High consumption of whole-grain foods is associated with a reduced risk of chronic diseases including CHD( Reference Jacobs, Meyer and Kushi 1 , Reference Liu, Stampfer and Hu 2 ), hypertension( Reference Whelton, Hyre and Pedersen 3 ) and type 2 diabetes( Reference Salmerón, Manson and Stampfer 4 , Reference Meyer, Kushi and Jacobs 5 ). Suggested mechanisms of action include reduction in serum lipid concentrations( Reference Anderson and Hanna 6 ) and blood pressure( Reference Tighe, Duthie and Vaughan 7 ), increased insulin sensitivity( Reference Anderson 8 ) and reduction in thrombotic and inflammatory markers( Reference Marckmann, Sandström and Jespersen 9 , Reference King, Egan and Geesey 10 ). However, the results of the two most comprehensive, well-designed randomised control trials ever conducted with whole-grain foods found no significant effects on the major risk factors for CVD( Reference Tighe, Duthie and Vaughan 7 , Reference Brownlee, Moore and Chatfield 11 ).
Whole grains consumed in a western diet consist mainly of wheat, rye, maize and oats. These cereals have different chemical compositions, which could explain the different responses with regard to CVD risk markers. Research has focused on β-glucan-rich cereals such as oats for their potential effect on serum cholesterol concentration( Reference Anderson, Kritchevsky and Bonfield 12 ) and postprandial glycaemia( Reference Tosh 13 ), with inconsistent results.
Although numerous studies suggest that there is a beneficial effect of oat consumption on markers of CVD risk, there is a need for a rigorous assessment of the strength of this evidence. The present study aimed to systematically review the literature describing intervention studies that had investigated the effect of regular consumption of whole-grain oat-based products (including oat bran) on risk factors for CVD. The objectives of the study were (i) to summarise the extensive literature on the subject, (ii) to describe the relative strengths and weaknesses of the studies and (iii) to evaluate the need for large intervention trials.
Methods
Literature search and study selection
The methods for the present literature review have been previously described (Thies F, Masson LF, Boffetta P et al., in this supplement). Briefly, Embase, Medline and the Cochrane library (Cochrane Central Register of Controlled Trials) were searched for articles describing intervention studies with oat-based products published before 26 November 2012. A total of 1174 articles were identified (Fig. 1). Titles and abstracts of 654 articles were reviewed independently by two reviewers who agreed that the full text should be obtained for 244 articles. A further 178 articles were then excluded, following agreement by two reviewers. Ten additional articles were identified by searching the reference lists in relevant articles obtained from the database search.
Fig. 1 Flow diagram of article selection.
Data extraction
Data were extracted by one reviewer into pre-prepared tables and the data extraction from a random 10 % of articles was checked and agreed by a second reviewer. The primary outcomes of interest included blood lipids/lipoproteins, blood pressure, glucose and insulin.
Quality of reporting and reporting preferences
Use of the Jadad scale for reporting randomised control trials( Reference Halpern and Douglas 14 ) to score the quality of reporting of each article, as well as the criteria for classifying studies as randomised control trials, is described elsewhere (Thies F, Masson LF, Boffetta P et al., in this supplement). For studies that showed a statistically significant (P< 0·05) effect of oats consumption, the percentage change from baseline in the intervention group relative to the control group was the preferred measure to present. If this was not available in the article, it was calculated from the results given, and such values are indicated in the tables. Interventions that involved products with altered molecular weight of β-glucan were not included.
Results
Study characteristics
We identified seventy-six articles( Reference Tighe, Duthie and Vaughan 7 , Reference Abrahamsson, Goranzon and Karlstrom 15 – Reference Wolever, Gibbs and Brand-Miller 89 ) describing sixty-nine studies that assessed the effect of oat consumption on CVD risk factors (Fig. 1). Online Web Tables S1–S3 describe the characteristics of these sixty-nine studies with less than thirty subjects (forty-one studies, 59 %), between thirty and fifty-nine subjects (seventeen studies, 25 %), and at least sixty subjects (eleven studies, 16 %) in the oat intervention group, respectively. These tables are sub-grouped according to the quality of reporting of the articles: forty-six articles (61 %) had a low quality of reporting, and thirty articles (39 %) had a high quality of reporting.
Over half (54 %) of the studies were carried out in North America (thirty-four in the USA and three in Canada). Six studies were carried out in Australia, five in Sweden, four each in the UK and New Zealand, three in Finland, two in the Netherlands, and one each in Austria, France, Germany, Denmark, China, Mexico and Brazil. One multicentre study was carried out in sites in Canada, the UK and Australia.
Lipids
Tables 1–3 show the results of sixty-four studies that assessed the blood lipid response to oat intervention in studies with less than thirty subjects, thirty to fifty-nine subjects or at least sixty subjects, respectively, in the oat intervention group. Of these sixty-four studies, thirty-seven (58 %) and thirty-four (53 %) studies identified a statistically significant reduction in total cholesterol and LDL-cholesterol, respectively, mostly in hypercholesterolaemic subjects; the rest of the studies found no significant response. This significant reduction ranged from 2 to 19 % for total cholesterol and from 4 to 23 % for LDL-cholesterol. In the eleven studies that contained at least sixty subjects in the oat intervention group, a higher proportion of studies found significant reductions in total cholesterol and LDL-cholesterol (eight studies (73 %) and seven studies (64 %), respectively), but the magnitude of these reductions was more conservative: 3–6 % for total cholesterol and 4–8 % for LDL-cholesterol (Table 3).
Table 1 Oats and blood lipids (studies with <30 subjects in the oat intervention group)
LFLC, low-fat, low-cholesterol diet; MW, molecular weight.
* % change from baseline relative to comparison group estimated.
Table 2 Oats and blood lipids (studies with thirty to fifty-nine subjects in the oat intervention group)
NCEP, National Cholesterol Education Program.
* % change from baseline relative to comparison group estimated.
† Group A started using kilned oats and group B started using unkilned oats.
Table 3 Oats and blood lipids (studies with ≥60 subjects in the oat intervention group)
CNA, controlled-release nicotinic acid.
* % change from baseline relative to comparison group estimated.
Three studies found that oat consumption significantly increased HDL-cholesterol levels by 4–11 %. The ratios of total cholesterol:HDL-cholesterol and of LDL-cholesterol: HDL-cholesterol were reduced significantly in three studies (by 2–7 %) and in five studies (by 9–21 %), respectively. Five studies found a statistically significant reduction in TAG concentrations (by 11–24 %) following oat-based intervention.
There is currently no evidence that oat consumption influences concentrations of HDL2 and HDL3 subfractions( Reference Anderson, Gilinsky and Deakins 18 , Reference Turnbull and Leeds 45 , Reference Davy, Davy and Ho 49 , Reference Van Horn, Emidy and Liu 83 ), intermediary density lipoprotein (IDL) cholesterol( Reference Anderson, Gilinsky and Deakins 18 ), VLDL cholesterol( Reference Abrahamsson, Goranzon and Karlstrom 15 , Reference Anderson, Gilinsky and Deakins 18 , Reference Dubois, Armand and Senft 23 , Reference Hegsted, Windhauser and Morris 24 , Reference Kristensen and Bugel 33 , Reference Saltzman, Das and Lichtenstein 41 , Reference Spiller, Farquhar and Gates 42 , Reference Swain, Rouse and Curley 57 , Reference Mackay and Ball 62 , Reference Van Horn, Liu and Parker 87 ), or LDL, HDL or VLDL particle size( Reference Davy, Davy and Ho 49 ).
Blood pressure
Table 4 shows that three( Reference Saltzman, Das and Lichtenstein 41 , Reference Keenan, Pins and Frazel 55 , Reference Pins, Geleva and Keenan 75 ) of twenty-five studies found that oat consumption significantly reduced systolic blood pressure by 4–6 %. Two of these studies( Reference Saltzman, Das and Lichtenstein 41 , Reference Keenan, Pins and Frazel 55 ) had less than thirty subjects in the oat intervention group. Pins et al. ( Reference Pins, Geleva and Keenan 75 ) found that 73 % of participants receiving treatment for hypertension were able to stop or reduce their medication by one-half after 6 weeks of consuming oats compared with 42 % in the wheat-based cereal (control) group (P< 0·05). In addition, participants in the oat intervention group whose medication was not reduced had a significant 4 % decrease in systolic blood pressure in comparison with the control group. The other twenty-two studies found no significant effect of oat consumption on systolic blood pressure.
Table 4 Results of studies assessing the effect of oat consumption on blood pressure
OBC, oat bran concentrate.
* % change from baseline relative to comparison group estimated.
Glucose and insulin
Blood glucose levels changed significantly in response to oat consumption in five out of twenty-one studies (Table 5). Glucose levels increased in three( Reference Noakes, Clifton and Nestel 37 , Reference Conceicao de Oliveira, Sichieri and Sanchez Moura 51 , Reference Uusitupa, Ruuskanen and Makinen 59 ) of these studies relative to the comparison group or baseline, and glucose levels decreased in the other two studies( Reference Frank, Sundberg and Kamal-Eldin 52 , Reference Pins, Geleva and Keenan 75 ). Four of these five studies had less than thirty subjects in the oat intervention group, and the other study, with n 45 in the oat intervention group, found a 13 % decrease in glucose after 12 weeks of an oat-rich diet compared with the control group who consumed wheat cereals( Reference Pins, Geleva and Keenan 75 ).
Table 5 Results of studies assessing the effect of oat consumption on glucose and insulin
* % change from baseline relative to comparison group estimated.
Fifteen out of sixteen studies found no significant effect of oats on insulin concentrations. One relatively small study with twenty-two participants found that high-molecular weight oat bran significantly increased insulin concentrations by 23 % compared with baseline( Reference Frank, Sundberg and Kamal-Eldin 52 ). None of the studies that measured glucose:insulin ratio( Reference Conceicao de Oliveira, Sichieri and Sanchez Moura 51 ), HbA1c( Reference Abrahamsson, Goranzon and Karlstrom 15 , Reference Maki, Galant and Samuel 35 ), homeostasis model assessment( Reference Tighe, Duthie and Vaughan 7 , Reference Saltzman, Das and Lichtenstein 41 , Reference Charlton, Tapsell and Batterham 66 ), quantitative insulin sensitivity check index (QUICKI)( Reference Tighe, Duthie and Vaughan 7 ), insulin sensitivity( Reference Saltzman, Das and Lichtenstein 41 , Reference Davy, Davy and Ho 49 , Reference Davy, Melby and Beske 50 , Reference Keenan, Pins and Frazel 55 ) or the acute insulin response to glucose( Reference Davy, Davy and Ho 49 ) found a significant effect of oat consumption on these variables. One study found that glucose effectiveness decreased in the oat intervention group by 5 % but increased in the wheat intervention group by 19 % (P= 0·03 for interaction)( Reference Davy, Davy and Ho 49 ).
Other outcomes
None of the studies that measured C-reactive protein( Reference Tighe, Duthie and Vaughan 7 , Reference Jenkins, Nguyen and Kendall 25 , Reference Sturtzel, Dietrich and Wagner 44 , Reference Theuwissen, Plat and Mensink 77 , Reference Maki, Beiseigel and Jonnalagadda 82 , Reference Wolever, Tosh and Gibbs 88 ), lipoprotein(a) (Lp(a))( Reference Kelley, Hoover-Plow and Nichols-Bernhard 28 , Reference Berg, Konig and Deibert 79 ), fibrinogen( Reference Bremer, Scott and Lintott 20 ) or IL-6( Reference Tighe, Duthie and Vaughan 7 ) found a significant effect of oat consumption on these variables. There is also a lack of evidence for a beneficial effect of oats on endothelial function( Reference Katz, Nawaz and Boukhalil 70 – Reference Katz, Evans and Nawaz 72 ).
One study( Reference Sturtzel, Dietrich and Wagner 44 ) reported measuring plasma homocysteine and found that concentrations were reduced by 16 % in response to 12 weeks of oat bran. A Danish study found that plasminogen activator inhibitor-I and factor VII levels decreased significantly by 27 and 7 %, respectively, following a 2-week oat bran v. a low-fibre diet( Reference Kristensen and Bugel 33 ), but no other studies reported measuring these outcomes. Another study measured serum NEFA that increased by 19 % after consuming 35–50 g/d of oat bran for 4 weeks, relative to the group consuming an oat bran-free diet( Reference Berg, Konig and Deibert 79 ).
Discussion
Lipids/lipoproteins
The present systematic review supports the results of observational studies suggesting that increased oat consumption has a beneficial effect on serum cholesterol concentration, particularly in hypercholesterolaemic subjects. This is consistent with Ripsin et al.’s( Reference Ripsin, Keenan and Jacobs 90 ) rigorous meta-analysis that concluded that about 3 g/d of soluble fibre from oat products can lower total cholesterol by 0·13 to 0·16 mmol/l, with a greater reduction in individuals with higher initial cholesterol concentrations. A 1 % reduction in total cholesterol or LDL-cholesterol is associated with a 2–3 % or 1 % decreased risk, respectively, of CHD( 91 ). The magnitude of the effect found in the present review (3–6 % for total cholesterol and 4–8 % for LDL-cholesterol when considering studies with a sufficient sample size) would translate to a 6–18 % decrease in CHD risk, which would equate to a substantial health benefit at a population level. However, increased oat consumption does not seem to significantly benefit other systemic lipid/lipoprotein markers associated with CVD risk, such as serum TAG and HDL-cholesterol concentration.
Lipoprotein particle number and size, particularly for LDL, are possibly strong predictors of CVD( Reference Superko and Gadesam 92 ) and could provide an independent measure of atherogenicity, which may be superior to total cholesterol determination. However, only a few studies evaluated the effect of oat intervention on the size and concentration of lipoprotein particles, with inconclusive results. More evidence is needed to establish whether increased oat consumption favourably affects the lipoprotein particle profile.
Blood pressure
Few studies found a significant effect of increased oat consumption on blood pressure. However, none of the studies carried out to date was adequately powered to rigorously evaluate the effect of oats or oat bran on this outcome. Furthermore, blood pressure results from these studies were most likely averaged from only two or three consecutive measurements. Such methodology, recommended by the British Hypertension Society, might be useful to identify hypertensive subjects but does not represent a precise method for measuring blood pressure, as recently suggested( Reference Tighe, Duthie and Vaughan 7 ).
Tighe et al. ( Reference Tighe, Duthie and Vaughan 7 ) found a significant reduction in systolic blood pressure after 12 weeks intervention with whole grain (wheat or oats plus wheat) compared with a refined cereals group. Blood pressure was measured using additional consecutive readings until the last three measurements varied by less than 8 %, and a significant reduction would not have been identified using the conventional method of measuring blood pressure. This demonstrates the requirement for all types of intervention trials (pharmaceutical, supplement, food-based, lifestyle interventions, etc.) where blood pressure is an outcome to adopt procedures designed to accurately measure blood pressure rather than those used for diagnostic classification. The best method to accurately measure blood pressure is to carry out 24-h ambulatory measurements. Thus, adequately powered and controlled intervention trials are required to determine the effects of oats on blood pressure.
Glucose and insulin
Impaired fasting glycaemia and impaired glucose tolerance are major risk factors for type 2 diabetes, and are strongly associated with an increased risk of CVD and all-cause mortality( Reference Barr, Zimmet and Welborn 93 ). The present review indicates that interventions with oats or oat bran do not affect fasting glycaemia or insulin concentration. Similarly, evidence to date suggests that markers for insulin resistance (homeostasis model assessment) or sensitivity that use algorithms including fasted glucose and insulin concentrations are also unchanged after intervention with oats or oat bran.
Other outcomes
Many inflammatory markers (including C-reactive protein, IL-6 and soluble intercellular adhesion molecule 1 (ICAM-1)) have been linked to CVD risk, but only high-sensitivity C-reactive protein is currently considered an independent marker of CVD risk( Reference Pearson, Mensah and Alexander 94 ). Observational studies suggest that a high dietary fibre intake may reduce C-reactive protein levels( Reference Ajani, Ford and Mokdad 95 , Reference Ma, Griffith and Chasan-Taber 96 ). However, only a few intervention studies reported the effect of long-term consumption of oats and oat bran on inflammatory markers or markers of endothelial dysfunction (von Willebrand factor, arterial stiffness and fibrinogen). None reported changes in these markers with increased oat consumption, suggesting that the benefits of oats on CVD are unlikely to be mediated by the modulation of these markers. However, more studies are needed to confirm the lack of effect, or otherwise, of oats on these putative markers.
Other systemic compounds that have been linked to an increased CVD risk include homocysteine( Reference Selhub 97 ), plasminogen activator inhibitor-I( Reference Raiko, Oikonen and Wendelin-Saarenhovi 98 ) and factor VII( Reference Noto, Barbagallo and Cefalu’ 99 ). However, the studies that examined the effects of increased oat consumption on these markers are scarce. One study reported measuring plasma homocysteine and found that concentrations decreased by 16 % in response to 12 weeks of oat bran( Reference Sturtzel, Dietrich and Wagner 44 ). A Danish study( Reference Kristensen and Bugel 33 ) found that plasminogen activator inhibitor-I and factor VII levels decreased significantly by 27 and 7 %, respectively, following a 2-week oat bran v. a low-fibre diet. No other studies reported measuring these outcomes, which deserve further investigation.
Weight gain is associated with an increased risk of high blood pressure and hyperlipidaemia. Whilst some studies suggest that increased oat consumption may aid weight loss( Reference Anderson, Story and Sieling 16 , Reference Hegele, Zahariadis and Jenkins 69 ) and reduce waist circumference( Reference Maki, Beiseigel and Jonnalagadda 82 , Reference Zhang, Li and Song 84 ), the majority of studies reviewed herein found no significant effect of oat consumption on weight( Reference Abrahamsson, Goranzon and Karlstrom 15 , Reference Anderson, Spencer and Hamilton 17 – Reference Bremer, Scott and Lintott 20 , Reference Demark-Wahnefried, Bowering and Cohen 22 , Reference Hegsted, Windhauser and Morris 24 – Reference Judd and Truswell 26 , Reference Kelley, Hoover-Plow and Nichols-Bernhard 28 – Reference Kirby, Anderson and Sieling 30 , Reference Kristensen and Bugel 33 , Reference Marlett, Hosig and Vollendorf 36 , Reference Onning, Akesson and Oste 38 – Reference Romero, Romero and Galaviz 40 , Reference Spiller, Farquhar and Gates 42 , Reference Turnbull and Leeds 45 , Reference Beck, Tapsell and Batterham 48 – Reference Davy, Melby and Beske 50 , Reference Frank, Sundberg and Kamal-Eldin 52 , Reference Keenan, Pins and Frazel 55 – Reference Uusitupa, Miettinen and Sarkkinen 58 , Reference Kemppainen, Heikkinen and Ristikankare 60 – Reference Van Horn, Moag-Stahlberg and Liu 64 , Reference Charlton, Tapsell and Batterham 66 – Reference Kashtan, Stern and Jenkins 68 , Reference Lepre and Crane 73 – Reference Theuwissen and Mensink 76 , Reference Berg, Konig and Deibert 79 , Reference Maki, Beiseigel and Jonnalagadda 82 – Reference Johnston, Reynolds and Patz 85 , Reference Van Horn, Liu and Parker 87 , Reference Wolever, Tosh and Gibbs 88 ), BMI( Reference Abrahamsson, Goranzon and Karlstrom 15 , Reference Robitaille, Fontaine-Bisson and Couture 39 – Reference Saltzman, Das and Lichtenstein 41 , Reference Stewart, Neutze and Newsome-White 43 , Reference Davy, Davy and Ho 49 , Reference Davy, Melby and Beske 50 , Reference Uusitupa, Miettinen and Sarkkinen 58 , Reference Kemppainen, Heikkinen and Ristikankare 60 , Reference Van Horn, Moag-Stahlberg and Liu 64 , Reference Van Horn, Liu and Gerber 78 , Reference Berg, Konig and Deibert 79 , Reference Zhang, Li and Song 84 ) or waist circumference( Reference Robitaille, Fontaine-Bisson and Couture 39 , Reference Beck, Tapsell and Batterham 48 – Reference Davy, Melby and Beske 50 ). In order to assess the effect of oats on body weight, it is necessary to also consider the energy and macronutrient content of the intervention diets, which is beyond the scope of the present review. Whilst oats may be used to displace other (more energy dense) foods in the diet, their effect on hunger and satiety is not clear. Although three( Reference Beck, Tapsell and Batterham 48 , Reference He, Streiffer and Muntner 67 , Reference Keenan, Wenz and Myers 86 ) of the studies included in the present review found no significant effect of oats on satiety, hunger or appetite, positive comments from one study( Reference Beck, Tapsell and Batterham 48 ) included ‘feeling more full, for longer’ and ‘less peaks & lows’ in intake. However, satiety is an acute physiological effect of a single meal intake and does not necessarily equate to longer-term changes in dietary habits, which could result in weight loss and/or reduced weight gain. Therefore, the measurement of satiety cannot substitute for longer-term intervention studies measuring body weight and/or composition. The European Food Safety Authority Panel on Dietetic Products, Nutrition and Allergies (NDA) recently concluded that a cause-and-effect relationship has not been established between the consumption of β-glucans from oats and barley and a sustained increase in satiety leading to a reduction in energy intake( 100 ). However, this aspect requires further investigation.
Whilst advice to increase oat consumption is likely to have beneficial health effects, it should be noted that relatively minor side effects (which may only be initial or intermittent) may include flatulence( Reference Abrahamsson, Goranzon and Karlstrom 15 , Reference Stewart, Neutze and Newsome-White 43 , Reference Swain, Rouse and Curley 57 , Reference Uusitupa, Ruuskanen and Makinen 59 , Reference Johnston, Reynolds and Patz 85 ), abdominal distension or bloating( Reference Bremer, Scott and Lintott 20 , Reference Stewart, Neutze and Newsome-White 43 , Reference Swain, Rouse and Curley 57 , Reference Kashtan, Stern and Jenkins 68 , Reference Lepre and Crane 73 ), diarrhoea or loose stools( Reference Maki, Galant and Samuel 35 , Reference Swain, Rouse and Curley 57 , Reference Kashtan, Stern and Jenkins 68 ), and abdominal pain or cramping( Reference Beck, Tapsell and Batterham 48 , Reference Swain, Rouse and Curley 57 ). Taking such side effects into consideration, one study( Reference Stewart, Neutze and Newsome-White 43 ) found that 50 g/d of oat bran was considered ‘acceptable long term’ by 76 % of participants and ‘unacceptable long term’ by 24 % of participants.
Limitations
The majority of studies identified by the present review were relatively small and did not have sufficient power to detect an effect: only twenty-three of the seventy-six articles reviewed (30 %) described carrying out a sample size or power calculation. For many of the variables (such as total cholesterol, LDL-cholesterol, ICAM1, apo and glucose), variation among individuals has been found by other authors to be about 10–20 %. Baseline covariate adjustment should reduce this to 5–10 %. This means that sixty subjects per group should give sufficient experimental power (90 %) to detect intervention effects of 5–7 %. Even less variation is expected in total cholesterol (sd approximately 0·25 mm, range 5–6 mm), so that sixty subjects per group will provide sufficient power to detect differences of 0·2 mm. Larger sample sizes would be required to assess intervention effects on blood pressure and inflammatory markers.
Ideally, a meta-analysis would be carried out to assess whether oats have a significant effect on the outcomes reported in the present review, and if so the size of this effect. However, the authors concluded that a meaningful summary estimate could not be obtained by meta-analysis for several reasons. First, the studies were too heterogeneous. The amount and type of oat products used were varied, and the comparison/control groups included a range of different treatments, for example, refined wheat, whole-wheat products, rice bran, psyllium, farina, fruit or no control. The duration of the studies (from 2 weeks to 6 months) and the initial blood cholesterol concentration of the subjects were also varied. Secondly, many studies were considered of poor quality: 61 % of articles had a low modified Jadad score, and 59 % of studies had less than thirty subjects in the oat intervention group. Thirdly, the outcomes were reported inconsistently among studies, e.g. mean absolute difference, or percentage change, or simply a line in the text to say that there was no significant effect. Some changes were compared with a control group v. baseline, and some results were adjusted for confounding factors whereas others were not.
Furthermore, the present review did not consider the appropriateness of the control group, changes in body weight, energy intake and macronutrient intake during the intervention, or compliance with the intervention – all of which could impact the response to the intervention and thus a meta-analysis summary estimate. The question regarding what could be considered as an ideal control group is important, and depends on the study aim as well as the primary outcomes of the trial. The results tables highlight the disparity of control groups used as comparators in previous studies. Guar gum, undefined control diet, usual diet, wheat- or rice-based products, β-glucan-enriched products as well as products based on specific parts of the grain such as bran and not the whole-grain product have been used. Some designs involved increasing total whole-grain intake without substitution of existing dietary components. An ideal control group should at least consider unchanged total energy intake during the intervention, substituting whole-grain food items with a similar amount of refined cereal products (white breads, etc.). The level of oats/whole-grain intake in the control group should match the lowest quartile of consumption observed in the population studied. When trials aim to identify the active parts or components of the grain, a positive control (with whole grain) should also be included. Further analysis of the studies reviewed herein is required before sufficiently homogenous studies can be chosen for inclusion in a meta-analysis to obtain both a precise and meaningful estimate of the magnitude of the effect of oat consumption on CVD risk markers.
The lack of significant effects in some studies may have been due to the factors mentioned earlier, or could be due to the fact that the response may be modified by other factors. Some studies carried out sub-group analysis or tested for interaction (effect modification), e.g. by sex( Reference Kerckhoffs, Hornstra and Mensink 29 , Reference Noakes, Clifton and Nestel 37 , Reference Onning, Akesson and Oste 38 , Reference Frank, Sundberg and Kamal-Eldin 52 , Reference Van Horn, Emidy and Liu 83 ), sex and age group( Reference Keenan, Wenz and Myers 86 ), BMI group( Reference Maki, Galant and Samuel 35 ), ethnicity (Caucasians v. non-Caucasians)( Reference Wolever, Gibbs and Brand-Miller 89 ), genotype( Reference Uusitupa, Ruuskanen and Makinen 59 , Reference Hegele, Zahariadis and Jenkins 69 ), amount of target dose consumed( Reference Uusitupa, Ruuskanen and Makinen 59 ) and baseline total cholesterol level( Reference Romero, Romero and Galaviz 40 , Reference Van Horn, Emidy and Liu 83 ). The results of such analyses were not considered in the present review; however, these need to be further assessed in larger studies with sufficient power for subgroup analyses or assessing effect modification.
The present review only considered the effect of oats on fasting lipids, glucose and insulin. However, regular consumption of oats may alter the postprandial concentrations. For example, Anderson et al. ( Reference Anderson, Gilinsky and Deakins 18 ) showed that 110 g/d oat bran for 21 d significantly lowered postprandial serum total cholesterol and TAG concentrations v. a control diet. However, Kirby et al. ( Reference Kirby, Anderson and Sieling 30 ) found that a similar amount of oat bran (100 g/d) for a shorter time period (at least 10 d) did not significantly affect postprandial serum total cholesterol or TAG, when measured at hourly intervals throughout the day when compared with a control diet. A 12-week trial of oat consumption significantly lowered the mean peak insulin and incremental area under the insulin curve response (both by 7 %) compared with a control group, but there was no significant change in peak glucose or incremental area under the glucose curve( Reference Maki, Galant and Samuel 35 ). Other studies found that oats or oat bran did not significantly affect postprandial glucose or insulin responses( Reference Kirby, Anderson and Sieling 30 , Reference Maki, Davidson and Witchger 34 , Reference Kestin, Moss and Clifton 56 ). The efficacy of oats and barley products to lower postprandial blood glucose concentration has been reviewed recently( Reference Tosh 13 ). The author concluded that intact grain, as well as barley and oat products containing at least 4 g of β-glucan and 30–80 g of available carbohydrate can significantly reduce postprandial glucose concentrations. The health benefit of reducing postprandial glycaemia is still debatable, but a statement recently issued by the European Food Safety Authority indicates that ‘the reduction of postprandial glycaemic responses (as long as insulinaemic responses are not disproportionally increased) may be a beneficial physiological effect’( 100 ).
Conclusions
Regular consumption of oats or oat bran has a beneficial effect on total cholesterol and LDL-cholesterol, particularly in hypercholesterolaemic subjects. The intervention trials described in the present review can generally be divided into three groups depending on the product used in the intervention: oat bran; whole-grain oat cereals; oatmeal. For the studies that showed a significant reduction in total cholesterol and/or LDL-cholesterol, the range of doses used was 25–135 g/d for oat bran, 45–90 g/d for whole-grain oat cereals and 60–150 g/d for oatmeal. So it appears that the form of oats does not really affect the outcome. The doses required to reach a significant effect were also similar. However, studies using amounts below 50 g/d are scarce, and more well-designed dose–response studies are needed to confirm the minimum amount required to have a clinical beneficial effect. The 3–6 % cholesterol reduction described in the larger studies would translate to a 6–18 % decrease in CHD risk. Some studies reported significant effects on blood cholesterol only 2 weeks after beginning the intervention, so it is likely that the benefits of increasing oats intake start very shortly after changing the diet. How long these effects on blood cholesterol remain if subjects revert to their original diet remains to be determined. However, there is no indication that it would significantly modulate insulin sensitivity. It is still unclear whether increased oat consumption would significantly affect other risk markers for CVD risk. More comprehensive, properly controlled intervention trials with adequate sample sizes are required to answer this question. The present review also highlighted the heterogeneity of treatments used as a control and notes the importance of carefully defining appropriately controlled interventions.
Supplementary material
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S0007114514002281
Acknowledgements
The authors thank M. Mowett for sourcing the majority of the articles. F. T. reviewed articles for inclusion and drafted the paper. L. F. M. carried out the literature search, extracted the data and contributed to writing the paper, P. B. and P. K.-E. contributed to writing the paper.
F. T., P. K.-E. and P. B. received an honorarium from Quaker Oats Company (a subsidiary of PepsiCo) for attending the workshop in May 2012 to discuss the content of the supplement and the University of Aberdeen received an unrestricted grant from Quaker Oats Company. L. F. M. has no conflict of interest to report.
This paper was published as part of a supplement to British Journal of Nutrition, publication of which was supported by an unrestricted educational grant from Quaker Oats Co. (a subsidiary of PepsiCo Inc.). The papers included in this supplement were invited by the Guest Editor and have undergone the standard journal formal review process. They may be cited.
The Guest Editor to this supplement is Roger Clemens. The Guest Editor declares no conflict of interest.
