Approximately 1·6 billion adults are overweight globally and 400 million of these are obese( 1 , Reference Thompson, Holdman and Janzow 2 ). Obesity can lead to serious health consequences, including CVD, type 2 diabetes and some cancers, which places considerable burden on health services. Weight reduction can improve the health of overweight individuals. Even modest reduction of 5 % in body weight (3 kg with a body weight of 65 kg) can significantly reduce the risk of disease( Reference Blackburn 3 ). While it is generally accepted that energy balance plays the major role in weight management, an increasing number of studies have explored the potential role of Ca in weight loss( Reference Shapses, Heshka and Heymsfield 4 – Reference Harvey-Berino, Gold and Lauber 8 ). The results of several cross-sectional epidemiological studies have indicated that dietary Ca may be associated with lower body weight and/or adiposity, such that a diet rich in Ca may attenuate weight gain( Reference Barba and Russo 6 , Reference Davies, Heaney and Recker 9 – Reference Parikh and Yanovski 11 ). Possible mechanisms have also been explored( Reference Barba and Russo 6 , Reference Major, Chaput and Ledoux 12 ) and include the potential to increase fecal fat excretion( Reference Christensen, Lorenzen and Svith 13 , Reference Jacobsen, Lorenzen and Toubro 14 ), the suppression of calcitriol levels( Reference Zemel 15 ) and improving insulin sensitivity( Reference Szollosi, Nenquin and Aguilar-Bryan 16 , Reference Pereira, Jacobs and Van Horn 17 ), which has been related to reducing fat deposition. Few randomised controlled trials have been undertaken to specifically investigate if increasing dietary Ca reduces body weight( Reference Shapses, Heshka and Heymsfield 4 , Reference Harvey-Berino, Gold and Lauber 8 , Reference Yanovski, Parikh and Yanoff 18 , Reference Angeles-Agdeppa, Capanzana and Li-Yu 19 ).
This review comprehensively and systematically analyses the available evidence assessing the effects of dairy and added Ca supplementation on body weight. This extends an earlier systematic review of thirteen studies published up to 2004( Reference Trowman, Dumville and Hahn 7 ), which found no significant effect of Ca on weight loss. Since 2004, several additional published studies have reported on the effects of increased Ca intake (through Ca or dairy supplementation) on changes in body weight and/or body composition, including twenty-seven randomised controlled trials with 4000 participants. Whilst there has been three subsequent reviews( Reference Onakpoya, Perry and Zhang 20 – Reference Chen, Pan and Malik 22 ), these have limited generalisability due to the strict inclusion criteria. Onakpoya et al. ( Reference Onakpoya, Perry and Zhang 20 ) conducted a meta-analysis on a subset of seven studies, each providing Ca supplementation that had body weight as the primary outcome and found that Ca supplementation resulted in a small but significant reduction in body weight and body fat. However, this study excluded several studies (total of 1963 participants) that had bone health or other factor as a primary outcome. In contrast, two meta-analyses have been published since 2012 on dairy food supplementation and body weight and found no overall effect on body weight in the long term( Reference Abargouei, Janghorbani and Salehi-Marzijarani 21 , Reference Chen, Pan and Malik 22 ). The limitations of these reviews include the narrow selection of included studies (half of available published studies)( Reference Abargouei, Janghorbani and Salehi-Marzijarani 21 ); and inclusion of studies with intervention periods of very short duration ( < 12 weeks' duration)( Reference Chen, Pan and Malik 22 ). This systematic review encompasses a broader range of studies with the aim of addressing these limitations. Meta-analysis was used to determine if increased Ca in the form of Ca supplementation or increased dairy food intake consumed for at least 12 weeks leads to a decrease in body weight.
Methods
Literature search
Electronic databases Medline, EMBASE and CINAHL were searched from 1994 to December 2014 as well as the Complete Cochrane Library Database of Controlled Trials from 1800 (earliest record) to December 2014. Search terms included ‘calcium’ or ‘dairy’ with ‘weight’ or ‘composition’ or ‘fat’. These terms encompassed ‘calcium supplement(s)’, ‘calcium supplementation’, ‘dairy products’, ‘milk’, ‘cheese’, ‘yoghurt’, ‘body weight’, ‘body fat’ and ‘body composition’ (online Supplementary Fig. S1). Only articles published in full length and in English were selected.
Study quality, inclusion and exclusion criteria
Inclusion criteria included: randomised controlled trial of Ca supplements or increased dairy products and intervention in adults over the age of 18 years that assessed changes in body weight, with or without data on body composition. To ensure quality of included studies, studies were only included if the intervention length was at least 12 weeks; Ca supplementation was of at least 300 mg/d; increased dairy intake provided an increase of at least 300 mg of Ca/d; and Ca intake was at least 300 mg higher than the control group. Studies in which the Ca was supplemented in the form of non-dairy-based fortified foodstuffs or powder were included in the same category as Ca supplements. Studies in pregnant or lactating women were excluded as well as populations with severe illnesses such as cancer.
Study selection
A summary of study selection and exclusion is outlined in Fig. 1. Abstracts were reviewed independently by at least two reviewers (including C. E. H., N. W. and A. O. B.). Any disputes were assessed by a fourth reviewer (C. A. N.).
Flow chart of process of study inclusion and exclusion.
Five studies were excluded that did not report body weight at two time points( Reference Dawson-Hughes, Harris and Krall 23 – Reference Caan, Neuhouser and Aragaki 28 ) and one( Reference Dawson-Hughes, Harris and Krall 23 – Reference Caan, Neuhouser and Aragaki 28 ) reported in graph format that could not be estimated. Two of these studies( Reference Gunther, Legowski and Lyle 29 , Reference Manios, Moschonis and Koutsikas 30 ) reported body weight at study completion in related studies already found in the initial search and were included in the analysis. Of the remaining four studies that reported body weight at baseline but not study completion, authors were contacted for data to include in the analysis and one author was able to provide the relevant data on body weight( Reference Dawson-Hughes, Harris and Krall 23 ); thus, this study was added back into the included studies. Twenty-nine studies with thirty-two trial arms provided information on body composition.
Data extraction
Relevant data were extracted and, wherever necessary, weight was converted and reported as kg. Where mean changes were reported graphically with variance( Reference Reid, Ames and Mason 31 ), the graph was magnified 600 × and values were estimated by the authors (N. W.). Where studies reported data separately for males and females, the data were combined( Reference Higgins and Green 32 , Reference Barr, McCarron and Heaney 33 ). For studies that reported body weight at exit, change from baseline standard deviation was imputed using a correlation coefficient( Reference Borenstein, Hedges and Higgins 34 ) (Cochrane Handbook for Systematic Reviews of Interventions 16.1.3.2)( Reference Higgins and Green 32 ). Where mean change data were not reported in the required format, the authors were contacted( Reference Barba and Russo 6 ). A correlation value of r 0·6 was used as it is considered a conservative estimate of correlation between baseline and end body weights.
Statistical analysis
Review Manager (RevMan, version 5.3) was used to conduct meta-analyses with the change in body weight (kg) and body fat (kg) as the outcome. Separate analyses were conducted by intervention type: dairy food or Ca supplement. Subgroup analysis was performed for studies that included dietary energy restriction. Where variance was reported as se or CI, sd was calculated. Random effects models in which the trials were assumed to be a random sample from a population of trials (both actual and potential) were used( Reference Borenstein and Higgins 35 ). All treatment effects are presented with 95 % CI. A funnel plot was used to detect the possibility of publication bias, and Egger's regression test( 36 ) to measure funnel plot asymmetry was performed using SAS (version 9.3; SAS Institute, Inc.).
Results
Forty-one studies were identified that met the inclusion criteria. Of these, ten used Ca supplementation only, twenty-five used dairy product supplementation only and six used both Ca and dairy product supplementation. Study characteristics are summarised in Table 1 (Ca supplementation) and Table 2 (dairy product supplementation). Of the forty-one studies, fifty-one intervention trial arms were included; in thirty-three studies, two trial arms were included (intervention, control); in six studies, three trial arms were included (one dairy intervention group, one Ca supplement group and a control)( Reference Manios, Moschonis and Koutsikas 30 , Reference Reid, Ames and Mason 31 , Reference Barr, McCarron and Heaney 33 , Reference Zemel, Thompson and Milstead 37 – Reference Zemel, Teegarden and Loan 40 ); in two studies, four trial arms were included (one included two Ca supplement groups, one dairy group plus one control group and the other was divided into three supplement groups plus one control group)( Reference Shapses, Heshka and Heymsfield 4 , Reference Wagner, Kindrick and Hertzler 41 ).
Study characteristics (intervention with calcium supplements)
NA, not available.
* Numbers in parentheses indicate number in the intervention group.
† Calcium gluconate.
‡ Calcium lactobionate.
§ Reported body composition, method not stated.
∥ Calcium carbonate.
¶ Measured body composition via dual-energy X-ray absorptiometry.
** Calcium citrate.
†† Calcium citrate malate.
‡‡ Calcium phosphate.
§§ Calcium lactate gluconate.
Study characteristics (supplementation with dairy products)
NS, not stated; f, female; m, male.
* Numbers in parentheses indicate number in the intervention group.
† A mean age range is provided for studies that presented mean age separately for each group (control and intervention).
‡ Milk low fat or skim.
§ Yoghurt low fat or skim.
∥ Cheese low or reduced fat.
¶ Measured body composition via dual-energy X-ray absorptiometry.
** Prolibra with milk minerals and Ca.
†† Milk type not stated or no preference.
‡‡ Yoghurt type not stated or no preference.
§§ Cheese fat not stated or no preference.
∥∥ High protein, high dairy diet
¶¶ Dairy type not specified.
*** Milk powder low or no fat.
††† Cheese full fat.
‡‡‡ Low or reduced fat custard.
§§§ Ca fortified beverage.
∥∥∥ Cream.
Study characteristics: calcium supplementation
Sixteen studies reported nineteen trial arms with Ca supplementation. Of the nineteen trial arms that used Ca supplements, sixteen used a placebo control (nine of these with energy restriction or weight loss diet), two used usual diet as the control and one used an energy-restricted diet without a Ca supplement.
Of the sixteen studies, five reported bone parameters as the main outcome, two reported body weight as the main outcome, two reported body composition, one reported lipids and the remaining six studies reported more than one main outcome including bone plus weight, bone plus lipids, weight plus body composition, body composition plus lipids and weight, composition and metabolic factors. Of the nineteen trial arms within the sixteen studies, nine incorporated energy restriction, thirteen included obese or overweight participants, ten were in women only, and five were in postmenopausal women. Ten studies were from North America, two were from New Zealand, two were from Europe, and there was one each from Iran and Puerto Rico. Mean study length was 9 months and ranged from 3 to 24 months. Mean age was 43 (sd 14) years and ranged from 20 to 74 years. One study( Reference Jensen, Kollerup and Quaade 42 ) did not report age. The supplements provided a mean of 913 (sd 253) mg of Ca/d, range 350–1500 mg/d.
Study characteristics: dairy food supplementation
Of the thirty-one studies with dairy food supplementation, nineteen included energy-restricted diets, twenty-one included obese or overweight participants, twelve were in females only, and six were in postmenopausal women. Of the thirty-one studies (including thirty-one trial arms of dairy food supplementation), five studies reported bone parameters as the main outcome; five reported body weight as the main outcome; two reported body composition; five reported both body weight and body composition; three reported body composition plus lipids; two reported body weight plus appetite; one reported waist circumference, blood pressure and lipids; and one each of oxidative stress, fat oxidation, biomarkers of inflammation, body weight plus blood pressure, body weight plus bone parameters, body composition plus bone parameters, and waist circumference. One study that reported body weight as the main outcome used protein in milk as the treatment focus, not Ca.
Of the thirty-one trial arms that used dairy supplementation, seven used a usual diet control; nineteen used a low dairy, low Ca or placebo control; one used a moderate dairy control; and four used a special diet (lower in Ca) as the control. In all studies with energy restriction, this restriction was employed in both the intervention and control groups. Twenty were from North America, four were from Asia, three were from Australia, two were from Europe and two were from South America. Mean study length was 7 months with a range from 3 to 24 months. Mean age was 44 (sd 11·5) years and ranged from 20 to 70 years. Mean Ca provided by dairy foods was 1326 (sd 401) mg of Ca/d, range 482–2400 mg of Ca/d.
Twenty-one of the thirty-one studies (68 %) instructed participants to consume low-fat, reduced-fat or no-fat dairy foods and one study supplied a low-fat Ca-fortified beverage containing 7 % non-specified milk( Reference Haub, Simons and Cook 5 ). Seven( Reference Thompson, Holdman and Janzow 2 , Reference Harvey-Berino, Gold and Lauber 8 , Reference Zemel, Thompson and Milstead 37 , Reference Smilowitz, Wiest and Teegarden 39 , Reference Stancliffe, Thorpe and Zemel 43 – Reference Tanaka, Uenishi and Ishida 45 ) of the remaining nine studies did not specify dairy fat content, one( Reference Zemel, Teegarden and Loan 40 ) instructed participants to choose either low- or full-fat dairy foods, and the other( Reference Barba and Russo 6 ) specified daily energy from fat at 30 % in both the dairy supplemented and control groups.
Meta-analysis
Calcium supplementation and body weight
The Ca supplement studies had a combined total of 2711 participants. Fig. 2(a) shows the association between Ca supplementation and body weight change. The pooled analysis showed that there was no difference in weight change between the intervention and control groups ( − 0·17, 95 % CI − 0·70, 0·37 kg, P= 0·54).
(a) Association between calcium supplementation and change in body weight and (b) separated for trials with energy-restricted diets and (c) calculating body fat. (A colour version of this figure can be found online at http://www.journals.cambridge.org/bjn).
Of the Ca supplement studies, eight reported compliance rates. Of these, all reported high compliance of above 75 %( Reference Shapses, Heshka and Heymsfield 4 , Reference Manios, Moschonis and Koutsikas 30 , Reference Reid, Ames and Mason 31 , Reference Palacios, Bertran and Rios 38 , Reference Smilowitz, Wiest and Teegarden 39 , Reference Wagner, Kindrick and Hertzler 41 , Reference Reid, Horne and Mason 46 , Reference Winters-Stone and Snow 47 ). Pooled analysis of these eight studies (n 1701) showed there was no difference in weight change between the Ca supplemented and control groups ( − 0·02, 95 % CI − 0·31, 0·28k g).
Dairy food supplementation and body weight
The dairy food supplement studies had a combined total of 2091 study participants. Fig. 3(a) shows the association between Ca supplementation and change in body weight. The pooled analysis showed that there was no difference in mean weight change for the groups receiving dairy supplementation compared to the control groups ( − 0·06, 95 % CI − 0·54, 0·43 kg).
(a) Association between dairy food supplementation and change in body weight and (b) separated for trials with energy-restricted diets and (c) those that measured body composition and (d) trials with energy-restricted diets that also measured body composition. (A colour version of this figure can be found online at http://www.journals.cambridge.org/bjn).
Measurement of body composition
To assess if there was some beneficial effect of Ca or the inclusion of dairy on body composition, the trials that incorporated measurement of body composition were pooled and analysed (Figs. 2(b) and 3(b)). Ten Ca supplement studies (with twelve trial arms) and twenty-one dairy food supplement studies (with twenty-one trial arms) reported change in body fat (kg) or body fat at baseline and completion. Of these, thirty measured fat using dual-energy X-ray absorptiometry and one did not specify how body composition was measured( Reference Major, Alarie and Dore 48 ).
Calcium supplementation and body composition
The ten Ca supplement studies with twelve trial arms had a combined total of 2350 participants. Fig. 2(c) shows the association between Ca supplementation and change in body fat (kg). Pooled analyses showed there was no difference in mean body fat change compared to the control groups ( − 0·19, 95 % CI − 0·51, 0·13 kg, P= 0·24).
Dairy food supplementation and body composition
Twenty-one dairy food supplement studies had a combined total of 1289 study participants. Fig. 3(c) shows the association between dairy food supplementation and change in body fat (kg). The pooled analysis showed that there was no difference in mean body fat change for the groups receiving dairy supplementation compared to the control groups ( − 0·36, 95 % CI − 0·80, 0·09 kg, P= 0·81).
Effect of calcium and dairy intervention combined with energy restriction on body weight
To assess if there was some beneficial effect of Ca or the inclusion of dairy products in assisting with weight loss, the trials that incorporated energy restriction or encouraged weight loss were pooled and analysed (Figs. 2(b) and 3(b)). The nine Ca supplement studies with energy restriction had a combined total of 350 study participants with a mean study length of 4 months. There was no difference in mean weight change compared to control groups ( − 0·34, 95 % CI − 1·15, 0·46 kg).
Nineteen dairy food supplement trials with energy restriction had a combined total of 1010 participants. These are relatively short-term studies, with a mean study length of 4 months (range 3–12 months). Average dairy intake of Ca was 1318 mg/d, equivalent to approximately 3 servings of dairy/d. There was no difference in weight change between the intervention and control groups when combined with dietary energy restriction ( − 0·32, 95 % CI − 0·93, 0·30 kg, P= 0·31).
Effect of dairy food supplementation combined with energy restriction on body composition
The effect of dairy food supplementation on body fat when combined with energy restriction was analysed (Fig. 3(d)). Thirteen dairy supplement studies with a combined total of 564 participants (mean age 38 (sd 7·0) years) and a mean study length of 4 months were included in the analysis. Fat loss was significantly greater in the dairy supplemented group compared to the control, when combined with energy restriction ( − 0·96, 95 % CI − 1·46, − 0·46 kg, P < 0·001).
Test for publication bias
There was no evidence for publication bias among the Ca supplementation (Egger's regression P= 0·21) (Fig. 4(a)) or dairy supplement trials reporting body weight (Egger's regression P= 0·07) (Fig. 4(b)).
Publication bias was assessed using a funnel plot depicting the mean difference (MD) between the intervention and control groups (treatment effect) against the standard error (se) of the mean difference. (a) Funnel plot of all randomised controlled trials (RCT) of calcium supplementation reporting body weight, Egger's regression, P= 0·21 indicates no publication bias. (b) Funnel plot of all RCT of dairy food supplementation reporting body weight, Egger's regression, P= 0·07 indicates no publication bias. Mid-vertical line represents the zero mean difference or zero effect size. (A colour version of this figure can be found online at http://www.journals.cambridge.org/bjn).
For the trials reporting body composition, there was no statistical evidence of publication bias among the Ca supplement studies (Egger's regression P= 0·40) or the dairy supplement studies (Egger's regression P= 0·10) (online Supplementary Fig. S2 for funnel plots).
Discussion
To determine if increasing dietary Ca could reduce body weight or body fat, a meta-analysis of randomised controlled trials assessing whether Ca supplements or increased Ca from dairy foods resulted in weight or fat loss was undertaken. These analyses show that increased Ca provision in the form of dairy foods or Ca supplements does not increase weight or fat loss compared to control groups. This finding differs from Onakpoya et al. ( Reference Onakpoya, Perry and Zhang 20 ) who conducted a meta-analysis on seven Ca supplementation studies and found that Ca supplementation resulted in a small but significant reduction in body weight and body fat. However, this study included only seven studies that had body weight as the primary outcome, whereas the present study included all studies (n 16) that provided increased Ca in the form of Ca supplementation that also measured body weight or body composition at two time points.
The limitations of these reviews include the narrow selection of included studies (half of available published studies)( Reference Abargouei, Janghorbani and Salehi-Marzijarani 21 ); and inclusion of studies with intervention periods of very short duration ( < 12 weeks duration)( Reference Chen, Pan and Malik 22 ).
Contrary to two previous meta-analyses( Reference Abargouei, Janghorbani and Salehi-Marzijarani 21 , Reference Chen, Pan and Malik 22 ), the current analysis found that consumption of about 3 servings of reduced fat dairy foods on an energy-restricted diet did not result in a greater mean weight change compared to the control. However, the current meta-analysis had broader selection criteria, which resulted in additional studies being included( Reference Abargouei, Janghorbani and Salehi-Marzijarani 21 ), and strict inclusion criteria (studies with >12 weeks duration) resulting in a more robust analysis( Reference Chen, Pan and Malik 22 ). Although cross-sectional studies have found consistent associations between high dairy consumption and lower body weight and BMI( Reference Barba and Russo 6 , Reference Davies, Heaney and Recker 9 – Reference Parikh and Yanovski 11 ), these are subject to confounding, and the overall impact on body weight with increased dairy consumption from longitudinal studies have been inconsistent.
In support of the two previous analyses, the current analysis did find that consumption of about 3 servings of reduced-fat dairy foods, but not Ca supplements, on an energy-restricted diet resulted in a greater mean fat loss of 920 g over 4 months in twenty-one studies. These results indicate the importance of separating out trials that are designed to facilitate fat loss from those where energy intake is not restricted.
It appears that consuming 3 servings of reduced-fat dairy products each day may assist in appetite control, although this is probably not related to amount of dietary Ca and is more likely to be related to other components present in dairy products. Dairy products are high in protein and increasing dietary protein intake increases satiety and can lead to weight loss( Reference Dove, Hodgson and Puddey 49 – Reference Wycherley, Moran and Clifton 51 ). In addition to protein, there is evidence that especially casein proteins present in dairy products have a long-term satiating effect in humans( Reference Bendtsen, Lorenzen and Bendsen 52 ). In addition, the increase in dietary protein resulting from increased consumption of dairy products may be assisting in maintaining lean mass when combined with resistance training( Reference Barba and Russo 6 ). The specific mechanisms whereby consumption of dairy product within the context of an energy-restricted diet may assist with fat loss are not known. Importantly though, the findings indicate only a small difference in fat change over a short time frame and whether this is likely to continue over the long term remains to be determined. For most weight loss interventions, the greatest effects are usually seen early in the intervention period. However, the two studies( Reference Thompson, Holdman and Janzow 2 , Reference Harvey-Berino, Gold and Lauber 8 ) in this category that were conducted for longer than 6 months did not show a significant difference in fat change between the intervention and control groups, suggesting that the effect of increased consumption of dairy foods on fat loss may attenuate over time.
While this meta-analysis did not find an indication that Ca supplement had an effect on body weight or fat mass, there is some evidence that consuming additional supplementary Ca can have a small effect in increasing fecal fat excretion, leading to a small net energy loss; however, this effect has been predicted to be less than 1 kg/year provided no other counter-regulatory effects are present( Reference Christensen, Lorenzen and Svith 13 ). It has been suggested that Ca supplementation may aid weight loss if usual Ca intake is low( Reference Tremblay and Gilbert 53 ). However, there were too few studies within the current meta-analysis that combined low usual Ca intakes with energy restriction for a separate analysis. It has also been suggested that Ca combined with vitamin D may reduce visceral body fat( Reference Rosenblum, Castro and Moore 54 ); however, this could not be explored in the present study as there were no studies that provided vitamin D and also assessed for visceral fat.
Strengths and limitations
Only studies with intervention periods of 12 weeks or more that also resulted in an increased intake of at least 300 mg of Ca/d were included. Nine of sixteen supplementation studies and twenty-one of thirty-one trials of dairy supplementation had body weight or body composition as a key outcome, with the remaining studies having bone or stress as the key outcome with change in body weight as a secondary outcome, and it could be argued that these studies were not designed to evaluate weight change. However, weight change is easy to measure and is reproducible, which is necessary for the interpretation of studies evaluating bone outcome; therefore, the fact that it was not the primary measure should not influence these results. Another potential limitation may be the heterogeneity of the studies included; however, this was attenuated by using a random effects model. The included studies were predominantly Caucasian populations and these results may be different in other populations, particularly those consuming very low levels of dietary Ca. It is possible that other confounding factors may influence a relationship between dairy and body composition or body weight that were unable to be taken into consideration, such as physical activity and other dietary factors.
In summary, a robust meta-analysis method with large subject numbers showed that there was no evidence that increased Ca provision in the form of supplements or dairy foods reduces body weight or body fat. There was some evidence that consumption of approximately 3 servings/d of reduced fat dairy foods (approximately 1300 mg/d Ca) in the presence of energy restriction resulted in a small, but significant greater loss of body fat mass over a short period of 4 months. This indicates that low fat dairy foods can be included as part of a healthy weight loss diet without having negative effect on body weight or body composition in the short term( Reference Booth and Nowson 55 ).
Supplementary material
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S0007114515001518
Acknowledgements
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
None of the authors has a conflict of interest to declare.
C. A. N. and C. E. H. were responsible for the concept and design. N. W. and A. O. B. performed the literature search, study selection, data extraction and statistical analysis. A. O. B. wrote the original manuscript. All authors contributed to analysis and interpretation of the data and critical revision of the final manuscript. C. A. N. had final responsibility for the decision to submit for publication.
