Available via license: CC BY-NC 4.0
Content may be subject to copyright.
Uncorrected Author Proof
Nutrition and Healthy Aging xx (20xx) x–xx
DOI 10.3233/NHA-170036
IOS Press
1
Effects of 8-hour time restricted feeding
on body weight and metabolic disease risk
factors in obese adults: A pilot study
1
2
3
Kelsey Gabela, Kristin K. Hoddya, Nicole Haggertya, Jeehee Songa, Cynthia M. Kroegera,b,
John F. Trepanowskia, Satchidananda Pandacand Krista A. Varadya,∗
4
5
aDepartment of Kinesiology and Nutrition, University of Illinois at Chicago, Chicago, IL, USA6
bSchool of Public Health, Indiana University, Bloomington, IN, USA7
cRegulatory Biology Lab, Salk Institute for Biological Studies, La Jolla, CA, USA8
Abstract.9
BACKGROUND: Time restricted feeding decreases energy intake without calorie counting and may be a viable option for
weight loss. However, the effect of this diet on body weight in obese subjects has never been examined.
10
11
OBJECTIVE: This study investigated the effects of 8-h time restricted feeding on body weight and metabolic disease risk
factors in obese adults.
12
13
DESIGN: Obese subjects (n= 23) participated in an 8-h time restricted feeding intervention (ad libitum feeding between
10:00 to 18:00 h, water fasting between 18:00 to 10:00 h) for 12 weeks. Weight loss and other outcomes were compared to a
matched historical control group (n= 23).
14
15
16
RESULTS: Body weight and energy intake decreased in the time restricted group (–2.6% ±0.5; –341 ±53 kcal/d) relative
to controls over 12 weeks (P<0.05). Systolic blood pressure decreased in the time restricted feeding group (–7±2 mm Hg)
versus controls (P< 0.05). Fat mass, lean mass, visceral fat mass, diastolic blood pressure, LDL cholesterol, HDL cholesterol,
triglycerides, fasting glucose, fasting insulin, HOMA-IR, and homocysteine were not significantly different from controls
after 12 weeks (no group ×time interaction).
17
18
19
20
21
CONCLUSIONS: These findings suggest that 8-h time restricted feeding produces mild caloric restriction and weight loss,
without calorie counting. It may also offer clinical benefits by reducing blood pressure.
22
23
Keywords: Time restricted feeding, intermittent fasting, body weight, metabolic disease risk factors, obese adults24
1. Introduction
25
Intermittent fasting has gained considerable pop-26
ularity over the past decade. There are two major
27
subcategories of intermittent fasting: 1) fasting 1–428
d per week, i.e. alternate day fasting or the 5:2 diet29
[1]; or 2) fasting every day for a 14 to 20h period, i.e.
30
time restricted feeding [2, 3]. Alternate day fasting31
and 5:2 are the most widely studied forms of inter-
32
∗Corresponding author: Krista A. Varady, PhD, Associate Pro-
fessor of Nutrition, Department of Kinesiology and Nutrition,
University of Illinois at Chicago, 1919 West Taylor Street, Room
532, Chicago, 60612, IL, USA. Tel.: +1 312 996 7897; E-mail:
varady@uic.edu.
mittent fasting [4–10]. Human trials of alternate day 33
fasting and 5:2 generally demonstrate reductions in 34
body weight of 3 to 8% after 8 to 52 weeks of treat- 35
ment, accompanied by decreases in blood pressure, 36
LDL cholesterol, triglycerides, and insulin resistance 37
[4–10]. The effects of time restricted feeding on the 38
other hand, have only been tested in three human tri- 39
als to date [11–13]. Gill and Panda [11] examined the 40
effects of 10-h time restricted feeding in overweight 41
healthy adults and showed a 4% weight loss that was 42
sustained for one year. Findings from the other two 43
studies [12, 13] reveal that 4–8 h time restricted feed- 44
ing reduces caloric intake (without calorie counting) 45
and significantly decreases fat mass without chang- 46
ISSN 2451-9480/18/$35.00 © 2018 – IOS Press and the authors. All rights reserved
This article is published online with Open Access and distributed under the terms of the Creative Commons Attribution Non-Commercial License (CC BY-NC 4.0).
Uncorrected Author Proof
2K. Gabel et al. / Time restricted feeding for weight loss
ing lean mass in young resistance trained men. While47
these preliminary studies offer promise for the use of
48
time restricted feeding in reducing energy intake and49
fat mass, additional trials are necessary to confirm50
these findings.51
Obesity greatly increases the risk of metabolic
52
diseases, such a coronary heart disease and type 2 dia-53
betes [14]. Accumulating evidence suggests that even54
small amounts of weight loss can lead to improve-55
ments in metabolic health [15]. Although alternate
56
day fasting and 5:2 have been shown to be effec-57
tive for weight loss [4–9], recent research suggests58
that obese individuals may have difficulties sticking
59
to these diets long-term [10]. Time restricted feed-60
ing is an attractive alternative to alternate day fasting
61
and 5:2, as this diet allows for ad libitum feeding62
within a large window of time each day, and does not63
require any calorie counting [2, 3]. However, no trial
64
to date has examined whether time restricted feeding65
is indeed an effective diet therapy for weight loss and66
metabolic disease risk reduction in obese subjects.67
Accordingly, this study compared the effects of68
an 8-h time restricted feeding regimen versus a no-69
intervention historical control group on body weight70
and metabolic disease risk factors in obese adults.
71
We hypothesized that the time restricted feeding
72
group would lose weight and improve blood pressure,73
plasma lipids, and glucoregulatory factors versus
74
control subjects after 12 weeks of treatment.75
2. Methods76
2.1. Time restricted feeding subject selection77
A 12-week trial was implemented to test the effects78
of time restricted feeding versus matched histori-
79
cal controls on body weight and metabolic disease
80
risk factors. The experimental protocol was approved
81
by the University of Illinois Chicago Office for the
82
Protection of Research Subjects, and all research83
participants gave their written informed consent to84
participate in the trial. Time restricted feeding sub-85
jects were recruited from September 2016 to January
86
2017 from the Chicago area via advertisements87
placed around the University of Illinois Chicago88
campus. A total of 40 subjects were consented and
89
assessed for eligibility to participate in the time90
restricted feeding intervention (Fig. 1). Of these 4091
subjects, 11 subjects were excluded because they did92
not meet one or more inclusion criteria, and 6 subjects93
declined to participate after qualifying. Inclusion cri- 94
teria was as follows: BMI between 30 and 45kg/m2;95
age between 25 and 65 years; pre-menopausal or 96
post-menopausal (absence of menses for more than 2 97
years); sedentary to lightly active (<7500 steps/d); 98
weight stable for 3 months prior to the beginning 99
of the study (<4 kg weight loss or weight gain); 100
non-diabetic; no history of cardiovascular disease 101
(myocardial infarction or stroke); non-smoker; not 102
a shift worker; and not taking weight loss, lipid- or 103
glucose-lowering medications. 104
2.2. Historical control subject selection 105
Outcome measures were compared to a matched 106
historical control group from a previous weight loss 107
trial performed by our group from October 2011 to 108
January 2015 [10] (Fig. 1). This previous trial imple- 109
mented similar inclusion and exclusion criteria as the 110
present trial, and all control subjects were recruited 111
from the Chicago area via advertisements placed 112
around the University of Illinois Chicago campus. 113
There were 31 control subjects in the previous trial 114
[10]. A stratified random sampling protocol based 115
on age, BMI, and sex was used to match the his- 116
torical control group (n= 23) to that of the time 117
restricted feeding group (n= 23). Subjects were ran- 118
domly selected from each consecutive strata, one-by 119
one, until the sample size of 23 was reached. 120
2.3. Baseline period 121
Before commencing the study, both groups par- 122
ticipated in a 2-week baseline weight stabilization 123
period. All subjects were requested to remain weight 124
stable by consuming their usual diet and not chang- 125
ing their physical activity habits. Body weight was 126
measured weekly at the research center during this 127
period. 128
2.4. Time restricted feeding group protocol 129
Time restricted feeding subjects were instructed to 130
eat ad libitum from 10:00 to 18:00 h daily, and fast 131
from 18:00 to 10:00 h daily. During the 8-h feeding 132
window, there were no restrictions on types or quanti- 133
ties of foods consumed. Moreover, subjects were not 134
required to monitor caloric intake during ad libitum 135
feeding period. During the fasting period, subjects 136
were encouraged to drink plenty of water and were 137
permitted to consume energy-free beverages, such 138
Uncorrected Author Proof
K. Gabel et al. / Time restricted feeding for weight loss 3
Fig. 1. Study flow chart.
black tea, coffee, and diet sodas. Adherence to the 8-h139
time restricted feeding window was measured using140
a daily adherence log, which recorded the times each141
subject started and stopped eating each day. If the log142
indicated that the subject ate within the 8-h window
143
(10:00 to 18:00 h), that day was labeled “adherent”. If144
the log indicated that the subject consumed food out-
145
side of the 8-h feeding window, that day was labeled
146
as “non-adherent”. Adherence to the time restricted
147
feeding diet was assessed as the number of adherent148
days per week.149
2.5. Historical control group protocol150
Historical controls were selected from a previ-151
ous weight loss trial [10] performed by our group152
from 2011–2015. Controls were instructed to main-
153
tain their weight throughout the trial, and not to
154
change their eating or physical activity habits. Con-155
trols visited the research center on a weekly basis for156
weigh-ins. Body composition and metabolic disease157
risk variables were assessed in control subjects every158
12 weeks.
159
2.6. Dietary intake and physical activity
160
Time restricted feeding and control subjects com-
161
pleted a 7-d food record (on 7 consecutive days)162
during the baseline period and week 12. At base-163
line, a dietitian provided 15 min of instruction to each 164
participant on how to complete the food records. 165
These instructions included information and refer- 166
ence guides on how to estimate portion sizes and 167
record food items in sufficient detail to obtain accu- 168
rate estimates of dietary intake. Subjects were not 169
required to weigh foods but were asked to measure 170
the volume of foods consumed with household mea- 171
sures (i.e. measuring cups and measuring spoons). 172
The timing of food intake was also recorded. Food 173
records were collected at the weigh-in at baseline and 174
week 12, and were reviewed by the dietitian for accu- 175
racy and completeness. The food analysis program, 176
Nutritionist Pro (Axxya Systems, Stafford, TX) was 177
used to calculate the total daily intake of energy, fat, 178
protein, carbohydrate, cholesterol, and fiber. All sub- 179
jects were asked to maintain their level of physical 180
activity throughout the entire trial. We assessed activ- 181
ity level as steps/d. Step counts were measured over 182
7-d during the baseline period and at week 12 by 183
a pedometer (Yamax Digi-walker SW-200, Yamax 184
Inc., San Antonio, TX). 185
2.7. Outcome measures 186
2.7.1. Body weight and body composition 187
The primary outcome measure was body weight. 188
Body weight was assessed to the nearest 0.25 kg every 189
week without shoes and in light clothing using a bal- 190
ance beam scale (HealthOMeter, Boca Raton, FL). 191
Uncorrected Author Proof
4K. Gabel et al. / Time restricted feeding for weight loss
Height was assessed during the screening visit using192
a wall-mounted stadiometer (HealthOMeter, Boca
193
Raton, FL) to the nearest 0.1 cm. BMI was assessed194
as kg/m2. Body composition (fat mass, lean mass,195
visceral fat mass) was measured using dual x-ray196
absorptiometry (DXA; iDXA, General Electric Inc).
197
2.7.2. Metabolic disease risk factors
198
Blood pressure and heart rate were measured at
199
baseline and week 12 in triplicate using a digital200
automatic blood pressure/heart rate monitor (Omron201
HEM 705 LP, Kyoto, Japan) with the subject in a202
seated position after a 10-min rest. Twelve-h fast-203
ing blood samples were collected between 5:00 and204
9:00 h at baseline and week 12. The subjects were
205
instructed to avoid exercise, alcohol, and coffee for206
24 h before each visit. Blood was centrifuged for207
10 min at 520 Í g at 4◦C to separate plasma from208
red blood cells and was stored at –80◦C until ana-209
lyzed. Fasting plasma total cholesterol, direct LDL210
cholesterol, HDL-cholesterol, triglycerides concen-
211
trations were measured by a commercial lab (Alverno
212
Laboratories, Hammond, IN). Fasting glucose con-213
centrations were measured with a hexokinase reagent214
kit (Abbott, South Pasadena, CA). Fasting insulin was215
assessed as total immunoreactive insulin (Coat-A-
216
Count Insulin, Los Angeles, CA). Insulin resistance
217
(IR) was calculated using the HOMA (Home-
218
ostasis Model Assessment) method, by applying
219
the following formula: [HOMA-IR = Fasting insulin220
(lU/ml) ×Fasting glucose (mg/dL) / 405]. Plasma221
homocysteine was quantified using HPLC with fluo-222
rometric detection.223
2.8. Statistical analyses
224
All data are presented as mean ±standard error
225
of the mean (SEM). Statistical analyses were per-
226
formed using SPSS 24.0 for Windows (SPSS Inc.).
227
A two-tailed P value of less than 0.05 was consid-228
ered statistically significant. Tests for normality were
229
included in the model, and all data were found to be230
normally distributed. For the sample size calculation,
231
we estimated that time restricted feeding and con-232
trol group would reduce body weight by 3% and 0%,233
respectively, by week 12. We calculated that n=19234
participants per group would provide 80% power to235
detect a significant difference of 3% in body weight236
between the time restricted feeding and control group237
at week 12, using a two-tailed independent-samples
238
t-test with ␣= 0.05. We anticipated a dropout rate of239
20%. Thus, we aimed to recruit n= 23 subjects in the 240
time restricted feeding group, assuming that n=19 241
would complete the trial. 242
Differences between the time restricted feeding 243
and control groups at baseline were analyzed by an 244
independent samples t-test (continues variables) or 245
the McNemar test (categorical variables). Data were 246
included for 46 participants, and means were esti- 247
mated using an intention-to-treat analysis using last 248
observation carried forward. Repeated measures two- 249
factor ANOVA with groups (time restricted feeding 250
and control) as the between-subject factor and time 251
(week 1 and 12) as the within-subject factor was used 252
to compare changes in dependent variables between 253
the groups over time. When there was a significant 254
main effect but no interaction, post hoc compar- 255
isons were performed using Bonferroni’s correction 256
to determine differences between group means. 257
3. Results 258
3.1. Baseline characteristics and dropouts 259
As portrayed in Fig. 1, n= 23 subjects began the 260
time restricted feeding intervention and 6 dropped 261
out. No subjects reported dropping out of the time 262
restricted feeding group due to issues with the diet. At 263
baseline, there were no statistically significant differ- 264
ences between the time restricted feeding group and 265
the historical control group for age, sex, ethnicity, 266
weight, height or BMI (Table 1). 267
Table 1
Baseline characteristics1
Time restricted Control Pvalue2
feeding
N2323
Age (y) 50 ±248±2 0.43
Sex (F/M) 20/3 21/2 0.82
Ethnicity 0.69
African American 17 17
Caucasian 3 5
Hispanic 3 1
Weight (kg) 95±392±3 0.34
Height (m) 1.66 ±0.02 1.63 ±0.01 0.29
BMI (kg/m2)35±134±1 0.61
1All values reported as mean ±SEM. 2Pvalues for comparison
of baseline variables between groups using independent samples
t-test (continuous variables) and the McNemar test (categorical
variables).
Uncorrected Author Proof
K. Gabel et al. / Time restricted feeding for weight loss 5
Fig. 2. Weekly adherence to the 8-hour feeding window by the
time restricted feeding group. All values reported as mean ±SEM.
Each bar indicates the mean number of days per week that the
time restricted feeding subjects were compliant with the 8-hour
feeding window.On average, the time restricted feeding group was
compliant with the prescribed eating window (10:00 to 18:00 h) on
5.6 ±0.3 d/week, and this level of adherence did not change over
the course of the trial (P= 0.86, repeated measures ANOVA).
3.2. Adherence to the time restricted feeding268
window269
Subjects in the time restricted feeding group were
270
compliant with the prescribed eating window (10:00271
to 18:00 h) on 5.6 ±0.3 d/week (Fig. 2), and this level272
of adherence did not change over the course of the
273
trial (P= 0.86). At baseline, the time restricted feed-274
ing and control groups reported a similar mean daily275
eating duration of 11 ±1 h (Table 2). By week 12,276
mean daily eating duration in the time restricted feed-277
ing group was significantly shorter (8 ±1 h) than that278
of the control group (11 ±1 h) (group ×time inter-
279
action, P= 0.01). The start of the eating duration in
280
the time restricted feeding group was later than that
281
of the control group during the study (group ×time282
interaction, P= 0.01), and the end of the eating dura-
283
tion was similar between the groups during the study
284
(no group ×time interaction, P= 0.32).285
3.3. Dietary intake and physical activity
286
At baseline, energy intake was similar in the time287
restricted feeding and control groups (Table 2). Dur-288
ing the trial, energy intake decreased in the time289
restricted feeding group by 341 ±53 kcal/d relative290
to controls (group ×time interaction, P= 0.04). Self-291
reported intake of macronutrients, dietary cholesterol
292
and fiber did not differ between groups at baseline or
293
post-treatment. Activity level, measured as steps/d,
294
was similar at baseline in the time restricted feed- 295
ing and control groups, and did not change over the 296
course of the trial in either group. 297
3.4. Body weight and body composition 298
Body weight remained stable during the 2-week 299
baseline period (Fig. 3). Body weight decreased in 300
the time restricted group (–2.6% ±0.5) relative to 301
controls during the 12-week study (group ×time 302
interaction, P< 0.001). BMI decreased in the time 303
restricted feeding group relative to the control group 304
during the trial (group ×time interaction, P< 0.001) 305
(Table 3). There were no statistically significant dif- 306
ferences (i.e. no group ×time interaction) between 307
groups for fat mass, lean mass or visceral fat mass 308
(Table 3). 309
3.5. Metabolic disease risk factors 310
At baseline, there were no differences between 311
groups for any metabolic disease risk factor (Table 3). 312
Systolic blood pressure significantly decreased in 313
the time restricted feeding group (–7 ±2 mm Hg) 314
relative to controls during the study (group×time 315
interaction, P= 0.02). There were no statistically sig- 316
nificant differences (i.e. no group ×time interaction) 317
between groups for diastolic blood pressure, heart 318
rate, total cholesterol, LDL cholesterol, HDL choles- 319
terol, triglycerides, glucose, insulin, HOMA-IR, or 320
homocysteine. 321
4. Discussion 322
This study is the first trial to examine the impact of 323
time restricted feeding on body weight and metabolic 324
disease risk factors in an obese population. We show 325
here that 12 weeks of 8-h time restricted feed- 326
ing (i.e. limiting food intake to 10:00 to 18:00 h 327
daily) decreases body weight by ∼3% relative to 328
a no-intervention historical control group. We also 329
demonstrate that this fasting regimen produces sig- 330
nificant reductions in systolic blood pressure relative 331
to controls. 332
Adherence to the time restricted feeding window 333
was assessed daily via self-report. Our findings show 334
that subjects were adherent to the prescribed eating 335
window on ∼6 days per week, and that this level 336
of adherence remained constant throughout the 12- 337
week trial. On average, the time restricted feeding 338
Uncorrected Author Proof
6K. Gabel et al. / Time restricted feeding for weight loss
Table 2
Self-reported dietary intake, eating duration, and physical activity at baseline and week 12 1
Time restricted feeding (n= 23) Control (n= 23) P-value Time ×group 3
Baseline 2Week 12 Baseline 2Week 12
Energy (kcal) 1676 ±114 1335 ±162 1645 ±113 1654 ±191 0.04
Protein (%) 16 ±117±117±117±1 0.40
Carbohydrates (%) 47 ±246±246±245±2 0.61
Fat (%) 37 ±137±237±138±2 0.74
Cholesterol (mg) 279 ±24 214 ±27 275 ±27 265 ±37 0.32
Fiber (g) 16 ±213±114±115±2 0.17
Daily eating duration (h) 11 ±18±111±111±10.01
Start of eating duration (local time, h) 8:30 ±0:30 10:00 ±0:30 9:30 ±0:30 8:30 ±0:30 0.01
End of eating duration (local time, h) 19:30 ±0:30 18:00 ±0:30 20:30 ±0:30 19:30 ±0:30 0.32
Steps/day 6896 ±723 7443 ±880 6148 ±775 6967 ±584 0.84
1All values reported as mean ±SEM. Data for all variables were collected over a 7-d period at baseline (prior to the commencement of the
study) and week 12 in the TRF and control groups. Data were included for 46 participants; means were estimated using an intention-to-treat
analysis using last observation carried forward. 2Baseline variables: No difference between groups for any parameter (Independent samples t-
test). 3Pvalues reported for the time restricted feeding group relative to the control group (group ×time interaction) using repeated-measures
2-factor ANOVA.
Fig. 3. Weight loss by the time restricted feeding group versus con-
trols 1.1All values reported as mean ±SEM. Data were included
for 46 participants; means were estimated using an intention-to-
treat analysis using last observation carried forward. Body weight
remained stable during the 2-week baseline period (week B1 and
week B2). Body weight decreased in the time restricted feeding
group relative to controls during the 12-week intervention period
(P< 0.001 for time ×group interaction).
subjects reduced their daily eating duration by 3 h/d,339
i.e. from 11 h/d to 8h/d. The dropout rate in the time340
restricted feeding group (26%) was high for a short-341
term trial. However, no one in the time restricted
342
feeding group reported dropping out due to issues
343
with the diet. These preliminary findings suggest that344
time restricted feeding may be somewhat well toler-
345
ated over short periods in obese subjects.346
Our findings also indicate that reducing the daily347
eating window to 8-h/d decreases caloric intake by
348
∼300 kcal/d, without intentional calorie counting. As
349
a result of this daily energy deficit, subjects in the time 350
restricted feeding group lost ∼3% of body weight 351
over 12 weeks, versus controls. The degree of weight 352
loss demonstrated here is less than what was achieved 353
in the 10-h time restricted feeding study (4%) [11]. 354
This may be due, in part, to the shorter trial dura- 355
tion (3 months) implemented here, when compared 356
to the 10-h time restricted feeding study (4 months) 357
[11]. Degree of energy restriction was also lower in 358
the present study (∼300 kcal/d deficit), when com- 359
pared to that of the 10-h time restricted feeding study 360
(∼400 kcal/d deficit) [11]. Our results, however, are 361
similar to the 4-8-h time restricted feeding trials 362
which produced ∼1–3% body weight reductions [12, 363
13]. 364
In comparison to other forms of intermittent fast- 365
ing [4–10], time restricted feeding appears to produce 366
less weight loss. For instance, after 12 weeks of alter- 367
nate day fasting, body weight typically decreases by 368
4–6% from baseline [4–10]. We speculate that this 369
difference in weight loss is due to greater overall 370
caloric restriction achieved with other forms of inter- 371
mittent fasting, versus time restricted feeding. Accu- 372
mulating evidence suggests that alternate day fasting 373
produces an average caloric deficit of 25–35% daily 374
[10, 16], whereas time restricted feeding may only 375
produce a 20% caloric deficit daily [11]. The greater 376
degree of energy restriction achieved with alternate 377
day fasting is most likely the result of the vigilant 378
calorie counting on fast days. Since time restricted 379
feeding does not require subjects to monitor calorie 380
intake at all, this may explain why the average caloric 381
deficit achieved with time restricted feeding is lower. 382
Uncorrected Author Proof
K. Gabel et al. / Time restricted feeding for weight loss 7
Table 3
Body composition and metabolic disease risk factors after 12 weeks1
Time restricted feeding (n= 23) Control (n= 23) P-value Time ×group3
Baseline2Week 12 Baseline2Week 12
Body weight (kg) 95 ±392±392±392±3<0.001
Fat mass (kg)442 ±240±237±237±2 0.23
Lean mass (kg) 50 ±250±253±253±2 0.12
Visceral fat mass (kg) 1.2 ±0.1 1.1 ±0.1 1.2 ±0.2 1.2 ±0.2 0.19
BMI (kg/m2)35±134±134±134±1<0.001
Systolic blood pressure (mm Hg) 128 ±4 121±3 123 ±4 124 ±30.02
Diastolic blood pressure (mm Hg) 83 ±282±281±282±2 0.41
Heart rate (bpm) 69 ±271±273±273±3 0.33
Total cholesterol (mg/dl) 177±7 178 ±9 192 ±7 185 ±7 0.15
LDL cholesterol (mg/dl) 108 ±5 110 ±7 114±7 112 ±6 0.54
HDL cholesterol (mg/dl) 48 ±249±261±355±2 0.11
Triglycerides (mg/dl)4105 ±11 93 ±989±789±11 0.43
Fasting glucose (mg/dl) 79±482±287±287±2 0.77
Fasting insulin (uIU/ml)48.3 ±1.0 5.7 ±0.7 9.2 ±1.4 10.3 ±1.9 0.16
HOMA-IR41.6 ±0.2 1.0 ±0.2 2.0 ±0.3 2.2 ±0.4 0.21
Homocysteine (mol/l)49.9 ±0.6 9.0 ±0.5 10.1 ±0.5 9.4 ±0.5 0.83
1All values reported as mean ±SEM. Data were included for 46 participants; means were estimated using an intention-to-treat analysis using
last observation carried forward. HOMA-IR: Homeostatic model assessment Insulin resistance; RMR: Resting metabolic rate. 2Baseline
variables: No difference between groups for any parameter (Independent samples t-test). 3Pvalues reported for the time restricted feeding
group relative to the control group (group×time interaction) using repeated-measures 2-factor ANOVA. 4Significant main effect of time,
P< 0.05.
One potential confound in the current study is the383
lack of an objective measure to assess eating dura-
384
tion. As time restricted feeding is a recent concept,
385
methods to objectively record eating time have yet
386
to be optimized. Self-reporting of eating duration, as
387
used in the current study, may not be optimal. An app-388
based recording of all eating events was used in the389
10-h time restricted feeding study, and found a self-
390
reporting error of 10% [11]. Objective methods of391
recording eating events show a mean eating duration
392
that is different from what is widely believed [11].393
A study assessing eating pattern among non-shift
394
worker adults found the median daily eating duration
395
can be 15-h or longer, and less than 15% of adults
396
eat for less than a 12-h duration [11]. In comparison,
397
in the present study, the self-reported baseline eating398
duration was 11 h, which is likely inaccurate. Our399
study also permitted the consumption of low energy400
drinks including coffee, tea, and diet soda. These401
drinks contain caffeine, which is known to perturb402
circadian rhythm [17]. Since time restricted feeding403
is based on the principle of circadian rhythm regu-
404
lation of metabolism, low-energy caffeinated drinks405
may not count significantly towards energy consump-406
tion, but can have significant impact on circadian407
regulation.408
Future trials in this area can be improved by using409
objective measures to better assess daily eating dura-410
tions. It will also be of interest to investigate whether 411
shorter eating windows (4–6 h) produce a degree of 412
weight loss that is comparable to that of alternate day 413
fasting and 5:2. Moreover, how the placement of the 414
feeding window influences weight loss and adher- 415
ence will also be important to examine. We chose to 416
prescribe a feeding window of 10:00 to 18:00 h, so 417
that the intervention would be standardized across all 418
subjects. We assumed that this window would pro- 419
duce maximal adherence, as it would cause minimal 420
disruption to the typical eating schedule (i.e. sub- 421
jects could still have their breakfast in the morning, 422
lunch in the early afternoon, and dinner in the early 423
evening). However, recent trials [18, 19] have found 424
that consuming larger meals earlier in the day produce 425
better weight loss than similar sized meals consumed 426
later in the evening. Whether weight loss and adher- 427
ence can be improved by shifting the eating window 428
earlier in the day warrants investigation. 429
Metabolic disease risk indicators remained rel- 430
atively unaffected by the time restricted feeding 431
regimen. Systolic blood pressure was the only 432
parameter that improved over the course of the 433
study, relative to controls. Reductions in insulin, 434
insulin resistance, triglycerides, and homocysteine 435
were also observed over time, but these effects were 436
not statistically different from the control group. 437
In the study by Moro et al. [12], plasma lipids 438
Uncorrected Author Proof
8K. Gabel et al. / Time restricted feeding for weight loss
and inflammatory factors also remained unchanged439
with 8-h time restricted feeding. It is likely that the
440
degree of weight loss produced by 8-h time restricted441
feeding was not large enough to improve these442
outcome measures. Accumulating evidence suggests443
that > 5% weight loss is required to improve plasma
444
lipid concentrations and glucoregulatory factors [20].445
It should also be noted that the obese subjects in the446
present study were metabolically healthy at baseline,447
i.e. their blood pressure, plasma lipid, glucose, and
448
insulin levels were all within the normal range.449
Previous work indicates that intermittent fasting450
regimens [4, 8, 10] and other lifestyle regimens [21,
451
22] have little effect on cardiometabolic disease452
risk factors in healthy obese subjects. It will be of
453
interest to examine whether time restricted feeding454
can improve these risk factors in other groups of455
obese patients, such as those with compromised
456
insulin sensitivity or dyslipidemia [23, 24].457
This study has several limitations. First, the study458
was not a randomized controlled trial. We compared459
the effects of time restricted feeding to a matched his-460
torical control group from a previous weight loss trial461
conducted by our group. The trial [10] from which462
the controls were selected was conducted between
463
2011–2015. As such, the lapse of up to five years
464
between trials could have influenced what the con-465
trol subjects knew about weight control, and what
466
foods were available in the marketplace due to sea-467
sonality. These issues should be considered when468
interpreting the present findings. In order to truly
469
determine the effect of time restricted feeding on470
body weight and other metabolic disease variables,
471
future trials should implement a randomized design
472
where controls are enrolled concurrently. Second, the473
study was quite short (12 weeks). Longer-term tri-474
als will be needed to determine the degree of weight475
loss that can be achieved with time restricted feeding.476
Third, adherence and dietary intake were assessed477
by self-report, thus our estimates of eating dura-
478
tion and caloric deficit may be inaccurate [25, 26].479
Implementing mobile apps to assess adherence in
480
real-time [11] will help determine how well subjects481
can adhere to the prescribed eating window. Fourth,482
subjects completed a 7-d food record during the base-483
line period, which may have influenced their eating484
behaviors [27]. Recent findings suggest that subjects485
are more likely to report foods that are considered486
healthy and socially desirable, and omit foods that
487
are considered unhealthy, when completing study-
488
related food records [27]. Fifth, our study involved
489
only metabolically healthy obese subjects, so our
490
findings cannot be generalized to other population 491
groups. 492
In summary, these findings suggest that 8-h time 493
restricted feeding produces mild caloric restriction 494
and weight loss in obese adults, without intentional 495
calorie counting. This diet may also offer some clini- 496
cal benefit by reducing systolic blood pressure. These 497
preliminary data offer promise for the use of time 498
restricted feeding as a weight loss technique in obese 499
adults, but longer-term, larger-scale randomized con- 500
trolled trials will be required before solid conclusions 501
can be reached. 502
Acknowledgments 503
KG designed the research, conducted the clinical 504
trial, analyzed the data, performed the statistical anal- 505
ysis, and wrote the manuscript; KKH, NH, JS, CMK, 506
and JFT assisted with the conduction of the clinical 507
trial; SP assisted with data interpretation and wrote 508
the manuscript; KAV designed the research, analyzed 509
the data, and wrote the manuscript. 510
Funding 511
University of Illinois Chicago Campus Research 512
Board Pilot Grant; National Institutes of Health 513
R01HL106228, F32DK107157, T32HL007909. 514
Trial registration 515
Clinicaltrials.gov NCT02948517. 516
Disclosure 517
Dr. Varady received author fees from Hachette 518
Book Group for the book, “The Every Other Day 519
Diet”. Dr. Panda received author fees from Rodale 520
Books for the upcoming book, “The Circadian Code”. 521
The other authors have no conflicts of interest to 522
report. 523
References 524
[1] Longo VD, Mattson MP.Fasting: Molecular mechanisms and 525
clinical applications. Cell Metab. 2014;19(2):181-92. 526
Uncorrected Author Proof
K. Gabel et al. / Time restricted feeding for weight loss 9
[2] Longo VD, Panda S. Fasting, Circadian Rhythms, and
527
Time-Restricted Feeding in Healthy Lifespan. Cell Metab.528
2016;23(6):1048-59.529
[3] Chaix A, et al. Time-restricted feeding is a preventative
530
and therapeutic intervention against diverse nutritional chal-531
lenges. Cell Metab. 2014;20(6):991-1005.532
[4] Bhutani S, et al. Alternate day fasting and endurance exer-533
cise combine to reduce body weight and favorably alter
534
plasma lipids in obese humans. Obesity (Silver Spring).
535
2013;21(7):1370-9.
536
[5] Catenacci VA, et al. A randomized pilot study com-
537
paring zero-calorie alternate-day fasting to daily caloric538
restriction in adults with obesity. Obesity (Silver Spring).539
2016;24(9):1874-83.540
[6] Harvie MN, et al. The effects of intermittent or continuous541
energy restriction on weight loss and metabolic disease risk
542
markers: A randomized trial in young overweight women.543
Int J Obes (Lond). 2011;35(5):714-27.
544
[7] Heilbronn LK, et al. Alternate-day fasting in nonobese sub-
545
jects: Effects on body weight, body composition, and energy546
metabolism. Am J Clin Nutr. 2005;81(1):69-73.547
[8] Hoddy KK, et al. Meal timing during alternate day fasting:548
Impact on body weight and cardiovascular disease risk in549
obese adults. Obesity (Silver Spring). 2014;22(12):2524-31.550
[9] Klempel MC, Kroeger CM, Varady KA. Alternate day fasting551
(ADF) with a high-fat diet produces similar weight loss and
552
cardio-protection as ADF with a low-fat diet. Metabolism.553
2013;62(1):137-43.554
[10] Trepanowski JF, et al. Effect of Alternate-Day Fasting on555
Weight Loss, Weight Maintenance, and Cardioprotection
556
Among Metabolically Healthy Obese Adults: A Randomized557
Clinical Trial. JAMA Intern Med. 2017.558
[11] Gill S, Panda S. A Smartphone App Reveals Erratic Diurnal559
Eating Patterns in Humans that Can Be Modulated for Health560
Benefits. Cell Metab. 2015;22(5):789-98.561
[12] Moro T, et al. Effectsof eight weeks of time-restricted feeding562
(16/8) on basal metabolism, maximal strength, body com-
563
position, inflammation, and cardiovascular risk factors in
564
resistance-trained males. J Transl Med. 2016;14(1):290.
565
[13] Tinsley GM, et al. Time-restricted feeding in young men per-566
forming resistance training: A randomized controlled trial.
567
Eur J Sport Sci. 2017;17(2):200-07.568
[14] Jensen MD, et al. 2013 AHA/ACC/TOS guideline for the569
management of overweight and obesity in adults: A report of570
the American College of Cardiology/American Heart Asso-571
ciation Task Force on Practice Guidelines and The Obesity572
Society. J Am Coll Cardiol. 2014;63(25 Pt B):2985-3023.573
[15] Magkos F, et al. Effects of Moderate and Subsequent 574
Progressive Weight Loss on Metabolic Function and Adi- 575
pose Tissue Biology in Humans with Obesity. Cell Metab. 576
2016;23(4):591-601. 577
[16] Klempel MC, et al. Dietary and physical activity adaptations 578
to alternate day modified fasting: Implications for optimal 579
weight loss. Nutr J. 2010;9:35. 580
[17] Burke TM, et al. Effects of caffeine on the human 581
circadian clock in vivo and in vitro. Sci Transl Med. 582
2015;7(305):305ra146. 583
[18] Garaulet M, et al. Timing of food intake predicts weight loss 584
effectiveness. Int J Obes (Lond). 2013;37(4):604-11. 585
[19] Jakubowicz D, et al. High caloric intake at breakfast vs. 586
dinner differentially influences weight loss of overweight 587
and obese women. Obesity (Silver Spring). 2013;21(12): 588
2504-12. 589
[20] Ryan DH, Yockey SR. Weight Loss and Improvement in 590
Comorbidity: Differences at 5%, 10%, 15%, and Over. Curr 591
Obes Rep. 2017;6(2):187-94. 592
[21] Janiszewski PM, Ross R. Effects of weight loss among 593
metabolically healthy obese men and women. Diabetes Care. 594
2010;33(9):1957-9. 595
[22] Kantartzis K, et al. Effects of a lifestyle intervention in 596
metabolically benign and malign obesity. Diabetologia. 597
2011;54(4):864-8. 598
[23] Barnosky AR, et al. Intermittent fasting vs daily calorie 599
restriction for type 2 diabetes prevention: A review of human 600
findings. Transl Res. 2014;164(4):302-11. 601
[24] Harvie M, Howell A. Potential Benefits and Harms of Inter- 602
mittent Energy Restriction and Intermittent Fasting Amongst 603
Obese, Overweight and Normal Weight Subjects-A Narrative 604
Review of Human and Animal Evidence. Behav Sci (Basel). 605
2017;7(1). 606
[25] Black AE, Cole TJ. Within- and between-subject varia- 607
tion in energy expenditure measured by the doubly-labelled 608
water technique: Implications for validating reported dietary 609
energy intake. Eur J Clin Nutr. 2000;54(5):386-94. 610
[26] Lichtman SW, et al. Discrepancy between self-reported and 611
actual caloric intake and exercise in obese subjects. N Engl 612
J Med. 1992;327(27):1893-8. 613
[27] Ortega RM, P.-R.C., Lopez-Sobaler AM. Dietary assessment 614
methods: Dietary records. Nutr Hosp. 2015;31:38-45. 615
