclear Low Low Low Low
Winding [30] Unclear Unclear Unclear Low Low Low Low

Effects of HIIT on lipid control increase in ­VO2peak with HIIT over control interventions.
The random-effects model showed (Fig. 4c) a significant
Seven studies assessed low-density lipoprotein (LDL) cho- improvement in absolute ­VO2peak of 0.37 L/min (95% CI
lesterol as an outcome. Of these, five studies compared the 0.28 to 0.45, P < 0.0001) for patients in HIIT group versus
change in the HIIT group to that in the control group and those in the MICT group and there was a similar increase
six studies compared the change in the HIIT to that in the seen with respect to relative ­VO2peak (MD: 3.37 ml/kg/min,
MICT group. There was also a significant reduction in LDL 95% CI 1.88 to 4.87, P < 0.0001) (Fig. 3c). However, there
cholesterol (MD: − 0.25 mmol/L 95% CI − 0.46 to − 0.04, existed moderate heterogeneity in this analysis (I2 = 48%)
P = 0.02) with HIIT versus with the MICT group (Table 3). and the results should be interpreted with caution (Table 4).
Unfortunately, there was no significant change in total cho-
lesterol as compared with both the control and MICT groups
and a similar result was found with respect to high-density Discussion
lipoprotein (HDL) cholesterol. LDL cholesterol did not dif-
fer significantly between the HIIT group and the control The purpose of this study was to evaluate the effectiveness
group. Because studies comparing HIIT with control inter- of HIIT on body composition, glycemic control, and car-
ventions in relation to LDL and HDL cholesterol showed diorespiratory fitness in patients with T2D; to observe the
significantly more heterogeneity, we conducted sensitivity difference in such compared with MICT or non-exercise;
analysis that showed that the studies heterogeneity changed and to provide information on an ideal time-efficient physi-
significantly (I2 = 20% in LDL cholesterol, I2 = 0 in HDL cal activity program. The principal finding of the current
cholesterol) after the removal of Alvarez 2016, but there meta-analysis was that HIIT was more efficient than MICT
were no significant changes in the results. in increasing V­ O2peak in T2D patients; they also found that
reduction of BMI, body weight, and HbA1c (%) was less
Effects of HIIT on cardiorespiratory fitness conclusive because of low quality of the evidence.
Excess weight and obesity are important risk factors for
Cardiorespiratory fitness as measured using absolute V­ O2peak the occurrence of T2D and contribute to the development of
(L/min) and relative ­VO2peak (ml/kg/min) was analyzed insulin resistance in obese individuals [31, 32]. Even with a
using data from seven studies representing a total of 219 body weight that falls within the normal range, individuals
patients. As compared with baseline, there was a 4.75 ml/ with an abnormal BMI and waist circumference can also
kg/min (95% CI 2.94 to 6.56, P < 0.0001) (Fig. 3a; Table 2) present with an increased risk of abnormal glucose metabo-
or 0.35 L/min (95% CI 0.17 to 0.53, P = 0.0001) increase in lism [33]. Our work showed that HIIT improved body com-
­VO2peak with HIIT (Fig. 4a; Table 3). In addition, there was position, reducing BMI significantly by 0.85 kg/m2 and
a 4.12 ml/kg/min (95% CI 2.66 to 5.57, P < 0.0001) (Fig. 3b) reducing body fat by 1.86%. Notably, both body weight
or 0.24 L/min (95% CI 0.10 to 0.37, P = 0.0005) (Fig. 4b) and BMI were significantly decreased compared with the

13
Aging Clinical and Experimental Research (2019) 31:575–593 585

Table 3  Effect of HIIT on body composition, glycemic control, lipid control, and cardiorespiratory fitness in patients with T2D
Body composition Within groups Compared to CON Compared to MICT

Body weight N 11 6 8
ES (95% CI) MD: − 1.65 [− 4.76, 1.46] MD: − 0.78 [− 2.36, 0.80] MD: − 1.22 [− 2.23, − 0.18]
I2 (%) 0 0 0
BMI N 11 4 8
ES (95% CI) MD: − 0.85 [− 1.57, − 0.12] MD: − 0.80 [− 1.86, 0.27] MD: − 0.40 [− 0.78, − 0.02]
I2 (%) 0 0 0
Body fat (%) N 6 ND 5
ES (95% CI) MD: − 1.86 [− 3.68, − 0.04] MD: − 0.50 [− 1.18, 0.19]
I2 (%) 0 0
Waist circumference N 7 ND 6
ES (95% CI) MD: − 2.23 [− 5.00, 0.55] MD: − 0.15 [− 1.21, 0.91]
I2 (%) 0 0
Glycemic control
HbA1c (%) N 10 3 9
ES (95% CI) MD: − 0.29 [− 0.55, − 0.04] MD: − 0.39 [− 0.81, 0.02] MD: − 0.37 [− 0.55, − 0.19]
I2 (%) 0 0 0
Fasting glucose N 9 5 8
ES (95% CI) MD: − 0.41 [− 0.91, 0.09] SMD: − 0.31 [− 0.69, 0.06] MD: 0.10 [− 0.84, 0.65]
I2 (%) 0 0 0
Fasting insulin N 6 5 4
ES (95% CI) SMD: − 0.46 [− 0.81, − 0.11] SMD: − 0.46 [− 0.91, 0.02] SMD: − 0.19 [− 0.58, 0.20]
I2 (%) 41 26 0
HOMA-IR N 7 4 6
ES (95% CI) MD: − 0.43 [− 1.18, 0.32] MD: − 0.18 [− 0.79, 0.42] MD: 0.13 [− 0.10, 0.36]
I2 (%) 73 0 0
Lipid control
Total cholesterol N 8 6 7
ES (95% CI) SMD: − 0.13 [− 0.42, 0.15] SMD: 0.02 [− 0.32, 037] MD: − 0.18 [− 0.44, 0.07]
I2 (%) 0 9 0
HDL cholesterol N 11 5 9
ES (95% CI) SMD: 0.20 [− 0.07, 0.48] SMD: 0.60 [− 0.26, 1.45] MD: − 0.04 [− 0.10, 0.02]
I2 (%) 39 83 0
LDL cholesterol N 7 5 6
ES (95% CI) SMD: − 0.15 [− 0.44, 0.13] MD: − 0.60 [− 1.74, 0.54] MD: − 0.25 [− 0.46, − 0.04]
I2 (%) 0 52 0
Cardiorespiratory fitness
VO2peak (ml/kg/min) N 7 2 7
ES (95% CI) MD: 4.75 [2.94, 6.56] MD: 4.12 [2.66, 5.57] MD: 3.37 [1.88, 4.87]
I2 (%) 0 0 48
VO2peak (L/min) N 5 2 6
ES (95% CI) MD: 0.35 [0.17, 0.53] MD: 0.24 [0.10, 0.37] MD: 0.37 [0.28, 0.45]
I2 (%) 0 0 36

ES effect sizes, CI confidence interval, MD mean difference, SMD standardized mean difference, ND not enough data

MICT group, which suggests that HIIT may be more effec- of fat from visceral fat stores. Maillard et al. [24] studied
tive for improving body composition (even in the absence of and compared the effects of HIIT and MICT on abdominal
changes in body fat and waist circumference) in individuals fat in postmenopausal women with T2D, and observed that
with T2D. The underlying mechanism of HIIT-induced body only HIIT reduced the subcutaneous and visceral fat mass
weight loss may be related to the consumption and release significantly following 16 weeks of training. Cassidy et al.

13
586 Aging Clinical and Experimental Research (2019) 31:575–593

Fig. 2  Forest plot for change in of HbA1c (%), a before and after (within-group) high-intensity interval training (HIIT), b between HIIT and
control (CON) intervention, and c between HIIT and moderate-intensity training (MICT)

[28] reported, in their randomized study, that there was a in individuals with T2D. Notably, however, although this
39% relative reduction in liver fat following HIIT perfor- review shows that HIIT has favorable effects on body fat
mance and observed that there was a significant correlation reduction in individuals with T2D, the effects of HIIT on
with changes in HbA1c and 2-h glucose. Moreover, Karstoft blood lipids were limited. Only LDL cholesterol showed
et al. [16] found that patients with T2D had greater oxygen significantly lower levels after HIIT than after MICT, while
consumption during HIIT training than did those who per- total cholesterol and HDL cholesterol did not. Thus, more
formed MICT, suggesting that this may be responsible for studies are required to determine whether HIIT could be a
their greater weight loss. Recent studies have shown that successful training program for lipid control in T2D patients.
the positive effects of exercise on body composition may be HbA1c is not only the most widely used indicator of
related to the improvement of glycemic control. For exam- glucose: it is also an important risk factor of cardiovas-
ple, in a long-term randomized trial, Senechal et al. [34] cular disease in patients with T2D [35, 36]. The previous
found that changes in HbA1c were associated with changes studies have shown that if HbA1c levels are reduced by
in body weight, waist circumference, and trunk fat mass 1%, the risk of microvascular complications is reduced by

13
Aging Clinical and Experimental Research (2019) 31:575–593 587

Fig. 3  Forest plot for change in ­VO2peak (ml/kg/min), a before and after (within-group) high-intensity interval training (HIIT), b between HIIT
and control (CON) intervention, and c between HIIT and moderate-intensity training (MICT)

37% and that of death related to diabetes can be reduced review with a meta-analysis, it was concluded that exercise
by 21% [35]. A recent meta-analysis has shown that an intensity was a better predictor of weight MD in HbA1c
increase of 100 min in physical activity per week was asso- than exercise volume in T2D patients [38]. Unfortunately,
ciated with an average change of − 0.16% of HbA1c in we noted no difference in fasting glucose, fasting insulin,
individuals with T2D and pre-diabetes subjects [37]. In or insulin resistance changes in patients following HIIT as
our meta-analysis, HbA1c (%) was found to be lower after compared with the CON and MICT groups, even though
HIIT than at baseline (SMD: − 0.29, 95% CI − 0.55 to the previous studies have shown that the effects of aerobic
− 0.04). Similar to our findings, a recent meta-analysis of training on insulin intensity are more closely influenced
RCTs by Grace et al. identified the positive effects of aero- by high-exercise intensity than by low- or moderate-inten-
bic exercise in reducing HbA1c levels over with controls sity exercise [39]. The inconsistent results could partly
[11]. HIIT showed a 0.37% greater reduction of HbA1c be explained by the difference among methods used to
than MICT, which means that HIIT may have additional measure insulin sensitivity, as well as the difference in
benefits on glycemic control. This is inconsistent with the the baseline of glycemic control. Further research would
findings of a meta-analysis conducted by Jelleyman et al. need to include data on the HIIT intervention program
[18], which found that, while HIIT can reduce the levels (e.g., training intensity, duration of interval time, fre-
of HbA1c in patients with diabetes and metabolic syn- quency of training, and total duration) and the character-
drome, there is no significant difference in reduction ver- istics of patients (especially with respect to age, duration
sus with continuous training. Furthermore, in a previous

13
588 Aging Clinical and Experimental Research (2019) 31:575–593

Fig. 4  Forest plot for change in ­VO2peak (L/min), a before and after (within-group) high-intensity interval training (HIIT), b between HIIT and
control (CON) intervention, and c between HIIT and moderate-intensity training (MICT)

of diabetes, and the baseline glycemic control), which all Batacan et al. [43] revealed that HIIT yielded a significant
impact trial results. increase in V
­ O2peak by a large amount in normal-weight pop-
Both ­VO2peak and HbA1c are important predictors of car- ulations and a medium effect in overweight/obese popula-
diovascular events in T2D patients [35], and the previous tions, with an aggregate improvement of 3.8 and 4.43 ml/
studies have shown that low cardiorespiratory fitness was kg/min, respectively. A more recent meta-analysis includ-
associated with an increased risk for impaired glycemic con- ing 594 coronary artery disease patients by Gomes-Noto
trol [40, 41]. Aerobic exercise training represents an effec- et al. [44] reported that a higher improvement in V ­ O2peak
tive means to improve ­VO2peak and HbA1c, and a previous (MD: 1.3 ml/kg/min, 95% CI 0.6 to 1.9 ml/kg/min) was seen
meta-analysis has revealed that aerobic exercise intensity with HIIT versus with MICT. The underlying physiologi-
is the primary stimulus for improved ­VO2peak in people cal mechanisms of HIIT that improve peak ­VO2 could not
with T2D [11]. Our study further compared the difference be ascertained from the present study, but may involve a
between HIIT and MICT in increasing peak ­VO2 and found combination of central and peripheral adaptations, includ-
that the improvement of 3.37 ml/kg/min in relative V ­ O2peak ing an increase in cardiac output, an improvement in vas-
and 0.37 L/min in absolute ­VO2peak following HIIT is supe- cular/endothelial function, and increased muscle oxidation,
rior to those seen with MICT. Our findings are similarly to which together promote the enhanced availability, extrac-
those from other recent studies. A meta-analysis focused tion, and use of oxygen during exercise [45, 46]. Revdal
mainly on cardiac patients by Xie et al. [42] showed that et al. [47] studied the impact of HIIT on cardiac structure
HIIT is more effective than continuous training in improv- and function in T2D patients, and observed a 12% relative
ing ­VO2peak [MD: 1.76 ml/kg/min, 95% CI 1.06 to 2.46 ml/ increase in left-ventricular wall mass and increased end-
kg/min]. Another systemic analysis analyzing 65 studies by diastolic blood volume, thus demonstrating improvements

13
Aging Clinical and Experimental Research (2019) 31:575–593 589

Table 4  Summary of GRADE’s approach to rating quality of evidence

Outcomes Quality assessment
Comparison Participants Risk of bias Inconsistency Indirectness Imprecision Publication bias Overall quality of
(studies) follow evidence
up

Body weight MICT 185 (eight stud- None None Serious Serious Undetected ⊕⊕⊝⊝ Low due
ies) to indirectness
and imprecision
CON 136 (six studies) None None Serious Serious Undetected ⊕⊕⊝⊝ Low due
to indirectness
and imprecision
BMI MICT 207 (eight stud- None None Serious Serious Undetected ⊕⊕⊝⊝ Low due
ies) to indirectness
and imprecision
CON 72 (three stud- None None Serious Serious Undetected ⊕⊕⊝⊝ Low due
ies) to indirectness
and imprecision
Body fat (%) MICT 138 (five stud- Serious Very serious None Serious Undetected ⊕⊝⊝⊝ Very low
ies) due to risk of
bias, incon-
sistency and
imprecision
CON ND ND ND ND ND ND ND
Waist circum- MICT 140 (six studies) None Serious None Serious Undetected ⊕⊕⊝⊝ Low due
ference to inconsistency
and imprecision
CON ND ND ND ND ND ND ND
HbA1c (%) MICT 209 (nine stud- None Serious None Serious Undetected ⊕⊕⊝⊝ Low due
ies) to inconsist-
ency, impreci-
sion
CON 63 (three stud- None Serious None Serious Undetected ⊕⊕⊝⊝ Low due
ies) to inconsistency
and imprecision
Fasting glucose MICT 162 (eight stud- None Serious None Serious Undetected ⊕⊕⊝⊝ Low due
ies) to inconsistency
and imprecision
CON 114 (five stud- None Serious None Serious Undetected ⊕⊕⊝⊝ Low due
ies) to inconsistency
and imprecision
Fasting insulin MICT 103 (five stud- None Serious None Serious Undetected ⊕⊕⊝⊝ Low due
ies) to inconsistency
and imprecision
CON 85 (four studies) None Serious None Serious Undetected ⊕⊕⊝⊝ Low due
to inconsistency
and imprecision
HOMA-IR MICT 182 (seven stud- None Very serious None Serious Undetected ⊕⊝⊝⊝ Very low
ies) due to incon-
sistency and
imprecision
CON 99 (four studies) None Very serious None Serious Undetected ⊕⊝⊝⊝ Very low
due to incon-
sistency and
imprecision

13
590 Aging Clinical and Experimental Research (2019) 31:575–593

Table 4  (continued)
Outcomes Quality assessment
Comparison Participants Risk of bias Inconsistency Indirectness Imprecision Publication bias Overall quality of
(studies) follow evidence
up

Total choles- MICT 165 (seven stud- None Serious None Serious Undetected ⊕⊕⊝⊝ Low due
terol ies) to inconsistency
and imprecision
CON 137 (six studies) None Serious None Serious Undetected ⊕⊕⊝⊝ Low due
to inconsistency
and imprecision
HDL choles- MICT 204 (nine stud- None Serious None Serious Reporting bias ⊕⊝⊝⊝ Very
terol ies) strongly sus- low due to
pected inconsistency,
imprecision and
publication bias
CON 114 (five stud- None Serious None Serious Reporting bias ⊕⊝⊝⊝ Very
ies) strongly sus- low due to
pected inconsistency,
imprecision and
publication bias
LDL MICT 150 (six studies) None Serious None Serious Reporting bias ⊕⊝⊝⊝ Very
strongly sus- low due to
pected inconsistency,
imprecision and
publication bias
CON 114 (five stud- None Very serious None Serious None ⊕⊝⊝⊝ Very low
ies) due to incon-
sistency and
imprecision
VO2peak (L/ MICT 159 (six studies) None Serious None Serious Undetected ⊕⊕⊕⊝ Moder-
min) ate due to
inconsistency,
imprecision and
large effect
CON 40 (two studies) None Serious None Serious Undetected ⊕⊕⊝⊝ Low due
to inconsistency
and imprecision
VO2peak (ml/ MICT 182 (seven stud- Serious Serious None Serious Undetected ⊕⊕⊕⊝ Moder-
kg/min) ies) ate due to
inconsistency,
imprecision and
large effect
CON 40 (two studies) None Serious None Serious Undetected ⊕⊕⊝⊝ Low due
to inconsistency
and imprecision

in systolic function, as indicated by raised stroke volume activity of citrate synthesis and skeletal muscle mitochon-
and left-ventricular ejection fraction. A similar finding was drial protein content, suggesting that the increases in skel-
found by Hollekin et al. [22], who observed that both MICT etal muscle mitochondrial content and function following
and HIIT groups showed improved diastolic function at rest, low-volume HIIT may be contributing factors to improved
but that the HIIT group showed greater improvement than ­VO2peak.
did the MICT group. Moreover, Little et al. [48] found that
people with T2D who performed six sessions of low-volume
HIIT at an intensity of 90% of the maximal heart rate with
60-s rest over 2 weeks experienced an increase in maximal

13
Aging Clinical and Experimental Research (2019) 31:575–593 591

Strengths and limitations Author contributions Jing-xin Liu contributed to study conception and
design, drafting the submitted article, and critically revising the draft
for important intellectual content. Lin Zhu revised the draft critically
Our meta-analysis of randomized trials has several strengths. for important intellectual content and gave final approval of the version
First, to our knowledge, this is the first existing systematic for publication. Pei-jun Li, Ning Li, and Yan-bing Xu contributed to
review to compare the effects of HIIT and MICT or non- acquisition, analysis, and interpretation the data. All authors contrib-
exercise on glycemic control (e.g., HbA1c, insulin, and fast- uted at all stages of this study, gave final approval of the version for
publication, and agree to be accountable for all aspects of the work.
ing glucose); body composition (e.g., body weight, body fat,
BMI, and waist circumference); and cardiorespiratory fitness Funding This work was supported by the National Planning Office of
(e.g., ­VO2peak) among people with T2D. Second, this system- Philosophy and Social Science of China (No. 18BTY075), the research
atic review involved a large number of literature searches by projects of the Social Science and Humanity on Young Fund of the
Ministry of education of China (No. 13yjc890050), the research pro-
two reviewers who independently screened studies, assessed
jects of the Department of Education of Guangdong Province (No.
their quality, and extracted data to decrease publishing bias 2015KTSCX079), the research projects of the Department of Science
and increase credibility. and Technology of Guangdong Province (No. 2015A020219010 and
However, some limitations were still present in our No. 2014A020220010).
evaluation. First, there are some inconsistencies among the
included studies with respect to HIIT protocols and MICT Compliance with ethical standards
protocols, which may have affected the results obtained
with respect to the intervention and control groups. Second, Conflict of interest The authors declare that there is no conflict of in-
terest regarding the publication of this paper.
considering the low quality of evidence, these results may
have some limitations in guiding clinical applications. Third, Statement of human and animal rights This review does not contain
an important limitation is that most of the included studies any experiments involving human participants or animals performed
reported the pre- and post-intervention parameters but not by any of authors.
the differences between the baselines. Therefore, consider- Informed consent For this review, formal consent forms were not
ing the different baseline values that may be present between required.
the intervention and control groups in some studies, we used
equations to calculate the mean difference whenever it was Open Access This article is distributed under the terms of the Crea-
not reported to address the discrepancy of the baseline in tive Commons Attribution 4.0 International License (http://creat​iveco​
each group, and this could have resulted in a bias. Fourth, mmons​.org/licen​ses/by/4.0/), which permits unrestricted use, distribu-
tion, and reproduction in any medium, provided you give appropriate
the results of this meta-analysis are limited by the lack of
credit to the original author(s) and the source, provide a link to the
high-quality studies and the small number of patients in each Creative Commons license, and indicate if changes were made.
included study. Only four of the included studies clearly
indicated random sequence generation, while three studies
reported allocation concealment, and five studies blinded
participants in their experimental procedures. References
1. WHO (2016) Global Report on Diabetes. WHO. http://www.who.
int/diabe​tes/publi​catio​ns/grd-2016/en/. Accessed 6 April 2016
2. Whiting DR, Guariguata L, Weil C et al (2011) IDF diabetes atlas:
Conclusions global estimates of the prevalence of diabetes for 2011 and 2030.
Diabetes Res Clin Pract 94:311–321. https​://doi.org/10.1016/j.
In conclusion, we here demonstrated that HIIT is an effective diabr​es.2011.10.029
3. Johansen MY, MacDonald CS, Hansen KB et al (2017) Effect of
strategy for improving cardiorespiratory fitness in patients an intensive lifestyle intervention on glycemic control in patients
with T2D, preferable to MICT. Results related to other with type 2 diabetes: a randomized clinical trial. JAMA 318:637–
parameters associated with the prognosis of T2D, such as 646. https​://doi.org/10.1001/jama.2017.10169​
HbA1c, body weight, and BMI, were not conclusive. This 4. Ribisl PM, Gaussoin SA, Lang W et al (2012) Lifestyle interven-
tion improves heart rate recovery from exercise in adults with
review can still provide some suggestions for the clinical type 2 diabetes: results from the Look AHEAD study. J Obes
application of HIIT in T2D patients. Future studies should 2012:309196. https​://doi.org/10.1155/2012/30919​6
investigate the effects of HIIT in T2D patients through mul- 5. Colberg SR, Sigal RJ, Fernhall B et al (2010) Exercise and type
ticenter RCTs with large sample sizes over the long term. 2 diabetes: the American College of Sports Medicine and the
American Diabetes Association: joint position statement executive
summary. Diabetes Care 33:2692–2696. https​://doi.org/10.2337/
Acknowledgements We would like to thank LetPub (http://www.letpu​
dc10-1548
b.com) for providing linguistic assistance during the preparation of
6. Anderson JE, Greene MA, Griffin JW et al (2012) Diabetes and
this manuscript.
employment. Diabetes Care 35:S94–S98. https:​ //doi.org/10.2337/
dc12-s094

13
592 Aging Clinical and Experimental Research (2019) 31:575–593

7. Sigal RJ, Kenny GP, Boule NG et al (2007) Effects of aerobic dysfunction: a randomized controlled trial. J Am Coll Cardiol
training, resistance training, or both on glycemic control in type 2 64:1758–1760. https​://doi.org/10.1016/j.jacc.2014.07.971
diabetes: a randomized trial. Ann Intern Med 147:357–369. https​ 23. Lee SS, Yoo JH, So YS (2015) Effect of the low- versus high-
://doi.org/10.7326/0003-4819-147-6-20070​9180-00005​ intensity exercise training on endoplasmic reticulum stress and
8. Church TS, Blair SN, Cocreham S et al (2010) Effects of aerobic GLP-1 in adolescents with type 2 diabetes mellitus. J Phys Ther
and resistance training on hemoglobin A1c levels in patients with Sci 27:3063–3068. https​://doi.org/10.1589/jpts.27.3063
type 2 diabetes: a randomized controlled trial. JAMA 304:2253– 24. Maillard F, Rousset S, Pereira B et al (2016) High-intensity
2262. https​://doi.org/10.1001/jama.2010.1710 interval training reduces abdominal fat mass in postmenopausal
9. Jorge ML, de Oliveira VN, Resende NM et al (2011) The effects women with type 2 diabetes. Diabetes Metab 42:433–441. https​
of aerobic, resistance, and combined exercise on metabolic con- ://doi.org/10.1016/j.diabe​t.2016.07.031
trol, inflammatory markers, adipocytokines, and muscle insulin 25. Ramos JS, Dalleck LC, Borrani F et al (2016) The effect of dif-
signaling in patients with type 2 diabetes mellitus. Metabolism ferent volumes of high-intensity interval training on proinsu-
60:1244–1252. https​://doi.org/10.1016/j.metab​ol.2011.01.006 lin in participants with the metabolic syndrome: a randomised
10. Motahari-Tabari N, Ahmad Shirvani M, Shirzad EAM et al (2014) trial. Diabetologia 59:2308–2320. https​://doi.org/10.1007/s0012​
The effect of 8 weeks aerobic exercise on insulin resistance in 5-016-4064-7
type 2 diabetes: a randomized clinical trial. Global J Health Sci 26. Terada T, Friesen A, Chahal BS et al (2013) Feasibility and pre-
7:115–121. https​://doi.org/10.5539/gjhs.v7n1p​115 liminary efficacy of high intensity interval training in type 2 diabe-
11. Grace A, Chan E, Giallauria F et al (2017) Clinical outcomes tes. Diabetes Res Clin Pract 99:120–129. https:​ //doi.org/10.1016/j.
and glycaemic responses to different aerobic exercise training diabr​es.2012.10.019
intensities in type II diabetes: a systematic review and meta-anal- 27. Backx K, McCann A, Wasley D et al (2011) The effect of a sup-
ysis. Cardiovasc Diabetol 16:37. https​://doi.org/10.1186/s1293​ ported exercise programme in patients with newly diagnosed Type
3-017-0518-6 2 diabetes: a pilot study. J Sports Sci 29:579–586. https​://doi.
12. Ellingsen O, Halle M, Conraads V et al (2017) High-intensity org/10.1080/02640​414.2010.54466​6
interval training in patients with heart failure with reduced ejec- 28. Cassidy S, Thoma C, Hallsworth K et al (2016) High intensity
tion fraction. Circulation 135:839–849. https​://doi.org/10.1161/ intermittent exercise improves cardiac structure and function and
circu​latio​naha.116.02292​4 reduces liver fat in patients with type 2 diabetes: a randomised
13. Wisloff U, Stoylen A, Loennechen JP et al (2007) Superior car- controlled trial. Diabetologia 59:56–66. https​://doi.org/10.1007/
diovascular effect of aerobic interval training versus moderate s0012​5-015-3741-2
continuous training in heart failure patients: a randomized study. 29. Bellia A, Iellamo F, De Carli E et al (2017) Exercise individu-
Circulation 115:3086–3094. https​://doi.org/10.1161/circu​latio​ alized by TRIMPi method reduces arterial stiffness in early
naha.106.67504​1 onset type 2 diabetic patients: a randomized controlled trial with
14. Phillips BE, Kelly BM, Lilja M et al (2017) A practical and aerobic interval training. Int J Cardiol 248:314–319. https​://doi.
time-efficient high-intensity interval training program modifies org/10.1016/j.ijcar​d.2017.06.065
cardio-metabolic risk factors in adults with risk factors for type II 30. Winding KM, Munch GW, Iepsen UW et al (2018) The effect
diabetes. Front Endocrinol 8:229. https​://doi.org/10.3389/fendo​ on glycaemic control of low-volume high-intensity interval train-
.2017.00229​ ing versus endurance training in individuals with type 2 diabetes.
15. Støa EM, Meling S, Nyhus LK et al (2017) High-intensity aerobic Diabetes 20:1131–1139. https​://doi.org/10.1111/dom.13198​
interval training improves aerobic fitness and HbA1c among per- 31. Kahn SE, Hull RL, Utzschneider KM (2006) Mechanisms linking
sons diagnosed with type 2 diabetes. Eur J Appl Physiol 117:455– obesity to insulin resistance and type 2 diabetes. Nature 444:840–
467. https​://doi.org/10.1007/s0042​1-017-3540-1 846. https​://doi.org/10.1038/natur​e0548​2
16. Karstoft K, Winding K, Knudsen SH et al (2013) The effects of 32. Heianza Y, Kato K, Kodama S et al (2015) Risk of the develop-
free-living interval-walking training on glycemic control, body ment of Type 2 diabetes in relation to overall obesity, abdominal
composition, and physical fitness in type 2 diabetic patients: a obesity and the clustering of metabolic abnormalities in Japanese
randomized, controlled trial. Diabetes Care 36:228–236. https​:// individuals: does metabolically healthy overweight really exist?
doi.org/10.2337/dc12-0658 The Niigata Wellness Study. Diabetic Med 32:665–672. https​://
17. Mitranun W, Deerochanawong C, Tanaka H et al (2014) Con- doi.org/10.1111/dme.12646​
tinuous vs interval training on glycemic control and macro- and 33. Li S, Xiao J, Ji L et al (2014) BMI and waist circumference are
microvascular reactivity in type 2 diabetic patients. Scand J Med associated with impaired glucose metabolism and type 2 diabetes
Sci Sports 24:e69–e76. https​://doi.org/10.1111/sms.12112​ in normal weight Chinese adults. J Diabetes Complicat 28:470–
18. Jelleyman C, Yates T, O’Donovan G et al (2015) The effects of 476. https​://doi.org/10.1016/j.jdiac​omp.2014.03.015
high-intensity interval training on glucose regulation and insulin 34. Senechal M, Swift DL, Johannsen NM et al. (2013) Changes in
resistance: a meta-analysis. Obes Rev 16:942–961. https​://doi. body fat distribution and fitness are associated with changes in
org/10.1111/obr.12317​ hemoglobin A1c after 9 months of exercise training: results from
19. Higgins JP, Altman DG, Gotzsche PC et al (2011) The Cochrane the HART-D study. Diabetes Care 36:2843–2849. https​://doi.
Collaboration’s tool for assessing risk of bias in randomised trials. org/10.2337/dc12-2428
BMJ 343:d5928. https​://doi.org/10.1136/bmj.d5928​ 35. Stratton IM, Adler AI, Neil HA et al (2000) Association of gly-
20. Balshem H, Helfand M, Schunemann HJ et al (2011) GRADE caemia with macrovascular and microvascular complications of
guidelines: 3. Rating the quality of evidence. J Clin Epidemiol type 2 diabetes (UKPDS 35): prospective observational study.
64:401–406. https​://doi.org/10.1016/j.jclin​epi.2010.07.015 BMJ 321:405–412. https​://doi.org/10.1136/bmj.321.7258.405
21. Alvarez C, Ramirez-Campillo R, Martinez-Salazar C et al 36. Woerle HJ, Neumann C, Zschau S et al (2007) Impact of fasting
(2016) Low-volume high-intensity interval training as a therapy and postprandial glycemia on overall glycemic control in type 2
for type 2 diabetes. Int J Sports Med 37:723–729. https​://doi. diabetes Importance of postprandial glycemia to achieve target
org/10.1055/s-0042-10493​5 HbA1c levels. Diabetes Res Clin Pract 77:280–285. https​://doi.
22. Hollekim-Strand SM, Bjorgaas MR, Albrektsen G et al (2014) org/10.1016/j.diabr​es.2006.11.011
High-intensity interval exercise effectively improves cardiac 37. Boniol M, Dragomir M (2017) Physical activity and change
function in patients with type 2 diabetes mellitus and diastolic in fasting glucose and HbA1c: a quantitative meta-analysis

13
Aging Clinical and Experimental Research (2019) 31:575–593 593

of randomized trials. Acta Diabetol 54:983–991. https​://doi. 44. Gomes-Neto M, Duraes AR, Reis H et al (2017) High-intensity
org/10.1007/s0059​2-017-1037-3 interval training versus moderate-intensity continuous training
38. Boule NG, Kenny GP, Haddad E et al (2003) Meta-analysis of the on exercise capacity and quality of life in patients with coronary
effect of structured exercise training on cardiorespiratory fitness artery disease: A systematic review and meta-analysis. Eur J Pre-
in Type 2 diabetes mellitus. Diabetologia 46:1071–1081. https​:// vent Cardiol 24:1696–1707. https​://doi.org/10.1177/20474​87317​
doi.org/10.1007/s0012​5-003-1160-2 72837​0
39. DiPietro L, Dziura J, Yeckel CW et al (2006) Exercise and 45. Ramos JS, Dalleck LC, Tjonna AE et al (2015) The impact of
improved insulin sensitivity in older women: evidence of the high-intensity interval training versus moderate-intensity con-
enduring benefits of higher intensity training. J Appl Physiol tinuous training on vascular function: a systematic review and
(Bethesda Md 1985) 100:142–149. https​://doi.org/10.1152/jappl​ meta-analysis. Sports Med (Auckland NZ) 45:679–692. https​://
physi​ol.00474​.2005 doi.org/10.1007/s4027​9-015-0321-z
40. Nojima H, Yoneda M, Watanabe H et al (2017) Association 46. Moholdt T, Aamot IL, Granoien I et al (2012) Aerobic inter-
between aerobic capacity and the improvement in glycemic con- val training increases peak oxygen uptake more than usual
trol after the exercise training in type 2 diabetes. Diabetol Metab care exercise training in myocardial infarction patients: a ran-
Syndr 9:63. https​://doi.org/10.1186/s1309​8-017-0262-9 domized controlled study. Clin Rehabil 26:33–44. https​://doi.
41. Wei M, Gibbons LW, Mitchell TL et al (1999) The association org/10.1177/02692​15511​40522​9
between cardiorespiratory fitness and impaired fasting glucose 47. Revdal A, Hollekim-Strand SM, Ingul CB (2016) Can time effi-
and type 2 diabetes mellitus in men. Ann Intern Med 130:89–96. cient exercise improve cardiometabolic risk factors in type 2 dia-
https​://doi.org/10.7326/0003-4819-130-2-19990​1190-00002​ betes? a pilot study. J Sports Sci Med 15:308–313
42. Xie B, Yan X, Cai X (2017) Effects of high-intensity interval 48. Little JP, Gillen JB, Percival ME et al (2011) Low-volume high-
training on aerobic capacity in cardiac patients. Syst Rev Meta intensity interval training reduces hyperglycemia and increases
Anal. https​://doi.org/10.1155/2017/54208​40 muscle mitochondrial capacity in patients with type 2 diabetes.
43. Batacan RB, Duncan MJ, Dalbo VJ et al (2017) Effects of high- J Appl Physiol (Bethesda Md 1985) 111:1554–1560. https​://doi.
intensity interval training on cardiometabolic health: a systematic org/10.1152/jappl​physi​ol.00921​.2011
review and meta-analysis of intervention studies. Br J Sports Med
51:494–503. https​://doi.org/10.1136/bjspo​rts-2015-09584​1

13