How Does Meal Timing Impact My Blood Sugar?

In Sigma Statements by Alan Flanagan5 Comments

Meal Timing & Blood Sugar

Meal timing has been a popular, and at times controversial, topic of interest in nutrition.

Closer scrutiny reveals that most of that debate was centred on questions for performance nutrition, in particular body composition-related outcomes. For example, a common ground for debate in this respect was the question of meal frequency: i.e., did a higher meal frequency enhance the rate of weight/fat loss compared to a lower meal frequency.

However, it is now well-established that meal frequency is not an important consideration for that particular outcome [1]. In general, for outcomes such as energy expenditure [2], weight or fat loss [3], blood lipids and blood pressure [4], there is little difference in various meal timing regimens.

However, pause for caution against over-extrapolation. Just becasuse there is little evidence for the importance of meal timing in relation to those outcomes, this should not  be (although often is) interpreted to mean there is little evidence for the importance of meal timing at all.

One specific area where meal timing may be an important factor is in relation to glycaemic control. This effect may be more pronounced as glucose tolerance progressively deteriorates, i.e., the magnitude of effect appears to be greater in individuals with pre-diabetes or diagnosed type-2 diabetes (T2D).

In this Sigma Statement, the following sections will be addressed:

  1. The mechanistic underpinnings of potential differences in glycaemic control relative to time of day
  2. Evidence from interventions comparing breakfast consumption vs. breakfast omission (i.e., fasting until lunch)
  3. Evidence from interventions targeting distribution of energy across the day
  4. Evidence from interventions targeting meal frequency

Where available, evidence will be discussed from studies in healthy individuals, individuals with impaired glucose tolerance/prediabetes, and individuals with type-2 diabetes.

Diurnal Variation in Glucose Metabolism

The diurnal variation in glucose tolerance [5] is perhaps the most well-established feature of metabolism from a time-of-day perspective. 'Diurnal' meaning occurring daily, e.g., a 'diurnal rhythm' being a biological rhythm that fluctuates across the day.

The diurnal variation in glucose metabolism reflects rhythms in:

  1. Insulin action
  2. Glucose disposal
  3. Circulating non-esterified fatty acids (NEFA)

With regard to insulin action, research has demonstrated that incretin hormones, in particular glucose-dependent insulinoptropic polypeptide (GIP) and glucagon-like peptide-1 (GLP-1), follow a diurnal rhythm [6]. And there is an amplification in the early phase of the day (i.e. in the morning). Both GIP and GLP-1 act as important co-factors for the release of insulin in response to nutrient intake [7].

The first-phase insulin response (over 30 - 45 minutes) after food intake has been shown to be significantly greater in the morning compared to the evening [8], and insulin responses in the afternoon and evening show a delayed rise and prolonged elevation [9].

It has been proposed that the enhanced insulin response in the morning compared to the evening may be explained, in part at least, by the corresponding diurnal variation in incretin hormone activity.

Lindgren et al. [10] carried out a study that compared two meals, one consumed at 8 a.m. and the other at 5 p.m., which were matched for both calorie content and macronutrient composition. They showed there was a more pronounced elevation in GIP and GLP-1  in response to the morning meal, compared to the afternoon/evening meal. The elevations in GLP-1 and GIP corresponded to an augmented, rapid insulin response, and thus lower glucose levels after the meal.

However, as noted as early as 1974 by Zimmet et al. [11], impaired insulin responses in the afternoon do not fully explain the impaired glucose tolerance that is observed with later meals. In that early work, it had been suggested that higher circulating non-esterified fatty acids (NEFA) may play a role in the diurnal variation in glucose tolerance and insulin action.

In an elegant 1999 metabolic ward  study, Morgan et al. [12] investigated the relationship between insulin sensitivity and circulating NEFA levels in response to insulin tolerance tests administered at either 8 a.m. or 8.30 p.m. They demonstrated that the diurnal variation in insulin sensitivity mirrors the diurnal variation in circulating NEFA levels, which are elevated in the evening and contribute to impaired insulin sensitivity and glucose disposal.

The relevance of the circadian rhythm in circulating NEFA levels extends to the following day. Extended morning fasting may result in higher NEFA levels in the afternoon [13].  This may contribute to elevated postprandial glucose levels in response to both lunch and dinner meals [14].

Conversely, eating in the morning results in a suppression of circulating NEFA levels [15], which appears to have a legacy effect [16] and is associated with attenuated postprandial glucose responses to subsequent meals [17]. In this respect, the evidence suggests that the suppression of NEFA and glucose disposal are related. Using stable isotope tracers to trace the metabolic fate of glucose, Jovanovic et al. found that glucose uptake in skeletal muscle glycogen occurred at a 50% greater rate when a morning meal preceded lunch, which was associated with significantly reduced glucose response after lunch [18]. However, both the enhanced glycogen signalling and the lower postprandial glucose response strongly correlated with the pre-lunch NEFA levels, which were significantly lower following breakfast consumption.

Key Point

Cumulatively, the evidence shows a pronounced diurnal variability in glucose tolerance, which is enhanced in the early phase of the day and declines over the course of the day, resulting in impaired glycaemic responses to the timing, size, and composition of meals later in the day.

Breakfast vs. No Breakfast

A number of interventions have compared the postprandial glucose levels in situation where one either fasts until lunch or consumes breakfast before lunch.

In an intervention in healthy lean males [19], Kobayashi et al. compared two diets matched for calories: one that included breakfast (3 meals/day) and one that skipped breakfast (2 meals/day). The study showed significantly greater postprandial glucose responses after lunch and dinner in the breakfast skipping condition. Overall 24-hour blood glucose levels were significantly higher in the breakfast skipping condition. However, it should be noted that while the total diets were isocaloric, this meant that the lunch and dinner meals of the breakfast-skipping condition contained significantly more calories than the lunch and dinner meals of the breakfast condition, where calories were spaced across three meals rather than two. Thus, the magnitude of glycaemic response in this study may reflect the difference in energy content of the specific meals. Of note, however, was the prolonged elevation of blood glucose levels in the breakfast skipping condition in response to the dinner meal administered at 8 p.m., which remained significantly elevated through the overnight period. This is consistent with the well-established diurnal variation in glucose tolerance, described above.


Taken from: Kobayashi et al., Obes Res Clin Pract. May-Jun 2014;8(3):e201-98.
© 2014 Asian Oceanian Association for the Study of Obesity, Published by Elsevier Ltd. All rights reserved.

Figure above shows the diurnal variations of blood glucose recorded by Kobayashi et al. Mean values were plotted at every 5 minutes. Mean values for morning, afternoon, evening and sleep periods are also shown.

Second Meal Effect

The potential attenuation of postprandial glucose levels in response to lunch depending on whether breakfast is consumed or omitted may reflect a phenomenon known as the "second meal effect".

Officially termed the 'Staub-Traugott Effect', this phenomenon was first described a century ago during experiments using sequential oral glucose tolerance tests (OGTT). In these experiments it was noted that, despite the exact same amount of glucose being ingested, the rise in blood glucose measured by a second OGTT was much lower than the rise in blood glucose after a first OGTT.

The second meal phenomenon has been consistently demonstrated in metabolically healthy humans [20, 21]. Mechanistically, both the suppression of circulating NEFA and enhanced skeletal muscle glycogen uptake (as described above) appear to mediate this effect. In a controlled feeding study in participants with type 2 diabetes, Jovanovic et al. found that the postprandial glucose response to lunch was 95% lower when breakfast preceded lunch compared to fasting until lunch [22].

Lee et al. also investigated the presence of the 'second meal effect' in participants with type 2 diabetes, comparing breakfast consumption to fasting until lunch [23]. They found that the glucose response after lunch to was 35% higher when participants fasted until lunch, compared to when breakfast had preceded lunch. The improved glucose tolerance in response to lunch following breakfast correlated with the pre-lunch NEFA levels, which had been suppressed in response to breakfast and elevated only slightly in response to lunch. Jovanovic et al. also demonstrated that the blood glucose response to lunch following a preceding breakfast was significantly lower [18], and corroborating the findings by Lee et al., showed that the attenuated postprandial glucose response after lunch correlated with the pre-meal NEFA levels.

The effect of breakfast omission may be more pronounced as the state of underlying glucose intolerance progressively deteriorates. In a controlled feeding study in participants with type 2 diabetes, Jakubowicz et al. showed that glucose responses to lunch and dinner were 36.8% and 26.6% higher, respectively, when breakfast was omitted compared to when breakfast preceded those meals [24]. Unlike the study by Kobayashi et al. above, where the meals in the breakfast omission condition were larger, the meals in this study were isocaloric, such that the exaggerated postprandial glucose responses were not attributable to higher calorie content alone.

Another study, completed in a metabolic ward, compared the effects of two diets where one of the meals was left out; one that omitted breakfast omission vs. another that omitted dinner. The diets were matched for calories, being set at maintenance energy levels. Breakfast omission resulted in significantly higher glucose and insulin levels following lunch, compared to dinner omission [25]. When breakfast was skipped, it also resulted in greater insulin resistance and higher 24-hour glucose levels.

Key Point

While the majority of evidence with weight loss as an outcome do not suggest any particular advantage to morning energy intake, from the perspective of glycaemic control (particularly in states of impaired glucose tolerance) there is consistent evidence of a benefit to morning energy intake compared to later meal initiation for postprandial glucose responses.

Meal Timing: Energy Distribution

While the studies in the previous section compared morning energy vs. fasting until lunch, the distribution of energy across the day appears to be an important consideration for glycaemic control.

Bandín et al. conducted a controlled feeding intervention in otherwise healthy, lean females [26]. The study had both breakfast and dinner occurring at the same times (8 a.m. and 8 p.m., respectively) in each condition, but with lunch occuring either earlier (1.30 p.m.) or later (4 p.m.). In response to the late lunch (4 p.m.), the glucose response was 46% higher compared to the early (1.30 p.m.) lunch, and blunted carbohydrate oxidation. While the initial rise in glucose in response to both lunches was similar, what characterised the later lunch glucose profile was a prolonged elevation in blood glucose levels, consistent with the impaired glucose tolerance observed later in the day [27].

Cu et al. compared the metabolic effects of having dinner at 6 p.m. or at 10 p.m., in a controlled feeding study where the diets were matched for calories [28]. The times of the other meals were matched between the diets. In response to the 10 p.m. dinner, both glucose and insulin remained significantly elevated from 11 p.m. to 5 a.m., and the glucose peak was 18% higher. Glucose levels over the entire day were also significantly higher in response to the later dinner.

Leung et al. investigated the effects of low-glycaemic index meals consumed at 8 a.m., 8 p.m., and midnight in healthy individuals [29]. They showed that postprandial glucose levels were significantly greater after the later meals, compared to the meal at 8 a.m.. After the midnight meal, glucose levels remained significantly elevated above baseline three hours after the meal, while in the 8 a.m. and 8 p.m. conditions glucose had returned to baseline after three hours.

Morgan et al. investigated the effects of temporal distribution in a controlled feeding study [30], comparing:

  1. 60% energy and breakfast and 20% at dinner
  2. 20% energy and breakfast and 60% at dinner

Each of these conditions was also tested with both high and low glycaemic index (GI) meals. Compared to the other test conditions, the high-energy (60%) and high GI dinner resulted in significantly higher postprandial glucose and insulin levels.

Jakubowicz et al. have also conducted a number of interventions considering energy distribution. In one study in participants with type 2 diabetes [31], they compared two 1,500 kcal interventions:

  1. Large Breakfast: 700kcal at breakfast, 600kcal at lunch, and 200kcal at dinner
  2. Large Dinner: 200kcal breakfast, 600kcal at lunch, and 700kcal dinner

The 700 kcal breakfast intervention resulted in a 20% lower whole-day glucose levels. Postprandial glucose was 24% lower after the 700 kcal meal at breakfast compared to the 700 kcal meal consumed at dinner. Further, the timing of the peak in insulin secretion, the magnitude of the peak in insulin, and the post-prandial area under the curce (AUC) for insulin were all impaired in response to the 700kcal dinner, compared to the 700kcal breakfast.


Image origally from: Circadian Eating Lecture - Danny Lennon

In the Bath Breakfast Project [32], participants were randomised to either consume more than 700 kcal before 11 a.m. or to fast until lunch at 12 p.m.. Metabolic control was improved in the high-energy morning group compared to morning fasting. There was improved insulin sensitivity in the breakfast group, observed in both lean participants and participants with obesity. In addition, participants with obesity in the breakfast group had lower nocturnal blood glucose levels. While lean participants in the fasting group had  higher blood glucose variability in the afternoon/evening [33].

There is also evidence of an effect of macronutrient distribution. Pearce et al. compared the effects of distributing a majority of daily carbohydrates to breakfast or lunch, carbohydrates equally distributed between meals across the day, or majority of carbohydrates distributed to dinner, in participants with poorly controlled T2D [33].  Distributing a majority of carbohydrate to breakfast or lunch resulted in significantly lower daily glucose excursions, compared to equal distribution or a majority at dinner.

Kessler et al. investigated the effects of diurnal distribution of carbohydrates and fats on glycaemic control in participants with impaired glucose tolerance and participants with normal glucose tolerance [34]. The study compared two diets:

  1. HC/HF Order: Carbohydrate-rich meals consumed up until 1.30 p.m., followed by fat-rich meals in the afternoon/evening between 4.30 p.m. and 10 p.m.
  2. HF/HC Order: Fat-rich meals until 1.30 p.m and carb-rich meals in the afternoon/evening.

Each meal sequence was consumed for 4-weeks in a crossover design. The glucose response after test meals in the afternoon/evening was 4.5-fold higher in participants with impaired glucose tolerance and 2.5-fold higher in participants normal glucose tolerance. The overall effect of diet over 4-weeks demonstrated worsened glucose tolerance to the HF/HC sequence in participants with pre-existing impaired glucose tolerance, but not participants with normal glucose tolerance.

Key Point

In sum, the diurnal variation in glucose tolerance across the day suggests that the distribution of energy and carbohydrate may influence postprandial glucose metabolism. Interventions comparing both energy and carbohydrate distribution suggest enhanced glycaemic responses with earlier temporal distribution compared to evening distribution.

Meal Frequency

Meal frequency has been theorised to be a strategy to improve overall glycaemic control, particularly for diabetes management. Early research from Professor David Jenkins and colleagues suggested that 'nibbling' patterns of eating were preferable to 'gorging' patterns in participants with type 2 diabetes [35]. This gave rise to the idea that smaller, more frequent meals might be a better option than larger, less frequent meals.

A more recent study by Hibi et al. compared the blood glucose respose to a meal frequency of either nine meals per day or three meals per day [36]. The study had participants follow their meal frequency for three consecutive days, with blood glucose levels being constantly monitored with continuous glucose monitors (CGM). This was then followed by testing responses to an OGTT on the 4th day. The participants then crossed over and completed the experiment again with the opposite diet. The study included both participants with impaired and normal glucose tolerance. CGM data indicated that while average 24-hour blood glucose did not differ between meal frequency patterns, peak glucose levels were lower in the 9-meal condition and time spent in a hyperglycaemic state was higher with the 3-meal condition, regadless of the participants glucose tolerance. In response to the test meal, there was no significant differences in glucose metabolism in those with normal glucose tolerance. However, those with impaired glucose tolerance showed a significantly lower glucose peak following the 9-meal condition.

However, there is also evidence to the contrary from other studies. In crossover intervention in participants with type 2 diabetes, Kahleova et al. compared the effects of two diet structures [37]:

  1. Six meals spread across the day
  2. Two meals front-loaded to early in the day

Both diets targeted a 500 kcal calorie deficit, and participants completed each diet for 12 weeks. Fasting glucose levels decreased by 0.47mmol/L in the 6-meal condition vs. 0.78mmol/L in the 2-meal condition, and fasting C-peptide levels (higher levels indicating impaired insulin function) by 0.04nmol/L in the 6-meal condition vs. 0.14nmol/L in the 2-meal condition. Fasting insulin and HbA1c were comparably reduced by both conditions. However, in this study no data was presented on the distribution of energy and carbohydrates between meals.


Image from: Kahleova et al., Diabetologia. 2014; 57(8): 1552–1560.
Data are shown as changes from baseline in response to the regimen of six (A6) and two meals (B2) a day.

A more recent intervention by Jakubowicz et al. compared different meal frequencies but also useful information about energy distribution across the day [38].  In the study they compared:

  1. Three meals per day with calories and carbohydrate front-loaded to breakfast: 700 kcal breakfast, 600 kcal lunch, and 200 kcal dinner.
  2. Six meals per day with calories and carbohydrates equally distributed across meals over the day: breakfast, lunch, and dinner with 20-25% energy and 23% daily carbohydrates, and 3 daily snacks of 10% energy and 10% carbohydrate.

The primary outcome of the study was total daily insulin dose (TDID), and the 3-meal group saw their daily insulin use drop by 26 units (from 60 to 34 units per day) after 12 weeks. Conversely, the 6-meals-per-day diet saw their daily dosage increase by 4 units. Also measured was the amount of time spent hyperglycemic each day. The 3-meal diet saw daily hyperglycemia drop from 8hr 59min at baseline to 3hr 3min at 12-weeks. While there was no change in the 6-meal group. Of note, the 3-meal diet led to a loss of ~5 kg bodyweight, compared to no change in the 6-meal group. While this would be expected to influence the results, there was no correlation between body weight and TDID, suggesting that the reduction in TDID occurred - to an extent - independent of weight loss.

Thus, the findings in relation to meal frequency appear to be contradictory. However, it may be possible to reconcile the apparent differences. In the Hibi et al. intervention, the 3-meal condition had energy intake equally divided between meals, such that the 8pm meal contained 33.3% of daily calories compared to 11.1% in the 9-meal condition. This would have the effect of biasing the postprandial responses to a worse effect in the low meal frequency group.

Support for this may be evident in the Jakubowicz et al. energy distribution studies comparing high vs. low energy dinners, where 700kcal consumed at 7 p.m. resulted in significantly higher blood glucose levels compared to the earlier front-loading of energy, and comparing the combination of reduced meal frequency and front-loaded energy distribution.

However, both the early Jenkins research and the study by Hibi et al. using CGM data were studies conducted over 1-day and 3-days, respectively, and were diets and energy balance levels. The studies by Kahleova et al. and Jakubowicz et al. have been over a longer duration of 12-weeks, and were energy deficit diets. This suggests an effect of study duration, energy levels, or perhaps both.

To potentially reconcile these factors, a 12-week isocaloric intervention from Papakonstantinou might be instructive [39]. In participants with either impaired glucose tolerance or type 2 diabetes, a meal frequency of six meals per day was compared to three meals per day.  The higher meal frequency resulted in better postprandial glucose responses to an OGTT. And participants with type 2 diabetes specifically, saw better glycaemic control as measured by HbA1c. There were no significant differences in other markers of blood glucose and insulin regulation. However, carbohydrate intake was still lower at breakfast compared to other meals. It is possible that front-loading more energy earlier could result in more beneficial responses, as suggested by the studies by Pearce et al. [33] and Jakubowicz et al. [31].

Key Point

Cumulatively, these data suggest that in the context of energy balance, a slightly higher meal frequency may lead to some minor improvements in postprandial glucose tolerance.

It is arguable, however, that the true effect of frequency relates to the distribution of energy and carbohydrates across meals throughout the day. Most of these interventions, with the exception of the energy distribution manipulation by Jakubowicz et al., have a higher ratio of evening to morning energy, which when compared to lower meal energy more equally distributed may not yield any benefit.

The lower energy content of meals in a higher frequency condition, particularly with lower evening energy intake, could be expected to result in more favourable glycaemic responses compared to higher evening energy intake.

With an energy deficit, however, it may that a combination of reduced (~3) meal frequency and front-loaded energy and carbohydrate distribution enhances overall improvements in glycaemic control, particularly in participants with impaired glucose tolerance and type 2 diabetes.


  1. There is clear and consistent evidence of effects of diurnal distribution of both energy and macronutrients from interventions in participants across a spectrum of glucose tolerant states, from normoglycaemic to type 2 diabetes.
  2. In states of impaired glucose tolerance, in particular type 2 diabetes, there is also consistent evidence of a deleterious metabolic effect of breakfast omission, for glycaemic control.

  3. The findings from these interventions mirror the evidence from metabolic studies demonstrating diurnal variation in glucose tolerance, and show impaired glucose tolerance from the late afternoon into the evening.

  4. The effects are particularly pronounced compared to earlier temporal distribution of calories. And the magnitude of effect increases as the underlying state of glucose tolerance decreases, i.e., effect is greatest in individuals with impaired glucose tolerance and type 2 diabetes.

    Statement Primary Author: Alan Flanagan

    Alan is the Research Communication Officer at Sigma Nutrition. Alan is currently pursuing his PhD in nutrition at the University of Surrey, UK, with a research focus in chrononutrition. Alan previuosly completed a Masters in Nutritional Medicine at the same institution.


    1. In reading your brilliantly structured article here Alan, the question sitting most at the front of my mind – and granted, it’ll be too heterogeneous to have a simplistic answer in light of all the study participants listed above – is:

      Do these 24 hour (acute) increases in glycaemia reflect a potential pathological process, or are they simply presumed as such because chronic elevations of high blood glucose is known to be problematic?

      I’d think that in eucaloric or hypercaloric states it may not be great, but in hypocaloric conditions perhaps there’s not much to worry about given the basis of the deficit seems to do some potently positive things for people’s health by and large (relatively speaking and with context as appropriate).

      Not sure here, would appreciate anyone to jump in though to discuss.

      1. Author

        Hey Troy, thanks for the question.

        So yes, these are intermediary factors, and post-prandial glucose excursions are a completely normal part of the post-prandial process.

        The inference we’re left to make is, what would be the effect of cumulative exposure over time? The hyper/eucaloric/hypo factor is relevant to consider for magnitude of effect, but the differences between breakfast omission vs. consumption in participants with T2D has still been shown in hypocaloric conditions. So the difference isn’t abolished by an energy deficit alone, i.e., the inference would be that comparing two people with T2D for example, if one back-loaded energy and the other front-loaded energy, the magnitude of improvement would be greater in the latter.

        From a population health perspective, I think we always have to be careful with “yes but energy deficit”, because the majority of the population are obviously not in that state (based on population-wide adiposity levels, that is a fair inference). Given what we know about the effect of factors like energy distribution on post-prandial metabolism, the question would be: what is the effect of this cumulative exposure over time, and does that provide a reasonable biological plausibility for increased metabolic disease risk observed in prospective studies? And the answer to that would be ‘yes’, i.e., it is plausible that certain eating patterns would contribute to impaired postprandial glycaemia, insulin, and lipids, adding up over time to adverse outcomes.

        Hope that helps clarify the thinking a bit?



        1. Thanks very much for the response Alan.

          I can comment overall noting that I think the concept of the possible long-term effects of cumulative higher average blood glucose given it’s tight regulatory patterns would pose something to consider for further investigation. In the case of the diagnosed diabetic – as you noted in your reply – it appears to be a problem posed which probably doesn’t need much more investigation (or at least, may not need precedence when it comes to funding dollars given other things in the area less understood).

          Specifically on what you mentioned about energy balance, I hope I was also clear I didn’t suggest the difference was ‘abolished’ but rather it was less to be concerned about. I definitely agree if we wanted to be pedantic about where and when we place our meals over a 24-hour period that smaller things matter especially in the health conscious individual.

          But to your point about population health – whether this small of a thing matters, when a deficit matters much more (regarding the end-point of concern with blood sugar being ‘mostly’ diabetes) is also part of where the conundrum lies in my thinking. You could suggest this might be more realistic given the increasing adiposity rates as you’ve noted at the level of the public and hence this strategy might well be a good one for the mass of folks in the western world eating more than is required. In this case the surplus of calories might benefit from the research above to minimise the sequelae of what diabetes will do to people’s quality of life, but I’d respectfully still argue that at this level the deficit is more important given what we know about even more subtle weight loss with 5-10% from starting weight.

          I apologise I’m not being too clear either, the context I’m inferring is more so an internal one. I’m a clinician working with individuals but also work in public health nutrition research so I’m forever grappling with the nuances of these studies presented with what’s either clinically important, population-based, or a combination of the two.

          Thank you for your thoughtful response Alan, enjoying many of your discussions on the podcasts.

    2. This post must be read by everyone. Most of us forget our health we do not have any issues but when we start facing issues, we realize that we did not fill our tummy in time and most of the time we have skipped our meals. Thank you for the very helpful post.

    3. Hi Alan, thank you for this article. It is certainly a bit over my head, but I still can derive important information. I feel that many studies – and guidelines – take high blood sugar (and diabetes) as a starting point. But what about (chronic) low blood sugar and frequent hypoglycaemia? Are there any nutritional studies related to that? Or does such a condition not exist? Thank you, Eline

    Leave a Comment