Hospital patients often have increased energy needs due to acute or chronic illness(Reference Lochs, Dejong and Hammarqvist1,Reference Thibault, Abbasoglu and Ioannou2) . Nutritional fluids (i.e. oral, enteral, parenteral) may be used as a first-line intervention to feed individuals who are critically ill or nutritionally compromised(3–Reference Cawood, Elia and Stratton5). Healthcare professionals rely on the nutrition information panel (NIP) of products to ensure patients are provided with sufficient calories to meet their estimated requirements(Reference Lochs, Dejong and Hammarqvist1). Inaccurate calorie values could expose patients to risks associated with under- and over-feeding.
Nutritional fluids provided in hospitals vary in composition, with some classified as foods while others are considered medicines. In Australia, Food Standards Australia New Zealand (FSANZ) regulate fluids that are consumed orally (including clinically prescribed oral nutrition supplements, ONS) or delivered enterally via tube feeding (e.g. enteral nutrition; EN). On the other hand, the therapeutic goods administration is responsible for the governance of medically prescribed intravenous (IV) fluids and parenteral nutrition (PN). Mandatory thresholds for the accuracy of reported energy on food and beverage product labels in Australia do not exist(6). This resonates with other regions throughout the world, including the European Union(7), Canada(8) and UK(9). FSANZ stipulates that packaged products must contain an NIP, which includes information outlining the average energy content (kJ) for both a serving and unit quantity of the food(6,10) . Reported energy values can be estimated either by a summation of the energy contained within each macronutrient (FSANZ Schedule 11), or obtained from a comparable item listed in the Australian Food Composition Database(11). Therapeutic goods administration conveys that when a medicine is intended for use as an energy source it must display an energy (kJ) equivalent for the stated volume(12). Without a direct measure of the energy content, stated values have the potential to misrepresent the actual energy contained within a fluid.
Bomb calorimetry provides a method for directly ascertaining the maximum energy content of food (i.e. gross energy, GE), and via energy conversion factors, can be used to verify energy values (i.e. metabolisable energy, ME) reported on food labels(13,Reference Allison, Heshka and Sepulveda14) . Standardised food/fluid sample preparation methods and combustion procedures have been developed to ensure bomb calorimetry is performed reliably and accurately(Reference Hopper, Desbrow and Roberts15). Measured values can then be assessed against reported values and equivalence determined by comparing them to predetermined thresholds.
The aim of this study was to quantify the energy from different nutritional fluids provided in hospitals via bomb calorimetry and compare these to values determined from manufacturers’ specifications (i.e. NIP). Results will indicate the extent to which current food and medical standards support the accurate quantification of energy by manufacturers of fluids supplied to vulnerable populations.
Materials and methods
Study overview
This study employed a statistical approach to determining equivalence(Reference Lakens16) between measured (via bomb calorimetry) and reported energy in nutritional fluids available at a major tertiary hospital (750 beds) in Queensland, Australia. Predetermined equivalence thresholds of ±10 % were employed for all comparisons. Reported GE values determined from products’ NIP or from the hospital’s electronic food service system were calculated. Fluids then underwent direct caloric measurement via bomb calorimetry to establish a mean ± 90 % CI GE value. Reported values were considered ‘equivalent’ to measured values when the established ±90 % CI fell within the ±10 % thresholds(Reference Lakens16,Reference Walker and Nowacki17) .
Nutritional fluids provided in hospital
Nutritional fluids were defined as any liquid provided to patients orally, enterally or parenterally, to deliver nutrition. This included items available on the hospital’s ‘Free Fluids’ diet (i.e. any item that is a smooth liquid, with no lumps or pieces), including products classified by the International Dysphagia Diet Standardisation Initiative (IDDSI) framework as being a ‘drink’ (i.e. thin and thickened fluids)(18), as well as clinician (e.g. dietitian) prescribed ONS and EN. Medical practitioner prescribed fluids, where the intention was for use as an energy source (i.e. IV and PN), were also included. Fluids were categorised according to FSANZ standards(6,19,20) and therapeutic goods administration criteria(21) as described in Table 1. A complete list of manufacturer names and brands for all items analysed is provided in Supporting Information (File S1 – Manufacturer Names). Items were further identified according to major product constituents, including milk, nut/seed, juice and sugar (i.e. made from predominantly added carbohydrate/sugars) based fluids.
Table 1. Food and medical nutrition classifications of nutritional fluids using FSANZ and TGA criteria
FSANZ, Food Standards Australia New Zealand; ONS, oral nutrition supplements; EN, enteral nutrition; FMR, formulated meal replacement; SF, supplementary food; IV, intravenous; PN, parenteral nutrition; TGA, therapeutic goods administration.
Serving size (weight)
Nutritional fluids were available in either single-serve (i.e. tetra pack, EN/PN bag) or multi-serve (e.g. cordial bottle, decanted juice) packages. Items were weighed prior to decanting the fluid and the package weight was subtracted from the gross weight to ascertain net fluid weight (g). If packages contained multiple serves, fluid weight was divided by the number of serves declared on the NIP to provide serving size weight. To assess serving size consistency, five to six single-serve items from within each of the five oral/enteral fluid classifications were weighed (including packaging), and the CV was determined.
Establishing equivalence thresholds
The absence of mandated tolerance levels for energy accuracy on food and beverage product labels in Australia necessitated a priori consideration of an appropriate equivalence margin. Initially, the US Food and Drug Administration ±20 % tolerance level was considered(22). However, this threshold was deemed too lenient given the importance of accurate energy prescriptions/intake for hospital patients. As such, a more conservative ±10 % threshold was employed as a ‘consensus criteria’ based on the expert opinion of the Food Regulation Policy Advocacy Working Group of Dietitians Australia (personal communication).
Calorimetry sample preparation
Calorimetry sample preparation followed the principles outlined by Hopper et al. (2024)(Reference Hopper, Desbrow and Roberts15). Fluids were first agitated for ∼60 s, and then a sample (∼20 g) of each was measured on an analytical balance (Ohaus AX324, Ohaus Corporation) before being poured into a silicone mould. Samples were then dehydrated by oven-drying (LT28 500W 6-layer) at 70°C for 72 h, or until a constant weight was achieved (n.b., PN – protein component ∼120 h). Dehydrated samples were homogenised using a mortar and pestle. A pellet press (MTB12 Micro-tec, Haarlem, Netherlands) was then used to form ∼1 g samples from the homogenised powder. Liquids that initially came in the form of a dry powder (i.e. supplement powders requiring water for consumption) were immediately pelletised. Initially, three pellets were combusted for each fluid using an oxygen bomb calorimeter (DDS CAL3K-S, Digital Data Systems), following the manufacturers calibration (Benzoic acid) and operational instructions(23). Calibration was performed at the beginning of each day and after 15 combustions, or when the ambient temperature changed by greater than 2. If within-sample test differences for fluid pellets were > 0·5 kJ/g, or if the 90 % CI of the averaged data approached an equivalence margin, up to two additional pellets were combusted.
Gross energy calculation
Energy values reported on the NIP and electronic food service system database represent ME(11). This requires conversion to GE to permit comparison against values that are directly measured. GE values of fluids were established using the following calculation:
$$G{E_{kJ}} = \Sigma _{{W_i} \times {F_i}}^N$$
where
${\rm{G}}{{\rm{E}}_{kJ}}$
is the total GE per serve, N is the number of energy-contributing constituents per serve of fluid,
${W_i}$
is the weight of each macronutrient per serve of the fluid item and
${F_i}$
represents the GE factors for each energy-providing constituent (i.e. fat = 39 kJ/g, protein = 23 kJ/g and all carbohydrates including fibre = 17 kJ/g(13,Reference Capuano, Oliviero and Fogliano24,Reference Atwater and Benedict25) ). GE calculations assumed items contained the reported macronutrients (i.e. fat, protein, carbohydrate) per serve (g) as stated on the NIP (packaged items) or electronic food service system database (local recipes). Total EkJ
was divided by serving size actual weight (g) to provide kJ/g.
Gross energy measurement
GE values determined via bomb calorimetry were established using the following calculation:
$${{{{M_d}}}\over{{{M_w}}}} \times {\varepsilon _i} = \;{\varepsilon _f}\;$$
where ϵ f is the energy density (kJ/g) of the beverage (prior to dehydration), M d is the dry mass of the sample, M w is the wet (initial) mass of the sample and ϵ i is the energy density (kJ/g) of the dry sample (determined by calorimeter combustion).
Statistical analysis
Data were initially entered into a Microsoft Excel (Microsoft, Office 365™) spreadsheet. The mean % difference (Mean Δ) for measured v. label comparison (i.e. kJ/g) and 90 % CI were calculated in the Excel sheet (by imputing appropriate formula functions) for each test fluid to determine equivalence (i.e. the 90 % CI fits within the established equivalence margins), and the 95 % CI was used to test the null hypothesis (i.e. no statistical difference when the 95 % CI crosses zero, P > 0·05). When the 90 % CI of the Mean Δ extended beyond the established equivalence margins (i.e. ±10 %), this was interpreted as measured v. label comparisons being non-equivalent, in accordance with Lakens(Reference Lakens16). Equivalence plot figures were produced using RStudio™(26), with the ‘ggplot2’ package(Reference Wickham27).
Results
Serving size (weight) consistency
In total, 130 samples were assessed for serving size consistency (i.e. five serves each from five to six items within each oral/enteral fluid classification). Overall, the CV ranged from 3·4 % (local recipes) to 0·3 % (prescribed nutrition fluids – ONS/EN) as per Table 2. A comprehensive list of weight variation for all items analysed is provided in Supporting Information (File S2 – serving size consistency).
Table 2. Serving size consistency summary
EN, enteral nutrition; ONS, oral nutrition supplements.
Gross energy – reported v. measured
GE was measured for seventy-three fluids. Samples consisted of sixty-two commercially manufactured items displaying a NIP and eleven locally produced recipes for which nutrition information was derived from an electronic food service system database. Four items were excluded due to containing sugar alcohols (erythritol, n 2) and fatty acids (n 2). Accurate GE comparisons for these products could not be attained due to manufacturers not listing sugar alcohol amounts on their NIP and inability to successfully dehydrate products containing fatty acids. A summary of included fluids (n 69) and classifications outlining statistical equivalence (±10 %) is provided in Table 3. A detailed list of results for all individual products is provided in Supporting Information (File S3 – Results Table).
Table 3. Reported v. measured energy accuracy of nutritional fluid categories
ONS, oral nutrition supplements; EN, enteral nutrition; IV, intravenous; PN, parenteral nutrition.
When compared with the ±10 % equivalence margin, 80 % (n 55) of fluids demonstrated statistical equivalence. Of the fourteen products whose energy content was deemed non-equivalent to the NIP, most (n 11; 79 %) contained fewer kilojoules than that calculated from manufacturer’s reports.
Food and medical nutrition classifications
Local recipes were the least equivalent nutritional fluid classification (n 2; 18 %). Supplementary fluids (n 4; 80 %), pre-packaged general fluids (n 15; 84 %) and prescribed thickened fluids (n 7; 88 %) all had equivalence outcomes over four times greater. Prescribed ONS/EN (n 20; 100 %) and prescribed IV/PN (n 7; 100 %) all contained calories considered equivalent to that calculated from manufacturers reports. Equivalence plots relevant to each fluid classification are displayed in Figure 1.
Figure 1. Local recipes (a), pre-packaged general fluids (b), supplementary fluids (c), prescribed nutrition fluids – thickened (d), prescribed nutrition fluids – ONS and EN, (e) and prescribed medical nutrition – IV and PN and (f) directly measured (bomb calorimetry) v. food label energy comparison (Mean Δ (centre), 90 % CI (thick error bars) and 95 % CI (dashed thin error bars)). All measured values (mean ± 90 % CI) are normalised to the energy value reported from manufacturers’ specifications using the nutrition information panel (0). ONS, oral nutrition supplements; EN, enteral nutrition; IV, intravenous; PN, parenteral nutrition.
Product constituents – milk/juice/nut/seed/sugar-based
Milk-based samples most often met the equivalence criteria (n 38; 90 %), whereas all other items (i.e. juice/nut/seed/sugar-based) exhibited equivalence in only 50 % of cases.
Discussion
This study quantified the energy content of different nutritional fluids available at a major tertiary hospital, comparing values derived directly via bomb calorimetry against those determined from manufacturers’ reported values. Non-equivalence between reported and measured calorie values was identified for items across most fluid categories (except for prescribed ONS/EN and IV/PN) but was more common in non-prescribed fluids (e.g. locally prepared recipes). Results suggest that a direct measure of energy may be required for some hospital fluid categories to ensure accurate calorie provision for vulnerable patients.
The current analysis indicated that one-fifth of nutritional fluid products provided to patients at a large tertiary hospital exhibited non-equivalent energy values (i.e. 90 % CI crossed the ±10 % threshold). Previous research suggests that the reported energy content of foods and beverages can differ from directly measured values by up to 85 %(Reference Allison, Heshka and Sepulveda14,Reference Root, Toma and Frank28–Reference Basolo, Parrington and Ando31) . Greater magnitude of difference tends to occur when items are locally produced (i.e. restaurant, take away kitchen, or café) compared with those generated on a larger national and commercial scale(Reference Allison, Heshka and Sepulveda14,Reference Urban, Weber and Heyman29,Reference Urban, McCrory and Dallal30,Reference Urban, Dallal and Robinson32) . Local recipes can be prone to larger differences due to increased instances of human error (e.g. inaccurate recipe formulation)(Reference Urban, Weber and Heyman29,Reference Urban, McCrory and Dallal30,Reference Urban, Dallal and Robinson32) . However, evidence suggests that large-scale pre-packaged food producers still often underestimate reported energy despite improved processes and efforts for standardisation(Reference Allison, Heshka and Sepulveda14,Reference Jumpertz, Venti and Le33) . The current study indicates that approximately 80 % of all non-equivalent outcomes comprised energy levels under the reported value, whereby locally produced recipes did in fact exhibit the greatest energy variation. Conversely, pre-packaged prescribed nutrition fluids (ONS/EN) and medical nutrition fluids (IV/PN) displayed the most accurate energy values, with all items demonstrating statistical equivalence.
Fluids governed by a food/medical standard and requiring clinician prescription (i.e. FSANZ Standard 2·9·5(20) and therapeutic goods administration Order No. 91(21)) were typically more accurate (i.e. statistically equivalent) than those not meeting these criteria (i.e. FSANZ Standard 1.2.8(6) and 2.9.3(19)). Accuracy was greater for prescribed nutrition fluids (i.e. ONS/EN) where product consistency was regulated (e.g. milk-based). In Australia, milk composition is governed by FSANZ (Standard 2.5.1(34)), which permits manufacturers to add or withdraw milk components (such as fat) to standardise milk composition as a means of producing nutritionally consistent products. In contrast, juice and nut/seed beverages are regulated according to a different standard (i.e. FSANZ Standard 2.6.1(35)) with less emphasis on product consistency. As such, fruit and nut/seed-based items assessed in this study were least likely to demonstrate equivalence, typically containing less energy than reported. Fruit and nut energy levels can vary based on season(Reference Eriksson and Ehrlén36), ripening(Reference Gautier, Diakou-Verdin and Bénard37–Reference Adrian, Arancon and Mathews39), climate(Reference Rabadán, Álvarez-Ortí and Pardo40) and maturity(Reference Song, Rowland and Tillman41). With a lack of nutrient and energy regularity in unformulated raw produce, these fluids may be prone to higher levels of energy variation and inaccuracy.
Despite being categorised as a prescribed nutrition fluid, thickened fluids demonstrated a slightly lower likelihood of equivalence (88 %) compared with ONS/EN and IV/PN products (100 %). Thickening agents listed as ingredients (e.g. guar and tara gums) contain organic compounds such as acetic and pyruvic acid, which have unique FSANZ GE factors (i.e. 13 kJ/g(11)). It is unclear how much these compounds impact reported ME values, as amounts (g) for these are not specified on the NIP. Energy values were unable to be accounted for with these compounds when calculating GE. Further research determining the impact of thickening agents on reported energy provision is required.
Clinical implications and suitability of current standards
While some food and medical nutrition labelling standards may facilitate accurate energy provision of hospital nutrition fluids, others may not. A lack of defined thresholds for food and beverage energy reporting may ultimately lead to inaccurate determination of energy provision to patients in Australian hospitals. As the majority of non-equivalent fluids tested in this study contained less energy than reported on the product’s NIP, the inadequate dietary provision (and intakes) of hospitalised patients previously reported within our facility(Reference Roberts, Chaboyer and Leveritt42) may have been underestimated. Possibly more apparent among patients whose intakes consist largely of locally prepared recipes and pre-packaged fluids that are juice or nut/seed-based. This can potentially lead to clinically meaningful variances in energy intake, examples of which have been demonstrated in a previous study whereby a ∼12 % reduction in the total calorie provision/day was observed when the non-equivalent measured energy value replaced NIP-derived ME values(Reference Chapman, Irwin and Hopper43). Mandating energy accuracy thresholds may present an opportunity to ensure all food and beverage items (not just prescription-based fluids) provide accurate nutrition label information. Although generous, the US Food and Drug Administration has a threshold for allowable energy variation of ±20 %(22). The mere presence of this mandatory regulation may facilitate hospitals’ and clinicians’ ability to provide sufficient nutrition to meet patient energy needs. This could be a consideration for other countries, however, associated implications such as manufacturer costs(13) and willingness to undertake additional product testing remain relatively unexplored. Establishing thresholds and the impact of associated changes is likely to require further industry and stakeholder consideration. Ultimately, hospitals could achieve improved energy accuracy if menu items supplied under food service contracts were mandated to undergo independent testing for direct determination of calorie content.
Reported energy values – additional sources of error
FSANZ indicates reported energy values for food/fluids to be estimated either by a summation of the ME contained within each macronutrient, or obtained from a comparable item listed in the Australian Food Composition Database(11). However, macronutrient and total energy values can vary based on the methods used to quantify the composition of the food/fluid(44), or constraints associated with the scope of products included within publicly available nutrition databases(45). Previous research suggests that energy underestimation is prominent in nutrition databases(Reference Basolo, Parrington and Ando31,Reference Urban, Lichtenstein and Gary46) . Foods and beverages listed in databases often display energy and macronutrient values based on averages taken from multiple samples (i.e. different brands), whereby individual values for each sample may differ. Without a direct measure of energy content, reported values, either via calculation or food databases, may misrepresent the actual energy contained within a food/beverage item.
Energy provision may also be influenced by variability in serving size, whereby serving size variation (CV) in this study ranged from 0·3 % (prescribed nutrition fluids – ONS/EN) to 3·4 % (local recipes). Research indicates local recipes (i.e. restaurant meals and takeaway items) are likely to have weight discrepancies(Reference Urban, Weber and Heyman29,Reference Urban, McCrory and Dallal30,Reference Urban, Dallal and Robinson32,Reference Urban, Lichtenstein and Gary46,Reference Roberts, Das and Suen47) compared with pre-packaged items(Reference Conway, Rhodes and Rumpler48). Reduced variation within packaged fluids implies patients are likely to be provided with correct portions and nutritional values for these products. Improved weight consistency (and energy accuracy) may be achieved if hospital menus primarily consist of pre-packaged items rather than locally produced/decanted products. However, this may have financial and environmental implications that hospitals should consider.
Gross energy measurement – challenges
Due to the scope of this investigation, outcomes were only measured for one item relevant to each hospital nutrition fluid analysed. As such, factors such as seasonal variation, batch production variability and manufacturing location differences were not examined. Combustion of erythritol (sugar alcohol) containing products also exhibited considerable energy variation (i.e. reported v. measured = +19 %). While dietary erythritol has a GE content of 17·2 kJ/g(Reference Livesey49), the ME value has been estimated to be less than 1·7 kJ/g(Reference Noda, Nakayama and Oku50). The label value assigned by FSANZ is 1 kJ/g(11) (i.e. ∼95 % less than sugar and other carbohydrates). Accurate energy comparison for these products could not be attained because manufacturers did not list the amounts of these compounds on their NIP.
The inability to fully dehydrate fatty acids successfully has been documented when undertaking lyophilisation, relating to structural membrane properties causing increased integrity and resistance to freezing(Reference Muñoz-Rojas, Bernal and Duque51–Reference Velly, Bouix and Passot54). This may also occur with oven-drying, with outcomes from this study indicating fluids containing fatty acids (e.g. mixed PN) were unable to be dehydrated (and combusted) completely, even after 5 days of dehydration. Label comparison results for these samples were therefore considered inaccurate. As a result, protein and carbohydrate PN chambers were dehydrated and analysed separately for this study. Further research is required to assess the impact of successfully dehydrating fatty acids for bomb calorimetry.
Conclusion
Current food labelling requirements in many countries, including Australia, provide limited imperative (i.e. mandatory thresholds) for food manufacturers to ensure accurate energy values are represented on NIP. This study determined that the reported energy value of nutritional fluids provided in hospitals can at times be non-equivalent (and lower) than the actual measured value. This is especially relevant for non-prescribed fluids, such as locally prepared recipes and fluids consisting of predominantly raw produce (such as fruit, nuts and seed additives). Fewer inaccuracies were evident among nutrition fluids governed by a food or medical standard dictating clinical prescription and/or product formulation. The direct measurement of the energy density of locally prepared hospital nutritional fluids is recommended to ensure appropriate energy provision to patients.
Acknowledgements
This work was supported by a PhD scholarship from Griffith University.
This research received no external funding.
Z.H.: Conceptualisation, Methodology, Formal analysis, Investigation, Writing – Original draft preparation, Writing – Review and editing. C.I.: Conceptualisation, Methodology, Data Curation, Writing – Review and editing, Supervision. B.D.: Conceptualisation, Methodology, Resources, Writing – Review and editing, Supervision. S.R.: Conceptualisation, Methodology, Writing – Review and editing, Supervision.
The authors have no conflicts of interest.
The authors affirm that this manuscript is an honest, accurate and transparent account of the study being reported. No important aspects of the study have been omitted and any discrepancies from the study as planned have been described.
Supplementary material
For supplementary material/s referred to in this article, please visit https://doi.org/10.1017/S0007114525000601
Open access
