There is considerable evidence that a high dietary salt intake is a major contributor to raised blood pressure and that reduction of salt intake at the population level has the potential to achieve meaningful reductions in cardiovascular events and medical costs(Reference He and MacGregor1, Reference Bibbins-Domingo, Chertow and Coxson2). While intervention studies with hard end points to determine the efficacy of salt reduction are difficult and expensive to conduct, there has been one long-term follow-up study 10–15 years after the end of two randomised intervention studies, the Trials of Hypertension Prevention (TOHP) Phase I and TOHP Phase II(Reference Cook, Cutler and Obarzanek3). A 25 % lower risk of a cardiovascular event was observed among those in the intervention group despite relatively small reductions in Na intake and a 30 % lower risk after adjustment for baseline Na excretion and weight. The Na reductions in the intervention groups were relatively small at 44 mmol/24 h (approximately 2·6 g NaCl/d) and 33 mmol/24 h (approximately 2 g NaCl/d) in TOHP Phase I and Phase II, respectively(Reference Whelton, Appel and Charleston4, 5).
There are other harmful effects of a high Na intake including impaired endothelial function, which is associated with the development of atherosclerosis, stiffening of the arteries as measured by increased pulse wave velocity and left ventricular hypertrophy(Reference Dickinson, Keogh and Clifton6–Reference Langenfeld and Schmieder9). A high Na intake has also been shown to have a negative impact on bone density(Reference Devine, Criddle and Dick10).
In Australia, the National Health and Medical Research Council (NHMRC) states that the Upper Level (UL) of intake should be no more than 100 mmol (2300 mg) Na/d (approximately 6 g NaCl/d) for the population and 70 mmol (1600 mg) Na/d (approximately 4 g NaCl/d) for chronic disease prevention(11).
The most recent population data on Na intake as assessed by urinary Na excretion in Australia are from the 1995 Hobart Salt Study, which showed that men had an excretion of 170 mmol Na/d (approximately 10 g NaCl/d) and women 118 mmol Na/d (approximately 7 g NaCl/d)(Reference Beard, Woodward and Ball12). More recently, Charlton et al. (2010) observed that Na excretion was equivalent to 111 mmol Na/d (approximately 7 g NaCl/d) in a group of women living in Wollongong, New South Wales, Australia(Reference Charlton, Yeatman and Houweling13). In contrast, recent data from 24 h urine collections from overweight volunteers who participated in weight-loss studies at the Commonwealth Scientific and Industrial Research Organisation (CSIRO) have shown that Na intake is relatively high in this sample of overweight people at 133 mmol Na/d (approximately 8 g NaCl/d) in women and 183 mmol Na/d (approximately 11 g NaCl/d) in men(Reference Keogh and Clifton14).
In developed countries more than 75 % of dietary Na intake comes from processed foods, making it difficult for an individual to follow a reduced or low Na diet(Reference Anderson, Appel and Okuda15, Reference Brown, Tzoulaki and Candeias16). Webster et al. (2010) collected data on Australian processed foods including 7221 products in ten food groups and found that the food groups highest in Na were sauces and spreads (1283 mg/100 g) and processed meats (846 mg/100 g)(Reference Webster, Dunford and Neal17). They found that cereal and cereal products (206 mg/100 g) and fruit and vegetables (211 mg/100 g) were the lowest in Na. Similar data have been reported from the New Zealand Total Diet Survey 2003–2004(Reference Thomson, Vannoort and Haslemore18, Reference Thomson19). Dietary modelling conducted by Food Standards Australia New Zealand (FSANZ) found that among adults dietary Na intake is estimated to be 1900–2500 mg/d(20). The aim of the present study was to identify the foods and food groups contributing to Na intake in a group of healthy women living in Adelaide. A secondary aim was to measure Na excretion in this group.
Experimental methods
Women living in the community in Adelaide, South Australia were recruited by advertisements in local newspapers to participate in a validation study of the CSIRO FFQ against weighed dietary intakes. That study has been described in detail elsewhere(Reference Lassale, Guilbert and Keogh21). Inclusion criteria were female, age 30–60 years and healthy as determined from a self-reported questionnaire. Exclusion criteria were having undertaken food intake measurement previously, evidence of illness/disease likely to entail dietary modification (e.g. diabetes, cardiovascular, renal, liver or gastrointestinal disease), past history of gastrointestinal surgery affecting dietary intake, medications affecting gastrointestinal motility or hunger/appetite (e.g. metoclopramide, domperidone and anti-cholinergic drugs, i.e. atropine or erythromycin) and non-English speaking. The study was approved by the CSIRO Food and Nutritional Sciences Human Research Ethics Committee and participants gave informed written consent.
Dietary intake methodology
Data collection was undertaken in two phases from September to December 2007 and March to July 2008. Participants attended the CSIRO research facility. They completed two 4 d weighed food records (WFR), 4 weeks apart, which the researchers believed was enough time to forget previous responses and maximise retention of participants in the study. Participants were asked to include three weekdays and one weekend day in each food record. They were given detailed instruction by the researchers on how to weigh and record all foods and fluids contributing to their dietary intake. Participants were asked to record the use of discretionary salt (i.e. salt added in cooking and at table) on their food record (salt cellars were not provided). When they returned the food record they were asked by the researcher to estimate the amount used in comparison to a 1 g sachet of salt. The estimated amount was then included in the dietary analysis. They were provided with printed instructions and record forms. They were asked to be specific about the food eaten (e.g. note the type and brand of food, whether meat was trimmed of fat or not, whether raw or lean weight), to weigh all food or use metric cups or spoons for fluids. They were asked to record all foods eaten, to weigh leftover food and record the amount eaten. They were asked not to deliberately change their eating pattern because they were participating in the study. Electronic weighing scales (Kitchen Craft®; Thomas Plant Ltd, Birmingham, UK) were provided. Each WFR was checked by the researcher with the participant while she was at the research facility for accuracy and for clarification of specific foods or amounts of food consumed if they seemed unusually high or low. WFR data were analysed using a computerised database of Australian foods (Foodworks Professional Edition 2007, version 5) based on the Australian nutrient database NUTTAB 2006(22). NUTTAB 2006 contains data for 2600 foods. The majority of the data are analysed values with some data from overseas food tables, supplied by the food industry or taken from food labels. The development of NUTTAB 2006 has been described by FSANZ(23). The numbers of foods in each set are as follows: Additives and Food Ingredients, Herbs, Seasonings and Spices (25), Beverages, Alcoholic (38), Non-Alcoholic (116), Cereals and Cereal Products, Biscuits (49), Bread and Bread Products (61), Breakfast Cereals (25) Cakes, Slices and Other Battered Products (55), Flours, Grains and Starches (48), Hamburgers, Pizza and Other Takeaway Products (23), Noodles and Pasta (33), Pastries, Pies and Tarts (34), Condiments, Dressings, Pastes and Sauces (57), Spreads (5), Dairy Butter (5), Cheese (35), Cream (12), Ice Cream and Edible Ice Products (11), Milk (26), Yoghurts and Dairy Desserts (14), Dairy and Meat Alternatives (22), Edible Fats and Oils, Edible Oil Spreads (22), Fats and Oils (23), Eggs (19), Fruit (161), Indigenous Foods, Plant Foods (6), Legumes (20), Meat and Meat Products, Beef (138), Game and Other Meat (27), Lamb (121), Mutton (20), Offal (31), Pork (75), Poultry (54), Processed Meats (67), Veal (40), Nuts and Seeds (30), Restaurant Foods, Asian Foods (84), Mediterranean Foods (36), Seafood and Seafood Products, Crustacea and Molluscs (12), Fish (59), Processed Fish, Crustacea and Molluscs (55), Snack Foods (36), Soups (43), Sugar, Confectionery and Sweet Spreads, Chocolate Based (21), Spreads and Toppings (12), Sugar Based (22), Sugars and Sweeteners (16), Vegetables (238).
Urinary sodium excretion
Participants collected two 24 h urine samples during the course of the study. Each collection was performed on 1 d during each of the two WFR. Participants were instructed by the researchers how to carry out a 24 h urine collection and provided with written instructions when the urine bottles were given out. These instructions were reiterated before the second collection. To minimise intrusion and promote compliance with the urine collections participants decided which day suited them best to collect the 24 h urine sample. Urine collection bottles were given to the participants in a carry bag at their visits to the research unit. The bottles did not contain preservative.
Completeness of the 24 h urine sample was verified by a creatinine excretion within the normal range.
Analysis for urinary Na and creatinine was undertaken by a certified laboratory (Institute of Medical and Veterinary Science, Adelaide, Australia). Results were averaged for the two visits.
Food group analysis
Foods from the WFR were categorised into groups (Table 1) similar to those specified by FSANZ, with modifications made to suit the purpose of the study.
Percentage contribution of individual foods and food groups to sodium intake among a sample of healthy women (n 70) living in metropolitan Adelaide, South Australia, 2007–2008
Statistical analysis
Average Na excretion was calculated from the two urinary Na excretions. Average Na intake from the dietary method was calculated from the 8 d of WFR. Data are presented as means and standard deviations and as percentages. Statistical analysis was performed using the statistical software package PASW Statistics 18.
Regression analysis
Prediction intervals from a linear regression of urinary Na excretion v. Na intake from the dietary method were used to assess the accuracy of WFR in estimating Na excretion both at the group level and for individuals.
Agreement analysis
The validity of the dietary intake in measuring Na intakes was tested by tertile classification. Individuals were ranked by tertile for both Na excretion and intake. The overall strength of agreement between tertile rankings was measured with the weighted kappa statistic.
Results
Seventy participants, with a mean age of 48·6 (sd 8·0) years, mean BMI of 28·4 (sd 6·3) kg/m2 and mean energy intake of 8·5 (sd 1·8) MJ/d, participated in the validation study.
Foods and food groups
Na intake from the WFR for the seventy participants who completed a WFR was 2378 (sd 690) mg/d. Proportions of Na intake contributed by the food groups are shown in Table 1. Overall the breads and cereals group contributed 32·5 % of Na in the diet, with bread alone contributing 19·0 %. Meat products and combination dishes contributed 22·8 % of Na intake, while discretionary salt use contributed 1·0 % of Na intake.
Urinary analysis
One participant did not complete the urine collections. Urine analysis was available for sixty-nine participants. Urinary Na excretion was 126 (sd 42) mmol/d. Seventy per cent (n 48) had Na excretion ≥100 mmol/d (146 (sd 34) mmol/d).
Urinary creatinine excretion was 12·0 (sd 2·6) mmol/24 h. Creatinine excretion for all participants was within the reference range of 6–16 mmol/24 h(24).
Urinary sodium excretion v. sodium intake from weighed food records
Urinary Na excretion was measured on sixty-nine women who had also completed WFR, from which daily Na intake was calculated. Average daily dietary Na intake was 2361 (sd 680) mg/d compared with 2898 (sd 966) mg/d (126 (sd 42) mmol/d) from urinary Na excretion (P < 0·001 for difference). Na excretion and intake were moderately correlated (r = 0·5, P < 0·001).
Determinants of differences between the methods
Women with BMI < 25 kg/m2 (n 28) had a smaller difference between Na excretion and intake than those with BMI ≥ 25 kg/m2 (n 41; 215 (sd 726) v. 752 (sd 888) mg Na/d, P = 0·01). There was an inverse relationship between the difference between Na excretion and intake and energy intake, such that it was lower in those reporting a higher energy intake (r = −0·3, P = 0·012). When age, BMI and energy intake were entered in a multivariate equation, only energy intake was independently related to the difference between Na excretion and intake. As urinary Na increased the difference between the two methods increased, with a correlation of r = 0·72 between the two.
To assess the strength of agreement between the two methods, a linear regression was estimated of the test method (WFR) v. the gold standard method (urinary excretion). The linear regression was significant (P < 0·001, R 2 = 0·246). The WFR intake regression coefficient was B = 0·705 (se 0·151).
Tertiles of agreement
When the data were analysed according to tertiles, approximately 42 % (n 24) of participants were in exactly the same tertile (weighted κ = 0·4).
Discussion
As in other studies(Reference Charlton, Yeatman and Houweling13, Reference Margerison, Nowson and Worsley25), the main finding of the present study was that bread alone contributed 19·0 % of Na intake with the breads and cereals group overall contributing 32·5 %. These proportions are similar to the results of a previous study in Adelaide, Australia in overweight and obese men and women with diabetes, in which breads and cereals contributed 29 % of Na intake(Reference Villani, Clifton and Keogh26). In a recent study by Charlton et al. (2010) the bread and cereals group contributed 27 % of Na intake but the proportion that bread alone contributed was not reported(Reference Charlton, Yeatman and Houweling13). Margerison et al. (2004) reported that bread contributed 8 % of the Na intake of the participants in their study and meat products/dishes contributed 17 %(Reference Margerison, Nowson and Worsley25). Comparison with other studies is difficult as the methodology used to create composite food groups varies. For example, the bread and cereals group in Charlton et al.'s study contains bread, cereal, rice, pasta and noodles; whereas in the present study the breads and cereals group contains these foods and in addition noodles, rice, cous cous and sweet foods such as biscuits, buns, cakes, pastries and fruit pies. The sweet foods were included in a grouping of snacks/desserts/extras in the study by Charlton et al., giving different percentage contributions to Na intake from the composite groups(Reference Charlton, Yeatman and Houweling13). Dietary modelling from FSANZ indicates that main contributors to Na intake are bread and bread rolls (25 %), but cereal products and cereal-based dishes (e.g. biscuits and pizza) are dealt with in a different group contributing 17 % and breakfast cereals are a group on their own contributing 4 %(20).
In the present study Na excretion was on average 126 mmol/d, with the majority of participants (70 %) having Na excretion in excess of the NHMRC-recommended UL of intake of 100 mmol Na/d. In the 1995 Hobart Salt Study women had Na excretion of 118 mmol/d and 64 % had Na excretion in excess of the NHMRC-recommended UL, suggesting that Na intakes have not changed since then(Reference Beard, Woodward and Ball12). In contrast, Charlton et al. (2010) found that average Na intake was in line with the NHMRC recommendations and only 43 % of women had an intake in excess of the recommendations(Reference Charlton, Yeatman and Houweling13). However, in the study by Charlton et al. completeness of the urine collection was not verified and it is likely that some of the participants had incomplete urine collections. The population in Charlton et al.'s study had an average BMI within the normal range whereas the population in the present study was on average overweight, which may also contribute to a higher proportion of women exceeding the NHMRC UL. While similar in other respects both studies were small and the discrepancies between them provide support for the need for a larger population study of dietary Na intake.
It is of interest that bread continues to be such a large contributor to Na intake given data from the Australian Longitudinal Study on Women's Health (2009) showing that on average most (75·4 %) women eat two slices or less daily, suggesting that amount of Na in bread remains high. In a recent survey of breads in Australia only four of twenty brands met the Australian Heart Foundation's benchmark of ≤400 mg/100 g, with some popular brands of bread having Na content 30 % higher than the Heart Foundation benchmark(Reference Cleanthous, Mackintosh and Anderson27). Reducing the Na content of bread to the Heart Foundation benchmark of 400 mg/100 g would decrease the average Na excretion by only 5 %. There is clearly an argument to be made for reducing the Heart Foundation benchmark further to achieve more substantial Na reductions. It has been shown recently that the Na content of bread can be reduced by 52 % without affecting bread consumption(Reference Bolhuis, Temme and Koeman28).
The most effective way to reduce Na intake at the population level is by decreasing the Na content of all processed foods(Reference Webster, Dunford and Neal17). Many Australian food manufacturers are already actively doing this, but 54 % of food products in the food supply have Na levels above the salt targets proposed by FSANZ in 2010(Reference Webster, Dunford and Neal17, Reference Grimes, Nowson and Lawrence29).
It is of interest that those women who had high Na intakes (as assessed by urinary Na excretion) had a greater disparity between the recorded dietary intake and the urinary excretion, suggesting that they were either using more added salt or failing to report salty snacks and other salt-rich foods. This was confirmed by the association of this difference with BMI, supporting under-reporting of food as the major problem. Under-reporting of dietary intake has been reported previously and increases as BMI increases(Reference Voss, Kroke and Klipstein-Grobusch30). This phenomenon emphasises the need for biomarkers of dietary intake such as urinary Na to validate dietary Na intake and urinary N to validate protein intake, together with p-aminobenzoic acid(Reference Bingham and Cummings31) or other markers to verify completeness of 24 h urine collections(Reference Bingham32). Those who reported a higher Na intake (or higher energy intake) had little or no disparity between intake and output.
Robustness of dietary intake data is dependent on the food composition tables used. In the present study the most recent version of the Australian food composition tables available at the time of the study was used(33).
We conclude that bread remains a major source of dietary Na and that the majority of women in the present study had Na intake exceeding the NHMRC UL. Effective reduction of Na requires specific targeting of those foods in the food supply that contribute most to Na intake, including breads and cereals and processed meats.
Acknowledgements
Sources of funding: This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. Conflicts of interest: The authors have no conflicts of interest in relation to this work. Authors’ contribution: J.B.K. designed the study, contributed to the interpretation of the data and the manuscript; K.L. undertook the statistical analysis and reviewed the manuscript; R.H. drafted the manuscript and contributed to the data analysis; P.M.C. oversaw the study interpretation of the data and the manuscript. Acknowledgement: Data from the Australian Longitudinal Study on Women's Health on bread consumption were used in this paper and the University of Newcastle and the University of Queensland are gratefully acknowledged.
