Total energy expenditure measured using doubly labeled water in adults with major chronic diseases: a systematic review

Abstract

Background

Energy requirement assessment is a cornerstone for nutrition practice. The extent to which total energy expenditure (TEE; indicator of energy requirements) has been measured in adults with chronic diseases has not been explored.

Objectives

This systematic review aimed to characterize evidence on TEE among individuals with chronic diseases and describe TEE across chronic diseases and in comparison to controls without a chronic disease.

Methods

A literature search using terms related to doubly labeled water and TEE was conducted in PubMed, MEDLINE, Web of Science, and Embase. Eligible articles included those that measured TEE using doubly labeled water in adults with a major chronic disease. Methodological quality was determined using the Academy of Nutrition and Dietetics quality criteria checklist. Sample size-weighted TEE was calculated in each chronic disease subgroup.

Results

Fifty studies were included, of which 15 had a control group. Median sample size was 20 participants, and approximately half of studies were published over 10 y ago. Thirty-five (70%) studies reported resting energy expenditure, and approximately half (k = 26) reported physical activity level. Methodological quality was neutral (k = 25) or positive (k = 23) for most studies. TEE among individual studies ranged from 934 to 3274 kcal/d. Mean weighted TEE was lowest among gastrointestinal (1786 kcal/d) and neurologic (2104 kcal/d) subgroups and highest among cancer (2903 kcal/d), endocrine (2661 kcal/d), and autoimmune (2625 kcal/d) subgroups. Excluding 1 article in cancer survivors resulted in a low TEE in the cancer subgroup (2112 kcal/d). Most studies with a control group reported no differences in TEE between controls and patients; however, only 1 study was powered for between-group comparisons.

Conclusions

Energy requirements vary across chronic diseases, although there is insufficient evidence to suggest that TEE in patients with chronic disease is different than that among controls. Further research is needed to inform energy requirement recommendations that consider chronic disease.

This review was registered at PROSPERO as CRD42022336500 (https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=336500).

Introduction

Chronic diseases are long-term health conditions that persist over extended periods of time, require ongoing medical care, and negatively impact activities of daily living [[1], [2], [3]]. Chronic diseases such as cancer, heart disease, diabetes, and respiratory conditions are associated with substantial psychosocial, physiological, and economic burden and are a major contributor to mortality [4]. Over half of adults in the United States have ≥1 major chronic disease [5], and up to 98% of older adults have multiple comorbidities [6]; these numbers are expected to increase in the coming years [7].

Given the prevalence and negative impact of chronic diseases, behavioral strategies to manage these conditions are essential for optimizing overall health [8,9]. Improved dietary intake and overall nutritional status are critical components of effective health behavior interventions [[10], [11], [12]]. A crucial step in designing nutrition interventions and recommendations includes characterizing energy requirements. These are often assessed in research settings by measuring total energy expenditure (TEE) with doubly labeled water (DLW; gold standard due to its precision and reliability) [13,14]. Certain attributes of chronic diseases such as altered physical activity, systemic inflammation, hormonal status, and nutrient absorption may impact TEE—and thus energy requirements [[15], [16], [17], [18], [19], [20], [21]]. Describing TEE in relation to chronic disease status would benefit the development of targeted nutritional approaches to prevent malnutrition, minimize adverse clinical outcomes, and support improved health among populations with the most urgent need for nutritional assessment and intervention. However, current energy requirement recommendations have been developed based on data from individuals without chronic diseases (aside from obesity) [14]. The extent to which TEE has been objectively measured across different populations or how this may differ according to chronic disease status has not been mapped. Therefore, the objective of this systematic review was to synthesize the available literature that has measured TEE using DLW among individuals with a chronic disease. A secondary aim was to describe TEE in relation to chronic disease status and in comparison with control populations without a chronic disease.

Methods

We conducted this systematic review following a predefined, registered protocol (International Prospective Register of Systematic Reviews: CRD42022336500) to answer the following research questions: 1) What are the characteristics of studies that have measured TEE using DLW in chronic diseases? 2) Does TEE differ across populations with different chronic diseases? and 3) Is TEE different between populations with or without chronic disease (as reported by the original studies)?. This review was planned, conducted, and reported according to the Conducting Systematic Reviews and Meta-Analyses of Observational Studies of Etiology [22] and Preferred Reporting Items for Systematic Reviews and Meta-Analysis recommendations [23]. No ethical approval was required.

Search strategy

Four databases were searched to identify relevant publications: PubMed, MEDLINE, Web of Science, and Embase. Given the large number of eligible studies initially anticipated, searching for gray literature or references of included articles was not part of the search strategy. The search was informed by previous literature [24,25] and developed in an iterative process in which the ability of each term to identify relevant articles was discussed between 2 authors (SAP, SAC). The final search consisted of subject headings and keywords related to “energy expenditure” and “doubly labeled water” (Supplemental Table 1). Searches were undertaken from each database inception (i.e., no lower date limit) to 16 October, 2022. No limits were applied for language during the search.

Study selection

Articles from each database search were imported to Covidence online software (Vertitas Health Innovation, Melbourne, Australia). Duplicate citations were excluded first by Covidence automatically, and then manually during screening. Two co-authors (SAP, SAC) independently screened titles/abstracts (20 October, 2022, to 1 March, 2023) and full texts (2 March, 2023, to 18 May, 2023) in duplicate to identify eligible articles. Disagreements were resolved through discussion and consensus between the 2 authors. When consensus could not be reached, disagreement was resolved by discussion with a third co-author (ATL-M or CMP). In cases where TEE data were not reported in the results or it was unclear whether articles included the same subjects from a similar article, the corresponding author was contacted for clarification. If no response was received, 3 authors (SAP, SAC, ATL-M) discussed which publications to include; studies with absolute TEE reported in a larger sample size and published most recently were prioritized.

Eligibility criteria

The PECO (Population, Exposure, Comparator, Outcomes) guidelines were followed to identify eligible studies [26]. Eligible articles included those conducted in adults (age 18 y or older) with a major chronic disease (population) and TEE measured using DLW (outcome) (Supplemental Table 2). Only studies that measured TEE with DLW were included owing to its accuracy and ability to measure TEE over an extended period (reflective of energy requirements). A comparator or control group was not a required eligibility criterion, as the primary aim of this review was to synthesize literature that has measured TEE among individuals with a chronic disease. Where applicable, TEE from controls were included in the synthesis of data to compare TEE between people with or without chronic disease (secondary aim).

Although obesity is a chronic disease with substantial social, health, and economic impact [27,28], articles that included only adults with obesity without any other comorbidities were not included in this review because a previous narrative review has summarized energy expenditure in obesity [29] and a large proportion of articles in our preliminary search included adults with obesity without any other comorbidities. Furthermore, current energy requirement recommendations from the dietary reference intakes (DRIs) do not exclude individuals with obesity [14]. Articles were excluded if the population consisted of adults with acute conditions without a chronic disease, such as patients in intensive care units. This was due to a previous review of TEE in this population [24] and the present review’s focus on chronic diseases. Conditions affecting mental health (e.g., depression) are considered chronic diseases by some governmental agencies [2,30]; however, studies in which participants exclusively had conditions affecting mental health were not included to focus the present review on chronic diseases that directly and overtly affect physical health, metabolic outcomes, or nutritional status. We also did not include individuals with dementia, as a recent systematic review assessed TEE in this population [31].

Given the nature of the research questions, included studies were not limited to randomized controlled trials. Specifically, peer-reviewed articles from observational, cohort, case–control, longitudinal, or intervention studies were included; review articles, case reports, editorials, conference abstracts, and book chapters were excluded. To be eligible, TEE had to be expressed in absolute terms or adjusted for anthropometric or body composition variables. Only articles published in English were included.

Study quality assessment

Methodological quality of the included studies was assessed independently in duplicate by 3 authors (SAP, SAC, ATL-M) using the Academy of Nutrition and Dietetics Quality Criteria Checklist for Primary Research (AND-QC) [32]. This quality assessment tool was selected because it was designed for nonrandomized trials and is specific to the field of nutrition and dietetics. The AND-QC uses a rating of positive, negative, or neutral based on an assessment of 10 categories: research question, subject selection, comparable groups, withdrawals, blinding, intervention/exposure, outcomes, analysis, conclusion support, and likelihood of bias. Only sub-questions relevant to cross-sectional studies were considered regardless of the study design (i.e., criteria for interventions or longitudinal studies were not included in the quality assessment), owing to the nature of our research question. The checklist was piloted in 4 studies by the 3 authors (SAP, SAC, ATL-M) to ensure a consistent assessment. Disagreements were discussed among the 3 authors until consensus was reached. Additional information on DLW procedures was extracted and primarily used to inform the exposure score; reported TEE methods (including calculation from isotope enrichments) and results informed the AND-QC outcome score.

Data extraction

Data were extracted in Covidence by 1 author (SAP, SAC, or ATL-M) and another author (SAP, SAC, or ATL-M) independently checked the data for accuracy. Disagreements were discussed among these 3 authors until consensus was reached. The following information was extracted: general study characteristics (i.e., first author’s last name, year of publication, country of data collection, study design, and study setting); study participants (i.e., sample size, age, body mass index (BMI) [in kg/m²], body composition, and distribution of sex); chronic disease details (i.e., general and specific chronic disease type, chronic disease stage/severity or relevant inclusion criteria); DLW techniques (i.e., dose amount, sampling protocols, internal quality control, equations used, and analysis techniques); and outcome (i.e., TEE). Data on resting energy expenditure (REE), activity energy expenditure (AEE), and physical activity level (PAL) were also extracted when possible, given their relationship to TEE and potential utility in nutritional assessment.

Where applicable, TEE was converted from megajoules and kilojoules to kilocalories. Data are presented as mean ± SD unless otherwise stated. Standard errors and data originally presented as 95% CIs were converted to SD. In cases where TEE data were presented as median and interquartile range or range, first and corresponding authors were contacted to obtain mean and SD or standard error; if no response was received, these data were converted to mean ± SD [33]. Two articles did not report TEE mean and SD for the whole group but provided raw data for each participant; in these instances, raw data were extracted to manually calculate the mean and SD of energy expenditure for the entire group.

Data synthesis

The original protocol proposed to assess publication bias, heterogeneity, and differences in TEE between chronic disease and control groups; however, the limited number of studies that included a control group precluded these analyses. Alternatively, TEE across all studies and within each chronic disease subgroup is presented as a weighted mean, with the sample size of each study as the weighting factor using R Studio (version 2023.12.1). No estimates of variability were calculated for weighted means, as there is no consensus on calculating this value. In cases where TEE data were only presented by separate chronic disease groups in a single study (e.g., with or without lipodystrophy), both groups were included in weighted mean calculations. Only studies in which unadjusted TEE data were presented as (or converted to) mean were included in data synthesis. TEE was also analyzed using both sample size and sex (i.e., the proportion of male participants in each study) as weighting factors in exploratory analyses. Additionally, we calculated sample size-weighted TEE separately for studies involving inpatients. Weighted mean data were assessed descriptively. There was a large discrepancy in sample size in 1 study compared with that in others [34]; as such, a post hoc sensitivity analysis was conducted in which TEE data from this study was removed from weighted mean calculations; k and n were used to describe the number of studies and individual participants, respectively.

Results

Description of studies

The initial search identified 11,825 potentially relevant references, of which 4820 were duplicates (Figure 1). Two authors (SAP, SAC) screened 7005 titles and abstracts and 590 full texts; most articles were excluded at this stage because of inclusion of an ineligible population (e.g., k = 420 were studies that included populations that were exclusively healthy without chronic disease) or were not primary research (e.g., k = 91 conference abstract). Four authors were contacted for TEE data; no response was received from 3 authors, so these articles were excluded. A total of 50 articles were included [[34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [59], [60], [61], [62], [63], [64], [65], [66], [67], [68], [69], [70], [71], [72], [73], [74], [75], [76], [77], [78], [79], [80], [81], [82], [83]].

FIGURE 1.

Flow diagram of study selection. DLW, doubly labeled water; EE, energy expenditure. Exclusively healthy population indicates that there were no individuals with a chronic disease in the study.

Study characteristics

Of the 50 studies, 35 did not include a control group (Supplemental Table 3). In total, there were 1699 participants with a chronic disease and 429 control participants. The number of participants with a chronic disease and DLW measurements in each article ranged from 4 to 317, with a median of 20 participants. The number of control participants in each article ranged from 6 to 259, with a median of 10 participants. The mean age of participants ranged from 26 to 73 y (not reported in 7 studies), and the mean BMI ranged from 18.3 to 39.9 kg/m² (not reported or unable to assess in 12 studies). Thirty-five studies reported REE and 33 reported any measure of physical activity (k = 26 reported PAL; k = 23 reported AEE). Thirty-one studies measured and reported body composition. There were 5 different body composition techniques used and a wide range of parameters used to depict lean tissues (e.g., fat-free mass, lean soft tissue, body cell mass). Over half of the included studies (k = 27; 54.0%) were published before 2014 (range: 1986–2022). The most common country of publication was the United States (k = 14; 28.0%), followed by the United Kingdom (k = 9; 18.0%), and Japan (k = 8; 16.0%).

Among the 35 articles without a control group, the most common chronic diseases were as follows: pulmonary conditions—k= 8 articles, all in chronic obstructive pulmonary disease (COPD) [[61], [62], [63], [64], [65], [66], [67], [68]]; neurologic conditions—k = 6 articles, including 3 in amyotrophic lateral sclerosis [[57], [58], [59]] and 1 each in cerebral palsy [55], Parkinson's disease [56], and Charcot neuroarthropathy [60]); and cancer—k = 5 articles, including 3 in patients with current cancer [39,40,43] and 2 in individuals discharged from cancer treatment/care (cancer survivors) [41,42]. Other chronic diseases included the following: autoimmune conditions—k = 4, including 2 in human immunodeficiency virus [35,38] and 2 in rheumatoid arthritis [36,37]; cardiovascular diseases—k = 4, including 1 article each of ischemic heart disease [44], peripheral arterial disease [45], previous stroke [47], and congestive heart failure [46]; gastrointestinal conditions—k = 3, all mixed causes [[50], [51], [52]]; diabetes—k = 2 in type 2 diabetes [48,49]; renal disease—k = 2 in chronic kidney disease [53,54]); and endocrine conditions—k=1 in polycystic ovary syndrome [69] (Table 1).

TABLE 1.

Characteristics of included studies without healthy control groups¹.

First author, y	Chronic disease category	Population characteristics				TEE (kcal/d)
		n	Sex (n), male/female	Age (y)	BMI (kg/m2)
Autoimmune disease
Macallan et al. [35], 1995	Human immunodeficiency virus	27	27/0	Median (range): 35 (28–62)	20.5 ± 2.7	2750 ± 670
Hagfors et al. [36], 20052	Rheumatoid arthritis	33 (n = 9 with DLW)	3/6	Mediterranean diet (n = 17): 59 ± 10 Control diet (n = 16): 60 ± 8	27.0 ± 4.53	2572 ± 619
Johansson and Westerterp, [37], 2008	Rheumatoid arthritis	9	3/6	60 ± 11	27.4 ± 4.5	2605 ± 645
Guimarães et al. [38], 20172	Human immunodeficiency virus	45 (n = 17 with DLW) with lipodystrophy: 12 without lipodystrophy: 5	17/0	49 ± 9 (with lipodystrophy: 47 ± 7; without lipodystrophy: 40 ± 13)	With lipodystrophy: 23.3 ± 2.7 Without lipodystrophy: 24.2 ± 4.4	With lipodystrophy: 2618 ± 415 Without lipodystrophy: 2691 ± 856
Cancer
Hayes et al. [39], 20032	Cancer—mixed tumor types	12 (n = 4 with DLW)	7/5	Mean (range): Exercise group: 40 (16–64) Control group: 55 (46–64)	27.2 ± 1.0	284 ± 169 (kcal/kg FFM0.5)
Moses et al. [40], 2004	Cancer—pancreatic	24	10/14	68 ± 10	20.0 ± 4.9	1732 ± 402
Copland et al. [41], 2008	Cancer survivors—gastric carcinoma	15	10/5	68 ± 7	22.9 ± 3.3	2430 ± 540
Carter et al. [42], 20192	Cancer survivors—breast	37 (n = 32 with DLW)	0/32	55 ± 9	31.8 ± 7.8	2012 ± 302
Purcell et al. [43], 2019	Cancer—colorectal	21	14/7	57 ± 12	28.3 ± 4.9	2473 ± 499
Cardiovascular
Taggart et al. [44], 1991	Ischemic heart disease	13	13/0	53 ± 7	25.5 ± 2.6	2183 ± 507
Gardner et al. [45], 1999	Peripheral arterial disease	61	57/4	70 ± 6	29.5 ± 4.2	2385 ± 447
Skotzko et al. [46], 2000	Congestive heart failure	33 (n = 14 with depression; n = 19 without depression)	With depression: 12/2 Without depression: 18/1	With depression: 64 ± 11 Without depression: 67 ± 7		With depression: 1654 ± 309 Without depression: 1674 ± 354
Moore et al. [47], 2012	Stroke survivors	9	6/3	73 ± 8	27 ± 2	2473 ± 468
Diabetes
Chong et al. [48], 19934	Type 2 diabetes	23	8/15	51 ± 11	29.9 ± 7.2	2878 ± 635
Rollo et al. [49], 2015	Type 2 diabetes	19	6/4	61 ± 7	31.0 ± 4.5	2820 ± 550
Endocrine
Broskey et al. [69], 2017	Polycystic ovary syndrome	28	0/28	29 ± 5	39.9 ± 8.3	2661 ± 373
Gastrointestinal
Schoeller et al. [50], 1986	Crohn's disease or ulcerative colitis	5	0/5	26 ± 6	19.6 ± 2.6	2020 ± 420
Novick et al. [51], 1988	Crohn's disease or stomach cancer	7 (n = 6 Crohn's disease; n = 1 stomach cancer)	0/7	34 ± 16	19.7 ± 5.0	1605 ± 397
Pullicino et al. [52], 1993	Crohn's disease, perforated intestinal desmoid, perforated colonic carcinoma, or perforated colonic diverticulum	13 (n = 10 Crohn's disease, n = 1 of each other condition)	Not reported	37 ± 12	18.3 ± 2.3	1717 ± 378
Neurologic
Johnson et al. [55], 1997	Cerebral palsy	30 (n = 18 male; n = 12 female)	18/12	Males: 35 ± 12 Females: 40 ± 12	Males: 23.6 ± 5.0 Females: 27.2 ± 7.8	Males: 2455 ± 622 Females: 1986 ± 363
Delikanaki-Skaribas et al. [56], 2009	Parkinson's disease	20	20/0	72 ± 10	Not reported	2237 ± 510
Ichihara et al. [57], 2012	Amyotrophic lateral sclerosis	10	7/3	66 ± 11	Not reported	934 ± 201
Kasarski et al. [58], 2014	Amyotrophic lateral sclerosis	80	52/28	59 ± 12	27.1 ± 4.7	2364 ± 647
Shimizu et al. [59], 2017	Amyotrophic lateral sclerosis	26	13/13	Median (IQR): 65 (62–70)	Median (IQR): 19.8 (17.6–22.3)	1615 ± 403
Grant et al. [60], 20212	Charcot neuroarthropathy	43 (n = 41 with DLW)	27/16	60 ± 10	Not reported	1887 ± 783
Pulmonary
Baarends et al. [61], 1997	COPD	20 (n = 10 with 'normal REE'; n = 10 without normal REE)	19/1	Median (range): With normal REE: 69 (55–78) Without normal REE: 65 (57–74)	Median (range): With normal REE: 24.4 (17.4–28.9) Without normal REE: 23.9 (16.6–32.4)	With normal REE: 2617 ± 373 Without normal REE: 2599 ± 311
Slinde et al. [62], 20034	COPD	10	5/5	63 ± 8	18.7 ± 1.2	1984 ± 466
Slinde et al. [64], 2006	COPD	14	5/9	64 ± 8	19.0 ± 1.9	1840 ± 287
Arvidsson et al. [63], 2006	COPD	13	4/9	63 ± 8	19.3 ± 1.9	1810 ± 302
Farooqi et al. [65], 2015	COPD	19	0/19	69 ± 6	24.5 ± 3.5	1904 ± 261
Nishida et al. [67], 2021	COPD	33	33/0	71 ± 7.3	21.7 ± 3.4	2245 ± 419
Sanders et al. [66], 2021	COPD	20	7/13	63 ± 7	24.1 ± 4.4	2133 ± 294
Shirahata et al. [68], 2022	COPD	36	36/0	70 ± 6	21.9 ± 3.2	2273 ± 445
Renal disease
Sridharan et al. [53], 2016	Chronic kidney disease	40	22/18	54 ± 17	26.8 ± 4.2	2481 ± 476
Vilar et al. [54], 2021	Chronic kidney disease	80 (n = 57 eGFR <50; n = 23 eGFR ≥ 50)	52/28	Median (IQR); intervention: 62 (48–75)	Median (IQR): 26.8 (22.5–31.1)	eGFR <50: 2388 ± 494 eGFR ≥50: 2704 ± 457

Abbreviations: BMI, body mass index; COPD, chronic obstructive pulmonary disease; DLW, doubly labeled water; eGFR, estimated glomerular filtration rate; FFM, fat-free mass; IQR, interquartile range; REE, resting energy expenditure; TEE, total energy expenditure. Age was rounded to the nearest whole number and BMI was rounded to the nearest 0.1 decimal.

¹Data are mean ± standard deviation unless otherwise stated.

²Participant descriptive data (e.g., age, BMI) represents whole study sample; TEE data represent subgroups with DLW.

³Data were manually extracted from data presented in the article.

⁴Data from communications with study authors.

Among the 15 articles that included control participants, the most common chronic diseases were as follows: autoimmune conditions—k = 3, including 2 in human immunodeficiency virus [70,71] and one in rheumatoid arthritis [72]; diabetes—k = 3 in type 2 diabetes [[74], [75], [76]]; neurologic conditions—k = 3, including 2 in Huntington's disease [78,79] and 1 in multiple sclerosis [80]; and pulmonary conditions—k = 3, all COPD [[81], [82], [83]]. Other conditions included the following: cancer—k = 1 in cancer survivors [34]; cardiovascular disease—k = 1 in chronic heart failure [73]; and gastrointestinal conditions—k = 1 in short bowel syndrome [77] (Table 2).

TABLE 2.

Characteristics of included studies with healthy control groups¹.

First author, y	Chronic disease category	Population characteristics				TEE (kcal/d)	TEE different in CD?
		n	Sex (n), male/female	Age (y)	BMI (kg/m2)
Autoimmune disease
Heijligenberg et al. [70], 1997	Human immunodeficiency virus	CD: 9 Controls: 9	CD: 9/0 Controls: 9/0	CD: 34 ± 6 Controls: 33 ± 5	CD: 23.1 ± 2.4 Controls: 23.6 ± 1.5	CD: 3274 ± 502 Controls: 3035 ± 359	No
Coors et al. [71], 2001	Human immunodeficiency virus	CD: 7 Controls: 7	CD: 7/0 Controls: 7/0	CD: 37 ± 11 Controls: 29 ± 3	CD: 21.8 ± 1.8 Controls: 22.8 ± 3.8	CD: 2629 ± 193 Controls: 2869 ± 338	No
Roubenoff et al. [72], 2002	Rheumatoid arthritis	CD: 20 Controls: 20	CD: 0/20 Controls: 0/20	CD: 47 ± 14 Controls: 48 ± 14	CD: 25.3 ± 4.5 Controls: 24.2 ± 3.3	CD: 2183 ± 319 Controls: 2504 ± 476	Yes; lower in the CD group
Cancer
Ness et al. [34], 20152	Cancer survivors—childhood acute lymphoblastic leukemia	CD: 365 (317 with DLW) Controls: 365 (259 with DLW)	CD: 162/155 Controls: 128/131	CD: 29 ± 6 Controls: 29 ± 8	CD: 28.1 ± 6.9 Controls-males: 27.2 ± 5.6 Controls-females: 27.8 ± 8.0	CD: Unadjusted: All CD: 3029 ± 7483 Adjusted for weight: CRT-males: 3290 ± 576 CRT-females: 2432 ± 419 No CRT-males: 3506 ± 632 No CRT-females: 2840 ± 438 Controls: Adjusted for weight: Males: 3423 ± 691 Females: 2598 ± 475	Yes; weight-adjusted TEE lowest in survivors exposed to CRT
Cardiovascular
Steele et al. [73], 1998	Chronic heart failure	CD: 12 Controls: 8	CD: 12/0 Controls: 8/0	Median (range): CD: 69 (58–81) Controls: 68 (66–77)	Median (range): CD: 23.1 (20.8–27.5) Controls: 24.4 (22.5–31.0)	CD: 3123 ± 1223 Controls: 2325 ± 619	No
Diabetes
Morino et al. [74], 2019	Type 2 diabetes	CD: 52 Controls: 15	CD: 28/24 Controls: 6/15	CD: 70 ± 5 Controls: 67 ± 5	CD: 23.3 ± 3.0 Controls: 22.7 ± 2.1	CD: 2159 ± 388 Controls: 2168 ± 727	No
Yoshimura et al. [75], 2019	Type 2 diabetes	CD: 12 Controls: 10	CD: 12/0 Controls: 10/0	CD: 55 ± 7 Controls: 55 ± 7	CD: 24.0 ± 1.8 Controls: 23.6 ± 1.8	CD: 2490 ± 379 Controls: 2284 ± 243	No
Ishikawa-Takata et al. [76], 2020	Type 2 diabetes	CD: 9 Controls: 10	CD: 5/4 Controls: 5/5	Median (IQR): CD: 52 (51–56) Controls: 54 (51–58)	Median (IQR): CD: 30.3 (28.5–33.5) Controls: 29.4 (28.7–30.7)	CD: 2831 ± 667 Controls: 2745 ± 435	No
Gastrointestinal
Fassini et al. [77], 2016	Short bowel syndrome	CD: 11 Controls: 11	CD: 5/6 Controls: 5/6	(both groups combined): 53 ± 8	CD: 21.5 ± 3.4 Control: 22.3 ± 2.5	CD: 1875 ± 276 Controls: 2393 ± 445	Yes; lower in the CD group
Neurologic
Pratley et al. [78], 2000	Huntington's disease	CD: 17 Controls: 17	CD: 11/6 Controls: 11/6	CD: 49 ± 4 Controls: 46 ± 4	CD: 25 ± 1 Controls: 26 ± 1	CD: 2365 ± 416 Controls: 2297 ± 388	No
Lazar et al. [79], 2015	Huntington's disease	CD: 38 Controls: 36	CD: 13/25 Controls: 18/18	CD: 43 ± 11 Controls: 44 ± 15	CD: 26.0 ± 5.0 Controls: 24.6 ± 3.0	Adjusted for age, sex and body weight: CD: 2689 ± 376 Controls: 2834 ± 387	No differences in age-adjusted, sex-adjusted, and weight-adjusted TEE
Silveira et al. [80], 20212	Multiple sclerosis	CD: 30 (n = 28 with DLW) Controls: 15	CD: 5/25 Controls: 2/13	CD: 41 ± 10 Controls: 35 ± 9	Not reported	CD (n = 28): 2124 ± 376 Controls (n = 15): 2286 ± 355	Not reported
Pulmonary
Baarends et al. [81], 1997	COPD	CD: 8 Controls: 8	CD: 8/0 Controls: 8/0	CD: 66 ± 6 Controls: 73 ± 6	CD: 21.8 ± 2.5 Controls: 22.4 ± 2.5	CD: 2499 ± 320 Controls: 2107 ± 88	Yes; higher in the CD group
Sato et al. [82], 2021	COPD	CD: 28 Controls: 8	CD: 28/0 Controls: 8/0	CD: 70 ± 6 Controls: 70 ± 7	CD: 22.1 ± 3.2 Controls: 21.2 ± 3.2	CD: 2283 ± 393 Controls: 2240 ± 629	No
Sanders et al. [83], 2022	COPD	CD: 20 Controls: 6	CD: 10/10 Controls: 4/2	CD: 62 ± 6 Controls: 60 ± 6	Median (range): CD: 22.4 (15.1–32.5) Controls: 23.13 (21.2–30.8)	CD: 2209 ± 394 Controls: 2631 ± 265	Yes; lower in the CD group

Abbreviations: BMI, body mass index; CD, chronic disease; COPD, chronic obstructive pulmonary disorder; CRT, cranial radiation therapy; DLW, doubly labeled water; TEE, total energy expenditure. Age was rounded to the nearest whole number and BMI was rounded to the nearest 0.1 decimal.

¹Data are mean ± standard deviation unless otherwise stated.

²Participant descriptive data (e.g., age, BMI) represents whole study sample; TEE data represent subgroups with DLW.

³Data from communications with study authors.

Sample characteristics and TEE data

Across studies with sufficient TEE and sample characteristics reported, mean age ranged from 26 [50] to 73 y [47] (k = 46 studies) and BMI ranged from 18.3 [52] to 39.9 kg/m² [69] (k = 39 studies). The mean proportion of males was 58.5% ± 34.3% (range: 0%–100%).

Across all studies, mean TEE in the chronic disease subgroups ranged from 934 [57] to 3274 kcal/d [70] (TABLE 1, TABLE 2). Two studies were not included in weighted mean calculations as data were presented as in kilocalories per kilogram of fat-free mass^0.5 (the coefficient indicates the power to which TEE was adjusted for fat-free mass using log-log regression) [39] or did not provide unadjusted TEE [79]. Mean weighted TEE across the remaining 48 studies was 2529 kcal/d (Figure 2). Among subgroups of conditions, mean weighted TEE was the lowest among studies that included individuals with gastrointestinal (1786 kcal/d), neurologic (2104 kcal/d), or pulmonary (2191 kcal/d) conditions and greatest among the cancer (2903 kcal/d), endocrine (2661 kcal/d), and autoimmune (2625 kcal/d) subgroups. Given the large discrepancies in sample size, 1 article with 365 participants [34] was removed for a sensitivity analysis. Without this large sample of acute lymphoblastic leukemia survivors [34], mean weighted TEE across all studies was 2262 kcal/d and TEE within the cancer subgroup was among the lowest (2112 kcal/d). Weighted mean TEE was 1987 kcal/d in the 7 studies that included inpatients [[50], [51], [52],57,61,79,81] and 2550 kcal/d in the remaining 43 studies that included outpatients. Weighted mean TEE in each subgroup was slightly altered when sex was included as an additional weighting factor. However, the general ranking of TEE from lowest to highest across subgroups remained similar to the ranking obtained when only the sample size was used as the weighting factor (Supplemental Figure 1).

FIGURE 2.

Mean weighted total energy expenditure (TEE) among studies grouped by chronic disease subgroup. TEE estimates were weighted by sample size. k = 48 studies included; k = 2 studies were not included in weighted mean calculations because unadjusted TEE data were not available. Of the 48 included studies, k = 4 were in gastrointestinal (total sample size across all pooled participants: n = 36 participants), k = 9 in neurologic (n = 252 participants), k = 12 in pulmonary (n = 221 participants), k = 6 in cardiovascular (n = 128 participants), k = 2 in renal (n = 120 participants), k = 5 in diabetes (n = 106 participants), k = 8 in autoimmune (n = 98 participants), k = 1 in endocrine (n = 28 participants), and k = 5 in cancer (n = 668 participants).

Among studies that included control participants, 9 reported matching groups based on age, height, weight, and/or BMI (Supplemental Table 3) [34,70,75,[77], [78], [79], [80], [81],83]. Fourteen studies presented results of statistical analyses comparing TEE between groups [34,[70], [71], [72], [73], [74], [75], [76], [77], [78], [79],[81], [82], [83]], and 1 [74] presented a sample size calculation to compare TEE between groups. Of these, 9 did not observe between-group differences [70,71,[73], [74], [75], [76],78,79,82], 4 reported lower TEE in the chronic disease group—rheumatoid arthritis [72], childhood cancer survivors [34], short bowel syndrome [77], and COPD [83]—and 1 reported higher TEE in the chronic disease group (COPD) [81]. The 5 studies that reported between-group differences in TEE also reported differences in PAL (i.e., lower PAL and lower TEE or higher PAL and higher TEE in chronic disease group); 1 reported a difference in REE (lower REE and lower TEE in chronic disease group), and none reported differences in body composition.

Study quality

Twenty-five studies were categorized as neutral, 23 as positive, and 2 as negative quality, (Supplemental Table 4). The most common reasons that contributed to lower quality were having a sample that was not free from bias (e.g., only including males with chronic diseases that impact both sexes equally) and lack of a description of handling withdrawals. In addition, lack of details regarding DLW protocols (i.e., exposure) contributed to lower quality ratings. As outlined in Supplemental Table 5, several approaches were used for isotope dosing and calculation of carbon dioxide production and TEE. None validated total body water using external body composition methods (not reported in Supplemental Tables).

Discussion

Characterizing TEE is fundamental for the development of energy requirement recommendations. This is especially important among individuals with chronic diseases who may benefit from nutrition assessment and intervention. Our systematic review synthesized data from 50 articles that assessed DLW-measured TEE in adults with chronic diseases primarily affecting metabolic health. TEE was most commonly measured in adults with pulmonary or neurologic conditions (k = 11 articles each). Sample size-weighted mean TEE among subgroups ranged from 1786 to 2903 kcal/d. The majority of studies that included control participants did not report group differences in TEE.

The lack of robust data on energy requirements for individuals living with chronic conditions presents a significant challenge in the nutrition care process, particularly given that chronic diseases are a leading cause for dietetic referrals in both outpatient and inpatient settings [84,85]. Current energy requirements set forth by the DRI committee are intended for the general population; data from individuals with chronic diseases (except for obesity) were excluded from the newest recommendations, with the rationale that these conditions may affect TEE [14]. The general population DRIs were developed using DLW data from 8600 adults residing primarily in the United States. By comparison, this systematic review uncovered data from ∼1500 individuals across all chronic diseases—a much smaller number than data available in generally healthy adults. This relative paucity of data are problematic considering the high prevalence of chronic diseases globally, especially in aging populations [[5], [6], [7]]. Furthermore, approximately half of the included studies were published >10 y ago and may not reflect changes in treatment approaches, medications, or body composition patterns. Although median sample size was small (n = 20), this is characteristic of many studies that measure TEE using DLW, given the substantial cost and expertise for this technique. This limited amount of data likely stems from the unique challenges associated with characterizing energy requirements among people with chronic disease. For example, diverse treatments and medications may hinder the selection of a population that is both specific and generalizable. Recruitment is particularly challenging, as individuals with chronic diseases may have additional obstacles to participation such as transport difficulties, lack of time, and challenges in managing other responsibilities (especially in the context of frequent medical appointments) [86]. Although collaboration with clinical staff may enhance recruitment efforts, their primary focus is on patient care, with research being a secondary priority. Furthermore, ethics boards may require additional processes or approvals for conducting research in clinical settings if the population is deemed “medically vulnerable” [87], and hospital ethics boards are typically structured to accommodate traditional medical clinical trials rather than nutrition research. The substantial disparity in available TEE data between chronic disease populations and healthy adults suggests a lack of funding prioritization for characterizing energy requirements or validating new equipment to measure energy expenditure in individuals with chronic diseases. More data are needed to robustly address nuances in energy requirements that relate to race and ethnicity, medications, and body composition abnormalities such as sarcopenic obesity [14,18,43,88].

Of the available data in chronic disease, there were a greater number of studies that measured TEE among individuals with pulmonary (i.e., COPD) or neurologic conditions than those measured TEE in individuals with other chronic diseases. This observation is expected to some extent, considering the documented effects of these conditions on nutritional status [89,90]. Conversely, there have been relatively few studies among individuals with endocrine (one study in polycystic ovary syndrome) or renal diseases (2 studies in chronic kidney disease). This lack of data is somewhat surprising, given individuals with polycystic ovary syndrome and chronic kidney disease have a greater risk of obesity and/or malnutrition, may have endocrine or behavioral alterations that impact TEE, and benefit from nutrition counseling [[91], [92], [93], [94]]. Furthermore, there were no studies in populations with other major chronic diseases such as liver disease, thyroid disease, or endometriosis, despite endocrinologic, behavioral, and nutritional factors that could theoretically impact TEE and energy requirements [[95], [96], [97], [98], [99]]. Thus, additional studies are needed to characterize TEE in these populations.

Across all subgroups, mean weighted TEE varied substantially, likely due to differences in age, body size, disease-specific characteristics, and physical activity. TEE was lowest in studies that included individuals with gastrointestinal, pulmonary, and neurologic conditions; mean weighted TEE was also over 500 kcal/d lower in studies of inpatients than that in studies of outpatients. Similar to trends in people without chronic diseases, it is likely that lower body weight (and potentially fat-free mass) and greater age contributed to reduced TEE in these subgroups. Articles in the gastrointestinal subgroup primarily included individuals with Crohn's disease, ulcerative colitis, or short bowel syndrome, which are conditions associated with a higher risk of malnutrition and low BMI [100]. Indeed, mean BMI was below 20 kg/m² in 3 studies with gastrointestinal patients included in this review [[50], [51], [52]]. Similarly, individuals with COPD or neurologic conditions may have challenges in consuming adequate amounts of energy due to anorexia, fatigue, or dyspnea while eating; some studies have also reported elevated REE in patients with severe COPD (perhaps due to a greater energetic demand from breathing) [89,101]. Conversely, the largest study in this review included survivors of acute lymphoblastic leukemia, with one of the youngest mean ages (29 y) and high mean BMI (28.1 kg/m²) compared with other included studies [34]. When this study was removed in a sensitivity analysis, mean weighted TEE in the cancer subgroup was among the lowest, highlighting the substantial variability in TEE observed within some subgroups. This observation also suggests that cancer survivors—including those in remission from acute lymphoblastic leukemia—may exhibit different energy metabolism characteristics compared to patients undergoing active cancer treatment. This within-group variability reflects the important nuance that should be considered within some chronic disease categories that is beyond the scope of this review and the availability of current evidence. Interestingly, inclusion of sex as a weighting factor did not substantially affect weighted mean TEE among subgroups, suggesting that biological sex may play a lesser role on group-level TEE independent of other factors such as body size or physical activity.

Characterizing individual components of TEE in individuals with chronic diseases is important for identifying those at increased risk of altered energy metabolism and accurately estimating their energy requirements. Seventy percentage of the studies included in this review reported REE and 66% reported PAL or AEE, which hinders our ability to comprehensively evaluate relationships between physical activity parameters and TEE. Similarly, a recent systematic review of energy expenditure parameters in children with chronic diseases reported that 27 of 49 studies (55%) with TEE included REE and only 12 (24%) reported PAL or AEE [102]. Many chronic diseases are associated with reduced lean tissue [103], which may impact REE. Because of the wide array of body composition parameters and techniques (which often provide discrepant estimates [104]), reported across studies, we are unable to delineate the role of body composition on REE or TEE in this review. Interestingly, the 5 studies that reported differences in TEE between individuals with chronic disease and controls attributed these differences to PAL or AEE rather than REE or body composition. For example, patients with COPD enrolled in a rehabilitation program exhibited similar body composition but had greater AEE and TEE than controls; higher AEE and TEE may have been a result of the higher cost of breathing during exercise as part of their rehabilitation program [81]. Conversely, sedentary patients with COPD had lower PAL and TEE than age-, sex-, and BMI-matched controls with no differences in body composition [83]. Our observation of lower TEE in inpatient studies further substantiates the notion that PAL/AEE is a primary determinant of TEE, similar to people without chronic diseases [15]. Many chronic diseases and associated treatments may impact one’s ability to engage in physical activity, thereby impacting TEE. Given the impact of physical activity on TEE and individual-level and population-level variability in PAL [105], continued exploration of individual components of TEE in relation to energy requirements among individuals with chronic diseases is warranted. Further exploration of the role of body composition on TEE and identification of accurate and feasible methods for estimating PAL in real-world settings would greatly enhance clinical translation.

Thirty percent of studies in this review included a control group, although not all were matched on criteria important for energy balance (e.g., age, BMI), and only 1 study was powered to detect differences in TEE between groups. Most studies did not report between-group differences in TEE, although some reported lower TEE among individuals with rheumatoid arthritis, survivors of childhood acute lymphoblastic leukemia, and short bowel syndrome (with conflicting results in COPD) [34,72,77,81,83]. Although the use of a control group has several strengths, it is inherently limited by the multitude of factors that contribute to TEE beyond age, sex, and body mass [14,106]. Given the small number of chronic disease-control comparisons, lack of reported statistical power, and complexity in selecting control groups, there is currently insufficient evidence to discern whether any chronic disease consistently or systematically impacts TEE or energy requirements compared with controls.

Although this review is the first to systematically synthesize TEE across populations with chronic diseases, some limitations should be considered when interpreting the results. There was considerable variability in terms of chronic disease types, severity, and disease trajectories among many subgroups. For example, studies with cancer populations included patients and survivors of different types of cancer, and studies with neurologic populations included 6 different conditions. Such nuances within each chronic disease category warrants consideration, which is beyond the scope of this systematic review. Additionally, the generalizability of our findings may be limited by the heterogeneity of the included studies and lack of consistent reporting on key predictors of TEE (e.g., body composition, age, sex); thus, conclusions about the determinants of TEE should be interpreted with caution. Given the relatively few studies that included a control group, a meta-analysis between chronic disease and control groups was not conducted (to avoid misrepresentation of data). In addition, only 2 studies included in this review evaluated TEE in relation to DRIs, preventing direct comparisons of energy requirements in chronic disease with those recommended for the general population. There was also a number of equations used to calculate carbon dioxide production across studies, which can substantially impact TEE estimates from DLW [107]; this is an inherent limitation of the field when comparing TEE across populations that cannot be controlled for without isotope ratio data from individuals. In summary, this review underscores the scarcity of TEE data across many chronic diseases. Although TEE may vary according to general chronic disease status, there are insufficient data to elucidate the determinants of TEE or to ascertain whether TEE differs from that of controls without a chronic disease on a population level. Given the prevalence and impact of chronic diseases, there is an urgent need to characterize TEE further, especially in populations that have been underrepresented in research to date. Further research should prioritize measurement of body size, physical activity, and disease-specific factors that may plausibly impact energy requirements in adults with chronic diseases. Such endeavors are imperative to facilitate the development of more robust and tailored energy requirement recommendations.

Author contributions

The authors’ responsibilities were as follows – SAP, ATL-M, SAE, CMP: designed the research; SAP, SAC, ATL-M: conducted research; SAP, SAC: analyzed data; SAP: wrote the manuscript; SAP: had primary responsibility for final content; and all authors: provided manuscript editing and read and approved the final manuscript.

Conflict of interest

The authors report no conflicts of interest.

Funding

SAP and CMP are partially supported by the Canada Research Chair’s Program through the Government of Canada. SAC is supported by a Canada Graduate Scholarhip - Master's though the Canadian Institutes of Health Research (CIHR). ATL-M is supported by the CIHR Postdoctoral Research Fellowship, TRIANGLE Canada Research Fellow through the ENRICH Stream Program, and the American Society for Parenteral and Enteral Nutrition Rhoads Research Foundation. PT is supported by CIHR grant 180548 and Alberta Innovates PRIHSG201900037. ELM is supported by the National Institutes of Health Grants P30 DK048520 and U01 NS113851. CMP is supported by CIHR FRN 159537.

Data availability

The authors confirm that the data supporting the findings of this study are available within the article [and/or] its supplementary materials. Data described in the manuscript, code book, and analytic code will be made available upon request pending application and approval.

Appendix A. Supplementary data

The following are the Supplementary data to this article: