Poor Reporting Quality in Basic Nutrition Research: A Case Study Based on a Scoping Review of Recent Folate Research in Mouse Models (2009–2021)

ABSTRACT

Transparent reporting of nutrition research promotes rigor, reproducibility, and relevance to human nutrition. We performed a scoping review of recent articles reporting dietary folate interventions in mice as a case study to determine the reporting frequency of generic study design items (i.e., sex, strain, and age) and nutrition-specific items (i.e., base diet composition, intervention doses, duration, and exposure verification) in basic nutrition research. We identified 798 original research articles in the EMBASE, Medline, Food Science and Technology Abstracts (FSTA), Global Health, and International Pharmaceutical Abstracts (IPA) databases published between January 2009 and July 2021 in which a dietary folic acid (FA) intervention was used in mice. We identified 312 original peer-reviewed articles including 191 studies in nonpregnant and 126 in pregnant mice. Most studies reported sex (99%), strain (99%), and age (83%). The majority of studies used C57BL/6 (53%) or BALB/c (11%) mice aged 3–9 wk. Nonpregnancy studies were more likely to use only male mice (57%). Dietary FA interventions varied considerably and overlapped: deficiency (0–3 mg/kg), control (0–16 mg/kg), and supplemented (0–50 mg/kg). Only 63% of studies used an open-formula base diet with a declared FA content and 60% of studies verified FA exposure using folate status biomarkers. The duration of intervention ranged from 1 to 104 wk for nonpregnancy studies. The duration of intervention for pregnancy studies was 1–19 wk, occurring variably before pregnancy and/or during pregnancy and/or lactation. Overall, 17% of studies did not report ≥1 generic study design item(s) and 40% did not report ≥1 nutrition-specific study design item(s). The variability and frequent lack of reporting of important generic and nutrition-specific study design details in nutrition studies limit their generalizability, reproducibility, and interpretation. The use of reporting checklists for animal research would enhance reporting quality of key study design and conduct factors in animal-based nutrition research.

Statement of Significance: Basic nutrition research suffers from a frequent lack of reporting of both important generic and nutrition-specific study design details, limiting their generalizability, reproducibility, and interpretation. The use of reporting checklists for animal research would enhance the reporting quality of key study design and conduct factors in animal-based nutrition research.

Introduction

Nutrition research is essential for establishing nutrient requirements and dietary guidance to ensure the proper growth, development, and maintenance of health for both humans and animals. Animal studies provide an opportunity to directly observe biochemical and physiological interactions and responses to food substances that otherwise cannot be observed in humans (1). As such, basic nutrition research, especially in animal model systems, informs causal relations between specific food substances and health outcomes and allows for the elucidation of specific mechanisms of action. However, the ability to incorporate animal-based research into nutritional risk assessment depends on its quality. Despite the large quantity of published animal studies, study quality and reporting impact the accuracy and reproducibility of observations.

Findings from preclinical studies provide the plausibility needed to support the design of clinical trials. However, knowledge translation from animal studies to human research is dependent on transparent reporting. Poor and incomplete reporting of animal studies limits their utility and validity, a concern for all stakeholders, including researchers, journals, funding agencies, academic institutions, and medical and public health professionals (2, 3). A systematic review of animal-based studies published between 1999 and 2005 demonstrated common major omissions of basic animal characteristics (e.g., strain, sex, age, etc.) and environmental conditions (2). These generic study design characteristics not only influence the results but are required to reproduce and appraise the findings (3, 4). For instance, underreporting of animal strains can mask existing biases towards the use of specific inbred strains and limits the generalizability of the findings to other strains and more broadly to other species and human populations (5, 6).

Reporting guidelines such as “Animal Research: Reporting of In Vivo Experiments” (ARRIVE) provide a checklist of the minimum details that should be reported for an animal-based study. Issues addressed include the reporting of both primary and secondary outcomes, key methodology (i.e., analytical method used, source and validation of reagents, quality-control measures), and results (i.e., unit of analysis, measure of precision) (9, 10). Animal characteristics (e.g., species, strain, sex, and developmental stages), their housing and husbandry conditions (including welfare-related assessments), and relevant baseline data are highlighted because they impact experimental outcomes (7). The guidelines address study design issues, such as sample size determination to allow replicability while minimizing selection bias, and the statistical methods used to analyze outcomes (7). Errors in study design and analysis can result in systematic biases and distort exposure–response relations, emphasizing the need for clear and transparent reporting of study design details. However, generic reporting tools do not highlight nutrition-specific reporting issues in basic nutrition research studies in addition to generic study design details. Nutrition-specific reporting issues include the composition of base diets, duration of dietary intervention, dietary intervention dose, and verification of intervention/exposure by assessing the diet directly or by assessing nutritional intake/status indicators.

Our objective was to perform a scoping review of recent basic nutrition research to determine the frequency of reporting of a selection of fundamental generic study methodology items, including sex, strain, and age of mice studied, as well as the frequency of reporting of nutrition-specific items, such as base diet, nutrient intervention dose, exposure verification, and intervention duration. As a case study for basic nutrition research, we focused the scoping review on recent folate research in mice (since 2009), a field with recent exponential growth in publications. We assumed that reporting in folate research would generally reflect reporting in other nutrition research areas.

Methods

This scoping review is reported based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses–Extension for Scoping Reviews (PRISMA-ScR) (8).

Search strategy

A literature search to retrieve articles that included studies of the effect of dietary folic acid (FA) in mouse models was conducted by a Health Canada librarian.

Electronic databases EMBASE, Medline, Food Science and Technology Abstracts (FSTA), Global Health, and International Pharmaceutical Abstracts (IPA) were searched in the Ovid platform. All database searches were restricted to publication dates from 2009 onwards and English-language articles. The original search period was 1 January 2009 to 3 May 2019. A search update was conducted to capture articles published up to 7 July 2021. Detailed search terms and strategies are presented in Supplemental Tables 1–4. The search results were uploaded into the web-based bibliographic manager RefWorks 2.0 (ProQuest LLC) database. Duplicate articles identified among databases were removed in RefWorks (Figure 1).

FIGURE 1.

Study selection flow diagram. Flow chart of the identification of original peer-reviewed research articles that reported folic acid dietary interventions in mice included in the scoping review. Only English articles published between 1 January 2009 and 7 July 2021 were considered.

Selection of studies

Literature search restrictions were used to focus on more recent evidence, a period during which folate research experienced an exponential growth in publications. The inclusion criteria used to select published articles for analysis in this scoping review were articles that 1) were original and peer-reviewed, 2) used a mouse model, 3) conducted a dietary FA intervention using food as a vehicle, 4) were in the English language, and 4) were published between January 2009 and July 2021. Articles were screened for inclusion by 1 reviewer (EM) and reviewed by 1 other reviewer (NAB or AJM).

Data extraction and charting

The selected articles were divided into 2 groups depending on whether the article described a study that examined FA interventions in pregnant or nonpregnant mice. Data from each article was extracted into a Microsoft Excel spreadsheet (Macintosh version 16.30; Microsoft Corporation). If an article presented multiple independent experiments, each experiment was treated as an independent “study”. Five articles included both a “nonpregnant” and a “pregnant” mouse experiment; therefore, the 5 articles were included in both study groups. For studies in which experimental mice were exposed to 2 different interventions, each intervention dose and duration was considered individually. Data were extracted by EM. In cases where an item was not clearly reported, a second reviewer (NAB or AJM) reviewed the article to determine whether an item was reported or not. Any disagreements were resolved through discussion between 2 reviewers or further adjudication by the third reviewer.

Full references for each article were extracted including the article title, authors, journal, pages, country, and publication year. Generic data extracted from each study included diseases and/or mechanisms under study, mouse strains, sex, and age at the start of the intervention(s). Substrains of mice were grouped by strain. Additionally, nutrition-specific data extracted included the FA dose of the control and intervention(s), base diets and their FA content, intervention duration, and method of folate status verification. The FA content in diets was recorded in milligrams per kilogram. Information on FA content in base diets was either obtained from the article (reported), from previously reported diet formulations when a reference was reported (declared), or the manufacturer's website (declared) when the diet's catalog number was provided. Closed-formula base diets and base diets reported only as normal or standard rodent chow were grouped together as standard rodent chow. Unless otherwise stated in the article, when a standard rodent chow was used, we defined it as the control diet. Studies that used a standard rodent chow as a base diet but did not report ingredient formulation and the formulation could not be found on the manufacturer's website were defined as having an unreported FA content.

For pregnancy studies, data (when available) were extracted separately for the dams, sires, and offspring, as applicable for the study design. Pregnancies whereby a live birth was achieved was classified as having an offspring endpoint. Studies whereby the dams were killed during the pregnancy for embryo analysis were classified as having an embryo endpoint. Multigenerational studies were treated as a single study when the dietary intervention was the same for all generations. For multigenerational studies in which different generations were given different dietary interventions, each generation was considered a unique “study”.

Results

The literature search identified 1446 articles published between 1 January 2009 and 8 July 2021. After screening and assessment, 312 original research articles were identified in which a dietary FA intervention in mice was described. The flow chart representing article identification and selection is shown in Figure 1. Included and excluded studies and extracted data can be found in Supplemental Table 5. Of these articles, 186 described a study in nonpregnant mice, 121 articles described a study in pregnant mice, and 5 articles described studies in both pregnant and nonpregnant mice. Review articles (n = 61), abstracts (n = 106), supplements (n = 63), short communications (n = 4), conference proceedings (n = 65), university theses (n = 10), cell culture studies (n = 23), articles with no FA intervention (n = 75), and articles with FA interventions via daily injections (n = 10), gavage (n = 16), or water or milk as a vehicle (n = 23) were excluded. Additionally, articles in humans (n = 7) or those that used an animal model other than mice for their dietary FA intervention were excluded [n = 12; rats (n = 10), avian (n =1), and monkeys (n = 1)]. Finally, non–English-language papers (n = 4), duplicate articles (n = 4), book chapters (n = 2), and an article whereby the full text could not be located (n = 1) were removed.

Health outcomes investigated

A wide variety of health outcomes and biochemical processes were investigated in the articles selected for this review. Articles were grouped into broad topic areas, including cancer, cardiovascular diseases, congenital defects, gastrointestinal diseases, genetic instability, immune responses, liver diseases, neurodegeneration, urinary disorders, as well as various biochemical/metabolic pathways (Table 1). Biochemical or metabolic pathways (n = 47; 25%) and neurodegeneration (n = 42; 22%) were the most common health outcomes examined in nonpregnant studies. In pregnancy studies, congenital defects (n = 47; 37%) and genetic instability (n = 29; 23%) were the most common outcomes examined.

TABLE 1.

Health outcomes investigated

Health outcome or biochemical processes	No. of articles with studies in nonpregnant mice, n (%)	No. of articles with studies in pregnant mice, n (%)
Biochemical or metabolic pathway	47 (24.6)	14 (11.1)
Cancer	33 (17.3)	8 (6.3)
Cardiovascular	23 (12.0)	5 (4.0)
Congenital defects	0 (0.0)	47 (37.3)
Gastrointestinal and urinary disorders	13 (6.8)	9 (7.1)
Genetic instability	15 (7.9)	29 (23.0)
Immune response	5 (2.6)	1 (0.8)
Liver disease	13 (6.8)	2 (1.6)
Neurodegeneration	42 (22.0)	11 (8.7)

Mouse strain

Most studies (99%) reported mouse strain. A total of 34 strains and 24 substrains were identified in the studies. The percentage representation of mouse strains used in the studies is shown in Figure 2. Most studies used C57BL/6 (n = 244; 53%) or BALB/c (n = 52; 11%) mice. Specific strains and substrains identified in the studies are presented in Supplemental Table 6.

FIGURE 2.

Representation of mouse strains used in folate research studies published between January 2009 and July 2021. Low representation strains were defined as strains used in <5% of all studies. These included: CD1, C3H, Swiss, AJ, CAST/EIJ, FVB, PWK/PHJ, WSB/EIJ, Big Blue, CBA, Intercrossed Outbred, Mixed Inbred, 634Mui/J, Athymic nu/nu, Beige Nude XID, SCID, Kumming, Mutamouse, Tg, SJL, SMV, SMV, DBA, LPT/LeJ, LM/BC, NIH, Curly Tail, CF1, SAMP8, ddY, ICR, and their substrains.

Sex

Among the 191 nonpregnant studies reviewed, 2 did not report the sex of the mice. The majority of the studies used male mice (n = 108; 57%), followed by studies with both male and female mice (n = 54; 28%) and fewer used only female mice (n = 27; 14%) (Figure 3A). Of the studies that used male mice, 2 examined male reproduction endpoints, which explains the use of only male mice. For obvious reasons, the majority of pregnancy studies reported a dietary intervention in female mice (pregnant dams), although 10 studies examined the paternal dietary effect in sires and their offspring. For the sex analysis of pregnancy studies, sex was identified for all mice for which an endpoint was measured. Of the pregnancy studies, 76 studies examined embryos and 70 studied live-born offspring. When embryos were analyzed, most studies examined both sexes (n = 74; 97%) (Figure 3B). When live-born offspring were analyzed, both sexes were most often examined (n = 42; 60%), followed by male offspring (n = 22; 31%); female offspring alone were infrequently studied (n = 6; 9%). When embryo/live-born offspring analyses were performed in only 1 sex, it was predominantly in males (n = 24/30; 80%).

FIGURE 3.

Representation of mouse sex reported in nonpregnancy (A) and pregnancy (B) studies published between January 2009 and July 2021. For pregnancy studies, the sex of the offspring is indicated depending on whether the study examined embryos or live-born offspring.

Age

Sixty studies (17%) did not report the age of the mice at the start of the dietary intervention (Figure 4A). This was more common among pregnancy studies, with 50 pregnancy studies (34%) compared with 10 nonpregnancy studies (5%) not reporting age (Figure 4B, C). The age at time of intervention among nonpregnancy studies varied substantially from 3 to 91 wk (Figure 4A), but the majority (n = 144; 67%) used younger mice aged 3–9 wk. In pregnancy studies, the age range of the dams and sires at the time of dietary intervention was smaller, varying from 3 to 25 wk, and the majority of the studies (n = 80; 54%) used younger mice aged 3–9 wk.

FIGURE 4.

(A) Representation of the age of mice at the start of the dietary intervention reported in studies published between January 2009 and July 2021. (B) Percentage of nonpregnancy studies that reported the age of the mice at the start of intervention. (C) Percentage of pregnancy studies that reported the age of the mice at the start of intervention.

FA intervention dose

All studies reported at least 1 FA intervention dose and 1 comparison group. The FA content of the control diet was not reported in 10% of studies (n = 30), primarily due to the use of a standard rodent chow with no FA content reported or declared. Of the studies that reported the FA content of the control diet, the dose ranged from 0 to 16 mg FA/kg (Figure 5). This range did not include 1 outlier study in the nonpregnant category that reported a 200 mg FA/kg control diet; it was assumed to be a typo but the authors could not be reached for verification. A control diet containing 2 mg/kg was most common (n = 182; 68%). Deficient diets ranged from 0 to 3.42 mg/kg of FA, with the majority (n = 171; 94%) containing 0 mg FA/kg. FA-supplemented diets were highly variable, ranging from 0 to 50 mg/kg of FA, with the most common supplemented diet containing 8 mg FA/kg (17%). This did not include 2 outlier studies in the nonpregnant category that reported 15,004 mg/kg and 252.6 mg/kg of FA; they were assumed to be typos but the authors could not be reached for verification. The ranges of FA content overlapped for all diet definitions, including deficient and supplemented.

FIGURE 5.

Folic acid dose and definition of intervention groups. Representation of the number of nonpregnancy (A) and pregnancy (B) studies that reported the FA content of intervention diets defined as control, deficient, and supplemented used in both nonpregnancy and pregnancy mouse studies. The size of the circle is relative to the number of studies reporting an FA dose in that range. FA, folic acid.

Base diet

Over half of the studies (n = 334; 63%) reported the use of a defined open-formula base diet (Figure 6A). The AIN-93 diet (9) was the most commonly used open-formula diet (n = 190; 36%). For analysis, normal and standard rodent chow diets were grouped as closed-formula standard rodent chow. By this definition, 71 (36%) nonpregnancy studies used a standard rodent chow base diet, of which 46 (65%) reported/had a declared FA content. Among the pregnancy studies, 125 (38%) used a standard rodent chow base diet, of which 107 (86%) reported/had a declared FA content (Figure 6B, C).

FIGURE 6.

(A) Base diets reported in both nonpregnancy and pregnancy mouse studies. (B) Number of studies that used an open-formula diet versus a closed-formula base diet in nonpregnancy and pregnancy mouse studies. (C) Number of studies that reported folic acid content (but not necessarily measured) in the closed-formula base diets used in nonpregnancy and pregnancy mouse studies.

Duration of intervention(s)

The duration of intervention among nonpregnancy studies ranged from 4 d to 104 wk (Figure 7A). Two studies (1%) did not report the duration of intervention. Among the pregnancy studies, FA dietary intervention duration ranged from 6 d to 19 wk. However, the initiation of the FA dietary interventions relative to the pregnancy varied, with some being initiated before mating, during pregnancy, and/or during lactation, and with variable intervention duration (Figure 7B). The majority of the studies initiated the dietary intervention before mating and maintained it throughout pregnancy (43%), followed by studies where the intervention started before mating and was continued until the pups were weaned at 3 wk old (mating, gestation, and lactation) (35%).

FIGURE 7.

(A) Duration of folic acid dietary interventions reported in nonpregnancy mouse studies. Studies were grouped into 3-wk intervals up until 18 wk. Longer-duration studies were grouped into 18–30, 30–60, and >60 wk. (B) Duration of folic acid dietary interventions reported in pregnancy mouse studies. Studies were grouped into 3-wk intervals up until 21 wk.

FA intervention verified by biomarker

The exposure to folate from a dietary intervention is dependent on both the folate content of the base diet (in nonformulated diets) and the intervention dose. Folate exposure can be assessed by measuring the folate content of the diet directly and by assessing folate status biomarkers in response to dietary intake. Here we assessed the reporting of plasma/serum/tissue folate or plasma/serum homocysteine to verify folate exposure. Approximately 60% of all studies verified folate status in exposed mice using 1 of these biomarkers (nonpregnancy studies: n = 120, 63%; pregnancy studies: n = 72, 57%) (Figure 8A). Studies that used standard rodent chow as the base diet were less likely to verify folate status in the mice (49%) than studies that used open-formula diets (68%) (Figure 8B).

FIGURE 8.

(A) Number of studies that reported using a biomarker to verify folate status of the mice used in FA intervention studies. Acceptable methods of folate status verification included circulating or tissue folate or circulating homocysteine analyses. (B) Number of studies that verified the reported FA content in the base diet. FA, folic acid.

Overall findings

Of the studies reviewed, 17% did not report at least 1 of the generic items assessed and 40% did not report at least 1 nutrition-specific item. Studies published in nutrition and dietetics journals were equally likely to not report generic items compared with studies from non-nutrition journals (median nonreported generic items: 0 vs. 0, respectively; P = 0.18, Mann-Whitney rank-sum test). However, studies from nutrition and dietetics journals were less likely to not report nutrition-specific items (median nonreported nutrition-specific items: 0 vs. 1; P = 0.003, Mann-Whitney rank-sum test).

Discussion

Incomplete reporting is not unique to nutrition science (2), but it needs to be addressed to minimize negative implications for reproducibility, ethical animal use, funding allocation, translation to human health, and policy making to name a few (3, 10, 11). Here we found that nutrition studies using mouse models frequently do not report a selection of key generic and nutrition-specific details. Animal models are vital for the investigation of causal pathways that underpin health outcomes and are the foundation on which clinical trials are designed. The unique characteristics of the animal models and study designs used can influence the experimental results and therefore must be accurately reported (4). Systematic reviews investigating the reporting quality of animal-based studies (2) highlight major reporting omissions, including the hypothesis, sample size, basic animal characteristics (e.g., age, sex, strain), actions to reduce bias, and statistical methods. Here, we identified key nutrition-specific items, including base diet source and composition, intervention dose, intervention duration, and exposure verification that warrant proper reporting in basic nutrition research studies. Of the studies reviewed, 17% did not report at least 1 of the limited generic items assessed and 40% did not report at least 1 nutrition-specific item, indicating that reporting omissions are common in basic nutrition research. Non-reporting of nutrition-specific items was more common in non-nutrition and dietetics journals. As a case study, we limited our scoping review to a handful of fundamental study design details (strain, sex, and age) as a proof-of-principle assessment to illuminate the reporting of nutrition-specific details in basic nutrition research. While the majority of studies in our review reported these basic details, it is most probably an underestimate of poor reporting quality.

Impacts of poor reporting

The majority of studies reported the basic animal characteristics assessed in this review, with 99%, 99%, and 83% of studies reporting sex, mouse strain, and age at intervention, respectively. The frequency of reporting of these foundational details was better than previously reported for mouse studies, where only 65% and 68% of studies reported sex and age or weight of mice, respectively (2). For example, Kilkenny et al. (2) found that approximately 45% of mouse studies reported the hypothesis, sex, strain, age and/or weight, and number of animals used, whereas >70% of studies demonstrated low-quality reporting of experimental design and statistical analyses. Reporting of these seemingly simple but foundational details is important because poor reporting can lead to erroneous conclusions and/or irreproducibility of experimental findings (6). For example, sex differences account for differences in gonadal steroid hormones, body size, metabolic rate, epigenetic programming, and gene expression (12–15), which can influence responses to nutritional interventions. Our study also highlighted the known publication bias towards the use of male animals in preclinical studies; 57% of the nonpregnancy studies were conducted only in male mice and only male live-born offspring were studied in 31% of pregnancy studies. The assumption that findings from single-sex (male) studies can be applied generally has been repeatedly disputed (12, 16). Such publication bias inflates interventions and treatment outcomes that may be ineffective in the generalized population (6). Accounting for variations due to sex in preclinical trials is an effective way to avoid costly and potentially ineffective clinical trials (17). Recognizing the influence of sex on experimental outcomes, international agencies such as the NIH, the Canadian Institutes of Health Research (CIHR), and the European Union (EU) now require the sex of animal models to be reported in publications and funding requests, and justification for the use of single-sex animals in studies (12, 17–19). These policy changes may account for the more consistent reporting of sex in the fairly recently published studies included in this review.

Animal strain–specific genetic and phenotypic variability can also impact metabolism and disease pathogenesis (17, 20). Genomic diversity and its manipulation through the use of transgenic tools allow researchers to investigate the influence of individual genes on disease pathogenesis, intervention effectiveness, metabolic rates, susceptibility to disease, and more (11, 20–22). Mouse strains are selected based on phenotypes defined by their genomic background and can be more or less susceptible to a nutritional intervention and/or the development of specific health outcomes, making this an important reporting detail (11). Almost all studies assessed in the scoping review reported mouse strain. Three inbred mouse strains dominated (˜70%) FA research studies, but we recognize that this is likely an underestimate of diversity because we grouped together substrains of major strain types, which can be genetically and phenotypically distinct (23). The use of single strains or transgenic lines can limit the generalizability of findings to other strains, and by extension to human populations.

The age of mice at the start of intervention was less likely to be reported than sex and strain. Age varied substantially among studies (i.e., nonpregnancy studies: 3 to 91 wk; pregnancy studies: 3 to 25 wk). Variability in age (or body weight as a proxy for age) at intervention can be necessary and health outcome–dependent (i.e., an aging study performed in older mice), but should be reported and rationalized given that metabolic rate, physiological response, and disease pathogenesis are age-dependent (24). Animal characteristics, including others not assessed in this scoping review, can interact to influence experimental outcome (25, 26), making it essential that all details are reported consistently and transparently.

Impact of nutrition-specific details

Animal-based research allows for the assessment of controlled nutritional interventions on outcomes, but this assumes we have an accurate assessment of the nutritional exposure. We found that about two-thirds of the studies utilized an open-formula diet, most often AIN-93. Open-formula defined diets were developed in the 1970s to address the growth, maintenance, and developmental needs of rodents, to standardize laboratory animal diets, and to facilitate interstudy comparisons (9, 27, 28). Surprisingly, over one-third of the studies used closed-formula diets (37%), which included studies that provided no details on the brand, manufacturer, or FA content of the “standard rodent chow” used. In contrast to open-formula diets, the ingredient list of closed-formula diets is often proprietary, only available upon request, and subject to change without public disclosure (27). Diet formulations can change in response to fluctuations in ingredient costs and seasonal or geographic variations in source materials (27). Standard rodent chow can therefore contain variable concentrations of the nutrient of interest, as well as variable amounts of bioactives and contaminants that may impact study outcomes (27, 29). Assessing diet–health outcome relations depends on the overall exposure from both the base diet and the added intervention dose. For these reasons, if a closed-formula diet is used, the absolute amount of exposure to a food substance may not be known and/or may vary considerably. Overall, the use of closed-formula diets with no information on nutrient content hinders the ability to compare and reproduce basic nutrition studies and should be considered a fatal flaw in study design. Ideally, investigators would use open-formula diets and declare information such as the manufacturer, catalog number, and lot number.

Building on the idea of exposure assessment, we investigated whether studies reported measuring a biomarker of exposure to validate the nutritional exposure, thereby confirming the relation between the exposure and health outcome (30, 31). In the case of folate, plasma/serum or RBC/other tissue folate were considered biomarkers of exposure (i.e., dietary intake or status), and homocysteine a functional biomarker (i.e., metabolism) (34–36). Less than two-thirds of studies reported measuring a folate exposure biomarker, and this was more common among studies that used an open-formula diet. A lack of nutritional exposure verification combined with the common use of closed-formula diets conveys a high degree of uncertainty in relating an exposure to the experimental outcome of interest. Ideally, all studies would verify exposure by measuring the food substance of interest directly in the diet and/or analyzing biomarkers of exposure in the animals.

Although not a reporting issue, we found a remarkable degree of variation and overlap among the definitions for what constituted a control, deficient, and supplemented FA intervention. Open-formula diets generally define adequate FA as 2 mg/kg (9, 28) and this was indeed the most commonly used control dose reported. However, the definition of the control diet ranged from 0 to 16 mg FA/kg. Similarly, the definitions of a deficient (0 to 3.42 mg FA/kg) or supplemented (0 to 50 mg FA/kg) diet contained a large and overlapping range of FA doses. The inconsistent definitions for “deficient,” “control,” and “supplemented” underscore the need for standardization of terms and definitions in nutrition research, along with clear reporting of intervention doses, to facilitate study comparisons, consolidate findings from different studies, and improve reproducibility.

Recommendations

Improving the reporting quality of publications in nutrition science is one way to enhance the accuracy, transparency, and reproducibility of findings. Based on our scoping review, we propose a number of recommendations to address under- and/or misreporting of generic study design and implementation details, as well as nutrition-specific items in animal studies (summarized in Table 2). First and foremost, we recommend the use of the ARRIVE guidelines; ideally, this would be a requirement for publication. The ARRIVE reporting checklist outlines essential study design details that should be included in animal-derived publications to facilitate accurate reporting and enhance reproducibility (2, 7, 10). The ASN journals encourage authors to follow and upload appropriate reporting checklists, but only the American Journal of Clinical Nutrition (which generally does not publish animal-based research) requires it.

TABLE 2.

Concepts that should be considered when reporting nutrition research that utilizes animal models

Topic and concept	Recommendations
Experimental animal model
Rationale for use	Provide rationale for model used including its relevance to the health outcome being investigated and the food substance under investigation
Characteristics	Report details about the species, strain, sex, developmental stage, age, and body weight of the animal model used. Provide brief details to explain choice of strain, sex, developmental stage, age, and body weight of animal model used. Indicate relevance to the food substance intervention and outcome. Comment on randomizations, blinding, statistical methods, and actions to reduce bias.
Outcome	Provide a rationale for investigating the relation between the food substance of interest and the health outcome. Report the use of a surrogate biomarker or health status indicator to verify experimental heath outcome under investigation. Comment on the choice of tissue(s) analyzed for health outcome and relevance to the food substance under investigation. Provide a rationale for investigating nonoral routes for food substance interventions and how it affects the interpretation of results.
Nutritional component
Base diet	Provide details of diet formulation (i.e., complete ingredient list and their relative concentrations), manufacturer and catalog number, and lot numbers. If an open-formula diet was not used, provide a rationale for why a closed-formula diet was used. Indicate whether the concentration of the food substance of interest was verified in the diet. Provide details on method used to verify composition of diet. Comment on environmental factors that may influence dietary intake or confounding factors that might otherwise impact food substance exposure.
Intervention dose	Give full details on the concentration of dose used in the intervention, including that of the comparison diet (control); explain whether or not nutritional component was modified in the base diet or administered differently. Total exposure (base diet plus added amount, if applicable) should be reported. This would include consideration of food consumption to ensure equivalent intakes among groups. Provide a rationale for the doses used and relevance to adequate intakes (when established for an essential nutrient) or usual intakes for other food substances.
Duration of exposure	Provide a brief description to justify chosen duration of dietary intervention, including how it relates to the development of the outcome being measured.
Verification of exposure	Comment on the method used to verify exposure, such as the measurement of biomarkers of intake or status. Provide a rationale for the choice of tissues analyzed for the biomarkers of exposure.

The ARRIVE guidelines consist of 2 checklists: the Essential 10 and the complementary Recommended Set composed of an additional 11 reporting items (32). The Essential 10 checklist addresses fundamental generic items including study design, sample size, inclusion/exclusion criteria, randomization, blinding, outcome measures, statistical methods, experimental animals, experimental procedures, and results. The Recommended Set expands on the Essential 10 to address reporting of the abstract, background, objectives, ethics, housing and husbandry, animal care and monitoring, interpretation/implications, generalizability/translation, protocol registration, data access, and declaration of interests.

Requiring the use of the ARRIVE guidelines will most certainly enhance the quality of reporting in basic nutrition research, but they do not specifically address important nutrition issues. The ARRIVE guidelines describe fairly broad general reporting items. We would suggest expanding the guidance descriptions of the study design to include a rationale for the control and intervention levels used, as well as the route, timing, and duration of exposure. For essential nutrients, an appropriate control intervention would be the level of intake required for adequacy, supported by evidence of an intake–response relation between the nutrient intake and a nutrient deficiency–related health outcome. For nonessential food substances, a more nuanced approach would be needed to determine appropriate levels of exposure for comparison groups, such as based on levels naturally occurring in diets. Either way, a rationale for the intervention doses will facilitate interpretation. In addition, a number of studies identified in our review used a route of exposure other than oral intake of diet. It would be important to include a rationale for exposing animals to a food substance via a nonoral route and how it affects the interpretation of the results. Furthermore, our scoping review saw a wide range in the timing and duration of the interventions. The timing of initiation/duration must be sufficient to plausibly affect a change on the outcome. For example, spermatogenesis takes 6–8 wk, so studies examining male fertility should take into account whether the timing and duration of the nutritional intervention of interest would plausibly affect fertility parameters.

Of critical importance is the reporting of details about the base diets used. The diet composition (either as a table in the article or citation), manufacturer, and catalog, and lot numbers should be reported. Lack of knowledge of total exposure (e.g., base diet content plus supplement) prohibits the appropriate interpretation of results; conclusions and inferences to human health are thereby also impossible. Open-formula diets should be used to better control the total exposure, minimize errors and uncertainties, and increase reproducibility of findings. Ideally, the amount of the food substance should be verified either in the diet itself or through validated biomarkers of exposure in the animals. In addition, food consumption can be measured to ensure equivalent intakes/exposure among experimental groups. In cases where a closed-formula diet is utilized, the complete diet formulation should be included, the food substance of interest level verified, and the biomarker of exposure assessed. Nutrition-specific items are described in Table 2 and this list could be used in tandem with the ARRIVE checklist to improve the reporting of nutrition details.

In addition to reporting checklists, there are numerous other tools available to enhance transparency and reproducibility in animal-based research. The Planning Research and Experimental Procedures on Animals: Recommendations for Excellence (PREPARE) checklist is a guideline for planning animal studies with consideration of issues such as ethics, experimental design, statistical analysis, animal care, and more (33). One step further is the preregistration of animal studies, with a public platform such as the Animal Study Registry, which includes a description of the planned study design and statistical analysis, thereby increasing transparency and reducing publication bias (34). Further, the Core Outcome Measures in Effectiveness Trials (COMET) initiative promotes the use of core outcomes defined as a “standardized set of outcomes that should be measured as a minimum” in clinical trials allowing for comparison and combination of data across studies (35, 36). Similarly, verified outcomes and biomarkers should be considered in preclinical studies. The effects and health outcomes examined in nutrition science are highly diverse, as observed in the suite of studies identified in this review. A diversity of studies is imperative to fill specific knowledge gaps, but the use of nonvalidated surrogate or disease outcomes can reinforce, undermine, or introduce inconsistencies in the evidence base, complicating interpretation. Finally, tools such as Systematic Review Centre for Laboratory Animal Experimentation (SYRCLE) can assess the risk of bias in animal studies—for instance, as part of a systematic review—facilitating their critical appraisal (37). Together, these tools can enhance transparency and reproducibility of animal studies and improve their potential to be translated to clinical practice.

Conclusions

Basic research informs the knowledge foundation used by the general public, research scientists, clinicians, and policy makers in decision making (38). However, research can be influenced by trends and underlying stakeholder interests, and by the media to either promote or undermine the credibility and trust in scientific evidence (39). When publications are found to be conflicting, incomplete, and/or irreproducible, they can erode the public's trust—poor reporting of scientific findings contributes to these challenges. As our scoping review demonstrates, there is a need for better reporting of animal-based nutrition research. The use of reporting guidelines such as ARRIVE to report animal study design details, and interpreted with consideration of nutrition-specific concepts, will improve the transparency, reproducibility, and reliability of these studies.

Notes

Supported by Health Canada A-base funding.

Author disclosures: The authors report no conflicts of interest.

Supplemental Tables 1–6 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/advances/.

Data Availability

Data described in the manuscript is available in the Supplemental Material.