Research: Treatment: B. T. Corley, R. W. Carroll, R. M. Hall, M. Weatherall, A. Parry-Strong and J. D. Krebs
Research: Treatment: B. T. Corley, R. W. Carroll, R. M. Hall, M. Weatherall, A. Parry-Strong and J. D. Krebs
Research: Treatment: B. T. Corley, R. W. Carroll, R. M. Hall, M. Weatherall, A. Parry-Strong and J. D. Krebs
Accepted Article
Article type : Research Article
Research: Treatment
and J. D. Krebs1,2
1
Centre for Endocrine Diabetes and Obesity Research, Wellington Hospital and 2Department
What's new?
This article has been accepted for publication and undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process, which may
lead to differences between this version and the Version of Record. Please cite this article as
doi: 10.1111/dme.13595
This article is protected by copyright. All rights reserved.
Intermittent '5:2' diets, 2 days of very-low-calorie diet per week with ad libitum intake
on the other 5 days, have popular appeal but implementation in people with diabetes
Accepted Article
can be daunting in view of the risk of hypoglycaemia caused by changing
overall risk of hypoglycaemia. Our protocol can be adopted for future studies on the
Abstract
Aims To establish whether the risk of hypoglycaemia is greater with 2 consecutive days of
intermittent fasting in adults. The participants had a BMI of 30–45 kg/m2, Type 2 diabetes
50–86 mmol/mol (6.7–10%). The participants followed a 2092–2510-kJ diet on 2 days per
week for 12 weeks. A total of 41 participants were randomized 1:1 to consecutive (n=19) or
non-consecutive (n=22) day fasts, of whom 37 (n=18 and n=19, respectively) were included
between the two study arms. Secondary outcomes included change in diet, quality of life,
Accepted Article
weight, lipid, glucose and HbA1c levels, and liver function.
Results The mean hypoglycaemia rate was 1.4 events over 12 weeks. Fasting increased the
rate of hypoglycaemia despite medication reduction (relative rate 2.05, 95% CI 1.17 to 3.52).
There was no difference between fasting on consecutive days and fasting on non-consecutive
days (relative rate 1.54, 95% CI 0.35 to 6.11). Improvements in weight, HbA1c, fasting
any type increased the rate of hypoglycaemia. With education and medication reduction,
fewer than expected hypoglycaemic events occurred. Although it was not possible to
Introduction
With rising rates of obesity and Type 2 diabetes mellitus worldwide [1,2], there is a need for
accessible, safe and cost-effective treatments for both conditions. A very-low-calorie diet can
facilitate weight loss and improve glucose homoeostasis [3,4]; however, in those taking
hypoglycaemic medication, such calorie restriction increases the risk of hypoglycaemia and
caloric restriction on some days combined with ad libitum calorie intake on others. The
degree of caloric restriction may vary from partial to complete restriction. The schedule may
involve restriction for several hours a day, alternate days or several days per week [5,6]. A
systematic review, but without meta-analysis, of studies comparing intermittent fasting with
options in nine out of 12 studies (75%). The individual study results, however, were more
Accepted Article
variable with regard to differences in HbA1c levels, fasting glucose levels and markers of
insulin sensitivity for intermittent fasting compared with daily energy restriction [7]. Some of
this variation may have been attributable to differences in the prescribed diet on fasting days
and the schedule of fasting. An issue yet to be investigated regarding intermittent fasting is
medication. There have been four studies to date of intermittent fasting in participants with
Type 2 diabetes [8–10] and in only two of these were participants taking medication with the
A popular form of intermittent fasting is the '5:2' schedule. This involves a very-low-calorie
diet (2092 kJ in women and 2510 kJ in men) for 2 days per week, with ad libitum intake on
the other 5 days. The best approach to medication adjustment to avoid hypoglycaemia on
fasting days in those with Type 2 diabetes undertaking intermittent fasting is not known.
Furthermore, it is not known whether consecutive days of fasting compared with non-
The aim of the present study, therefore, was to test the hypothesis that during a 5:2
non-consecutive days of caloric restriction along with medication adjustment would reduce
the overall risk of hypoglycaemia to a greater extent than consecutive days of caloric
restriction.
Participants aged >18 years with Type 2 diabetes who were taking medication for diabetes,
including metformin and/or any combination of hypoglycaemic agents, and who had an
30–45 kg/m2, were recruited from secondary care diabetes clinics, local community networks,
Accepted Article
and primary care practices. Exclusion criteria were: Type 1 diabetes; weight change of >5 kg
blood pressure > 180/100 mmHg despite medical therapy; previous bariatric surgery; and any
significant medical condition which, in the view of study investigators, would make
recruitment to the study inappropriate. Each participant provided written informed consent.
The study was approved by the New Zealand Health and Disability Ethics Committee
(14/NTB/33/AM03), was performed in accordance with the Declaration of Helsinki, and was
(ACTRN12614000402640). The final study protocol differed from the registry data as
follows: inclusion criteria for HbA1c range differs from that in the trial registry, which was
64–86 mmol/mol (8.0–10.0%), power calculations were based on hypoglycaemia event rate,
the diet intervention was based on a prescription of 2092 kJ in women and 2510 kJ in men,
and food diaries were gathered at 6 weeks as well as at baseline and 12 weeks.
All participants attended the Centre for Endocrine, Diabetes and Obesity Research,
Wellington Hospital, New Zealand on three occasions over 12 weeks: at baseline and at 6 and
12 weeks between August 2014 and November 2015. At baseline participants were
consecutive days of fasting on 2 days per week. Participants were free to choose which day of
the week to fast; this could vary from week to week to allow flexibility and improve
adherence. Treatment allocations were printed and placed in sequentially numbered sealed
envelopes by a member of the research team prior to enrolment of the first participant in the
study. The allocation was concealed from the staff member conducting enrolment and
Participants were given 9 days of written sample recipes with small variations between the
Accepted Article
energy content of each option. These were developed by a research dietician and structured as
two small snacks and one light meal, amounting to between 2092 and 2510 kJ per day, with
men instructed to consume approximately 400 kJ more than women. Each participant was
provided with written and verbal information about symptoms, management and common
follows: sulfonylureas and isophane insulin (NPH insulin) was reduced by 50% on fasting
days. Insulin lispro (Humalog), insulin aspart (Novorapid), insulin glulisine (Apidra), and
regular insulin (Humulin R) were reduced by 70% on fasting days, mixed insulins were
reduced by 25% on the night before a fast and 50% on the day of a fast, insulin glargine
(Lantus) was reduced by 50% on the morning of a fasting day and/or by 50% on the evening
before a fasting day. Doses of metformin or other medication not resulting in hypoglycaemia
were unchanged.
Hypoglycaemic events
The primary outcome was the total number of hypoglycaemic events during 12 weeks'
observation. A related secondary outcome was the risk of hypoglycaemia on fasting vs non-
fasting days. Participants were contacted weekly by telephone or email during the study.
Days of the week spent fasting, change in medication, the date, time, circumstances, severity
and capillary blood glucose concentration of any hypoglycaemic events were recorded.
Hypoglycaemia was defined as a capillary blood glucose level <4.0 mmol/l. Severe
event requiring the assistance of another person to actively administer carbohydrate, glucagon
the result of a missed meal, appropriate advice was given to prevent a further occurrence and
Accepted Article
hypoglycaemic medications were not adjusted. If there was no clear cause, or if
hypoglycaemic events were recurrent despite appropriate advice, the participants’ medication
Anthropometric measurements and food intake were assessed at all visits and included;
analysis (TBF-300; Tanita Corp., Arlington Heights, IL, USA) and blood pressure (Flexiport;
Welch Allyn Inc., Skaneateles Falls, NY, USA). Participants completed a 4-day food diary to
record calorie intake at baseline, 6 and 12 weeks [12]. At 6 and 12 weeks, food diaries
contained at least 1 fasting day which was identified as such. Quality-of-life and biochemical
The Audit of Diabetes-Dependent Quality of Life 19 (ADDQoL) Questionnaire was used for
and a measure of the impact of diabetes on quality of life in general. It also measures the
impact in 19 different domains. For each domain, participants are asked to rate how their life
would be if they did not have diabetes. The impact scales for each domain range from 'very
much better' (–3) to 'worse' (+1) and are multiplied by an importance score from 0 to +3,
which reflects the importance of that domain to the participant. Lower scores reflect poorer
quality of life. A mean weighted impact score is calculated as a summary score across all
domains.
non-fasting day.
HbA1c, fasting lipids, thyroid-stimulating hormone, free thyroxine, liver function, renal
Fasting venous blood was collected and analysed for lipids, glucose, liver function and renal
function (Cobas c-501; Roche, Basel, Switzerland); free thyroid hormone and thyroid-
stimulating hormone (Cobas e601; Roche); full blood count (XS-1000i, Sysmex, Hyogo,
Statistical analysis
The relative rate of hypoglycaemic events was estimated using a generalized linear mixed
model. Covariates in the model were baseline insulin use, baseline sulfonylurea use and
as random effects. A Poisson distribution for count data was used with the logarithm of the
number of days’ fasting as the offset variable. Anthropometric and biochemical outcomes
were assessed using a general linear model that incorporated baseline values and treatment
arm as fixed effects (Table 2). Quality-of-life data were analysed using two-way ANCOVA,
incorporating treatment arm as a predictor variable and baseline scores as a covariate in the
model. For glucose profiles on fasting and non-fasting days, only recordings that covered
>85% of the day and contained a minimum of three calibrations per day were used.
Participants who dropped out were not included in the primary analysis because of the risk of
underestimating the risk of hypoglycaemia. For secondary analyses, missing data were
analysis.
Accepted Article
Sample size
The sample size was estimated by simulation from two Poisson distributions, one with an
event rate of 4.4 per 12 weeks [15] and one with an event rate of 6.6 per 12 weeks. The aim
was to detect a clinically significant relative rate of events of 1.5 with 80% power and a two-
Results
Baseline demographic data and medication use are shown in Table 1. The flow of
Hypoglycaemia
hypoglycaemic events occurred over 851 fasting days, a crude rate of one event per 37 days
of fasting, and 30 over 2257 non-fasting days, a crude rate of one event per 75 participant-
participants in the consecutive days' fasting arm, with 1512 participant-days of observation
for a crude rate of one event per 43 days. A total of 20 events occurred in eight out of 19
observation for a crude rate of 1 event per 80 days. There were no reported severe
hypoglycaemic events.
2.05 (95% CI 1.17–3.52); P=0.013]. The risk of having a hypoglycaemic event was not
Accepted Article
different between treatment arms [relative rate 1.54 (95% CI 0.35–6.11); P = 0.51]. Although
use [relative rate 0.63 (95% CI 0.10–3.28); P = 0.56] or baseline insulin use [relative rate
2.16 (95% CI 0.38–3.29); P=0.141] and the risk of hypoglycaemia. Over 12 weeks, further
participants (24%). Of these, seven had their medications adjusted in the first 2 weeks, one at
3 weeks and one at 5 weeks. Six of these participants required one medication adjustment,
two participants required two medication adjustments and one participant required three
medication adjustments. Two participants in the consecutive days' fasting arm and three in
the non-consecutive days' fasting arm had their medications uptitrated by their general
physician or diabetologist for hyperglycaemia. The three participants who dropped out or
There were significant technical and participant challenges in acquiring the continuous
glucose monitoring data. Consequently data were only available from 27/39 participants who
were fitted with a continous glucose monitoring device. From these, glucose concentrations
from a total of 786 non-fasting days and 425 fasting days were obtained. There were seven
hypoglycaemic events in five participants while the monitor was being worn. Two occurred
on fasting days and five occurred on non-fasting days. The mean (SD) subcutaneous glucose
reading on fasting days was 8.34 (2.18) mmol/l, and on non-fasting days it was 8.93 (2.59)
mmol/l.
Baseline, 6-week and 12-week food diaries were completed by 33 (89%), 25 (68%) and 33
Accepted Article
participants (89%), respectively. The mean (SD) total energy intake was 3430 (1129) kJ on
fasting days and 7472 (2418) kJ on non-fasting days. There was a sustained reduction in total
calorie intake after 6 and 12 weeks compared with baseline (Fig. 2). There was no significant
any timepoint. After 6 and 12 weeks, self-reported adherence rates to the calorie target of
With the exception of a clinically insignificant difference in LDL cholesterol level, there
were no statistically significant differences between the consecutive days' fasting and non-
consecutive days' fasting arms with regard to secondary outcomes (Table 2). Weight, waist
circumference, fat mass, HbA1c and fasting glucose values improved from baseline to 12
levels achieved statistical significance, but were small and not considered clinically
Quality of life
Treatment arm had no significant impact on any quality-of-life or diabetes impact score in
any domain; however, in all participants there was a small but statistically significant
improvement in the global quality-of-life rating between baseline and week 12 . The effect
was an improvement of 0.66 on a scale where the maximum achievable difference is 6.0
[0.66 (95% CI 0.48, 0.85); P = 0.020]. This was offset by a small, statistically significant
increase in the global impact of diabetes on quality of life between baseline and week 12
[0.70 (95% CI 0.34–1.04); P <0.001]. The effect was an increase in the negative impact of
There was no significant change in the impact of diabetes on any of the 19 specific domains
Accepted Article
or the average weighted score based on these domains. No gross departures from underlying
Discussion
standardized medication reduction and weekly contact, intermittent fasting was associated
with a twofold increase in hypoglycaemia on fasting days in people with Type 2 diabetes who
were following a 5:2 diet. The overall risk of hypoglycaemia on both fasting and non-fasting
days, however, was lower than expected, there were no episodes of severe hypoglycaemia,
and most participants did not experience hypoglycaemia. These observations suggest that the
pattern of fasting. Continuous glucose monitor recordings supported the reported absence of
This is one of few studies examining the risk of hypoglycaemia in people with Type 2
diabetes who are following an intermittent calorie-restriction diet [9,16]. Ash et al. [16]
studied 51 overweight or obese men with Type 2 diabetes on oral hypoglycaemia medication
The intermittent energy restriction was 4 consecutive days of a 4184-kJ liquid meal for 12
hypoglycaemic events were not reported. More recently, Carter et al. [9] studied a medication
reduction protocol based on baseline HbA1c. This was altered during the study period because
occurring in insulin users was 4.3 ± 3.8 events over the 12-week study period. This compares
period, despite a greater proportion of insulin and sulfonylurea users at baseline (56%), with
Accepted Article
only two participants not on hypoglycaemic medication. This rate is substantially lower
hypoglycaemia. Most importantly HbA1c levels did not deteriorate because of medication
reduction at baseline.
Consistent with other studies on intermittent fasting there was a clinically relevant reduction
in weight, HbA1c and fasting glucose despite the relatively mild dietary intervention and short
We observed a difference in adherence to the target calorie intake at 6 weeks and 12 weeks;
We found that participants’ perceptions of the negative impact of diabetes on their quality of
life increased slightly during the study. This may have resulted from the dietary constraints,
requirement for regular testing and the emphasis on body shape and weight during a period of
calorie restriction. Despite this there was a small improvement in their global quality-of-life
rating.
The present study has some limitations. Our primary outcome was reliant on self-reported
rates are low in people with Type 2 diabetes [17], self-reported hypoglycaemia in response to
included who was not taking insulin or an oral hypoglycaemic agent at the time of the study
and therefore would not be expected to have any hypoglycaemic events. A sensitivity
analysis excluding this individual did not alter the interpretation of the data.
Generalized linear mixed models for count measures are prone to bias depending on the
estimation methods used and the specified distribution of the outcome variable [19]. We ran
the procedure in R (‘lme4’ package), which uses a Laplace approximation method [20]. For
our analysis, we categorized insulin use as a dichotomous variable which may have obscured
The lower-than-anticipated hypoglycaemia rate meant the study was underpowered to detect
Systematic underreporting of food intake in self-reported food diaries is well recognized [12].
In this study, there may have been greater risk of underreporting on fasting days; however,
the within-participant comparison between fasting and non-fasting days may have attenuated
any one individual’s tendency to underreport their intake. Systematic underreporting would
not be expected to differ between the consecutive days' fasting and non-consecutive days'
fasting arms.
In conclusion, the principle finding of the present study was that intermittent fasting was
intervention did result in weight loss, reduced HbA1c and a small improvement in quality of
life. Our study protocol could be adopted for the longer-term studies that will be required to
Funding sources
The study was supported by a grant from the New Zealand Society for the Study of Diabetes.
Competing interests
None declared.
Acknowledgements
The authors would like to thank all the participants and the administrative and nursing staff at
References
1. Collaboration NCDRF. Trends in adult body-mass index in 200 countries from 1975 to 2014:
a pooled analysis of 1698 population-based measurement studies with 19.2 million participants.
2. (NCD-RisC) NCDRFC. Worldwide trends in diabetes since 1980: a pooled analysis of 751
3. Lim EL, Hollingsworth KG, Aribisala BS, Chen MJ, Mathers JC, Taylor R. Reversal of type
2 diabetes: normalisation of beta cell function in association with decreased pancreas and liver
nonalcoholic hepatic steatosis, hepatic insulin resistance, and hyperglycemia by moderate weight
Accepted Article
reduction in patients with type 2 diabetes. Diabetes 2005; 54:603–608.
5. Antoni R, Johnston KL, Collins AL, Robertson MD. Effects of intermittent fasting on glucose
6. Patterson RE, Laughlin GA, LaCroix AZ, Hartman SJ, Natarajan L, Senger CM et al.
Intermittent Fasting and Human Metabolic Health. J Acad Nutr Diet 2015; 115:1203-1212.
7. Seimon RV, Roekenes JA, Zibellini J, Zhu B, Gibson AA, Hills AP et al. Do intermittent
diets provide physiological benefits over continuous diets for weight loss? A systematic review of
8. Williams KV, Mullen ML, Kelley DE, Wing RR. The effect of short periods of caloric
restriction on weight loss and glycemic control in type 2 diabetes. Diabetes Care 1998; 21:2–8.
9. Carter S, Clifton PM, Keogh JB. The effects of intermittent compared to continuous energy
restriction on glycaemic control in type 2 diabetes; a pragmatic pilot trial. Diabetes Res Clin Pract
2016; 122:106–112.
10. Arnason TG, Bowen MW, Mansell KD. Effects of intermittent fasting on health markers in
those with type 2 diabetes: A pilot study. World J Diabetes 2017; 8:154–164.
report from the American Diabetes Association Workgroup on Hypoglycemia. Diabetes Care 2005;
28:1245–1249.
12. Ortega RM, Perez-Rodrigo C, Lopez-Sobaler AM. Dietary assessment methods: dietary
individualized questionnaire measure of perceived impact of diabetes on quality of life: the ADDQoL.
Bjergaard U. Self-reported non-severe hypoglycaemic events in Europe. Diabet Med 2014; 31:92–
Accepted Article
101.
16. Ash S, Reeves MM, Yeo S, Morrison G, Carey D, Capra S. Effect of intensive dietetic
interventions on weight and glycaemic control in overweight men with Type II diabetes: a randomised
17. Frier BM, Jensen MM, Chubb BD. Hypoglycaemia in adults with insulin-treated diabetes in
the UK: self-reported frequency and effects. Diabet Med 2016; 33:1125–1132.
analyzing hypoglycemia data using bootstrap simulations. J Biopharm Stat 2015; 25:54–65.
20. Team RC. R: A language and environment for statistical computing. . Vienna, Austria: R
FIGURE 2 Mean daily calorie intake (kJ) at all time points. Self-reported energy intake was
consistently lower at 6 and 12 weeks compared with baseline in both treatment arms. There
was a trend toward lower energy intake in the non-consecutive group, but this did not achieve
statistical significance.
n = 19 n = 18
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination
and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi:
10.1111/dme.13595
This article is protected by copyright. All rights reserved.
ccepted Articl
Table 2 Changes in anthropometric and biochemical measurements from baseline to week 12
difference
Baseline 12 weeks
19)
(n = 18) (n = 18)
Weight, kg 109.8 (20.3) 108.7 (20.4) 106.2 (20.1) 105.6 (19.9) 0.5 (–1.5, 2.5) 0.65
BMI, kg/m2 36.8 (5.2) 36.6 (5.3) 36.0 (5.2) 36.1 (5.5) 0.0 (–1.5, 1.49) 0.96
Body fat, % 41.2 (6.7) 38.6 (7.2) 40.3 (7.0) 37.5 (8.1) –0.2 (–2.2, 1.8) 0.82
Waist circumference, cm 122.5 (13.6) 120.4 (17.0) 119.1(14.2) 118.8 (16.6) 0.0 (–10.7, 10.7) 0.99
HbA1c, mmol/mol 66 (7) 68 (10) 59 (8) 62 (10) 1.4 (–3.0, 5.7) 0.53
HbA1c, % 8.2 (1.3) 8.4(1.8) 7.5(1.5) 7.8 (1.8) 0.1 (–0.3, 0.5) 0.53
mmol/l
9.0 (2.1) 8.2 (2.8) 7.9 (1.7) 6.9 (2.1) –0.8 (–2.1, 0.5) 0.21
Thyroid stimulating 1.6 (0.8) 2.1 (1.9) 1.8 (0.9) 2.0 (1.2) 0.0 (–5.9, 5.9) 0.99
hormone, mU/l
Total cholesterol, 4.2 (1.0) 3.9 (0.8) 3.8 (0.8) 4.0 (1.0) 0.4 (0.1, 0.8) 0.01
mmol/l
HDL cholesterol, mmol/l 1.1 (0.2) 1.0 (0.2) 1.1 (0.2) 1.1 (0.2) 0.0 (–0.0, 0.11) 0.37
LDL cholesterol, mmol/l 2.1 (0.8) 2.1 (0.8) 2.0 (0.7) 2.25 (0.81) 0.3 (0.0, 0.6) .03
Cholesterol:HDL ratio 3.8 (1.0) 3.9 (0.9) 3.5 (0.9) 3.9 (1.0) 0.0 (–0.5, 0.6) 0.87
Triglycerides, mmol/l 1.8 (0.6) 1.8 (0.7) 1.7 (0.4) 1.7 (0.8) 0.0 (–0.3, 0.3) 0.80
Creatinine, µmol/l 79 (20) 100 (44) 82 (20) 93 (35) –5.3 (–12.2, 1.5) 0.12
Estimated GFR, 74 (17) 59 (20) 73 (17) 63 (20) 3.5 (–1.4, 8.4) 0.15
ml/min/.1.73m2
Systolic blood pressure, 133 (15) 132 (12) 129 (12) 129 (16) 1.2 (–7.4, 10.0) 0.77
mmHg
mmHg
78 (11) 74 (11) 75 (10) 72 (10) –6.5 (–5.8, 4.5) 0.80
This article has been accepted for publication and undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process, which may
lead to differences between this version and the Version of Record. Please cite this article as
doi: 10.1111/dme.13595
This article is protected by copyright. All rights reserved.
Accepted Article
Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.
Alternative Proxies: