Reliability and Validity of Nutrient Assessment Applications for Canadian Endurance Athletes: MyFitnessPal and Cronometer

ABSTRACT

Introduction

The purpose of the present observational study was to assess the inter‐rater reliability and validity of two free nutrition apps, MyFitnessPal (MFP) and Cronometer (CRO), among Canadian endurance athletes.

Methods

Two raters independently input 43 three‐day food intake records (FIR) (27M/16F) into MFP and CRO, and one rater input each FIR into ESHA Food Processor® using the reference standard 2015 Canadian Nutrient File (CNF) database.

Results

MFP showed a consistent difference in measurements between raters for total energy and carbohydrates (absolute reliability), but the difference was too small to be clinically meaningful; further, raters were inconsistent for sodium and sugar (relative reliability), particularly among men. MFP showed poor validity for total energy, carbohydrates, protein, cholesterol, sugar, and fibre, with discrepancies for total energy, carbohydrates and sugar being driven by women, and protein differences by men. Rationale for low reliability and validity may be due to the copious options for each food entry in MFP, including non‐verified consumer entries. Possible rationale for gender differences may be more detailed and descriptive FIRs (reliability) and generally more varied dietary patterns (validity) in women compared to men. Conversely, CRO showed good to excellent inter‐rater reliability for all nutrients and good validity for all nutrients except for fibre and vitamins A and D, with no differences between genders. Rationale for low validity of fibre may be due to how it is represented in the software (i.e., total vs. soluble), and rationale for vitamins A and D may be due to fortification practices differing between brands and countries. Bland‐Altman plots for inter‐rater reliability and validity revealed smaller bias, narrower LOAs, and better horizonal spread of data when using CRO compared to MFP.

Conclusion

Given the unique energy and nutrient needs of athletes, they should be aware that MFP may provide dietary information that does not accurately reflect true intake, and that CRO could serve as a promising alternative.

Summary

• Nutrition tracking applications can support athletic success and overall health.

• MyFitnessPal provides low reliability and validity for Canadian endurance athletes.

• Cronometer provides good reliability and validity for Canadian endurance athletes.

Abbreviations

BMI

body mass index

CNF

Canadian Nutrient File

CRO

Cronometer

ESHA

Elizabeth Stewart Hands and Associates

FIR

Food Intake Record

ICC

Intraclass Correlation Coefficient

LOA

Limits of Agreement

MDC

minimal detectable change

MFP

MyFitnessPal

MUFA

monounsaturated fatty acids

NCCDB

Nutrition Coordinating Centre Food and Nutrient Database

PUFA

polyunsaturated fatty acids

SEM

standard error mean

SFA

saturated fatty acids

USDA

United States Department of Agriculture National Nutrient Database for Standard Reference.

1. Introduction

Dietary assessments are valuable for evaluating an individual's health status and monitoring both short‐ and long‐term nutritional goals. Traditional dietary assessment methods include food frequency questionnaires, 24‐h food recalls, 3‐ or 7‐day food intake records (FIR), and weighted food records [1]. Though each dietary assessment method has its strengths, some common limitations include participant literacy and memory, participant and clinician burden, motivation to accurately record intake, and the need for trained personnel to administer certain tools [2]. Mobile and web‐based nutrition applications are an increasingly popular alternative that minimise the limitations of traditional methods of tracking dietary intake [1]. In a cross‐sectional web‐based survey of Canadian dietitians, 57% reported the use of an application in their own practice, 54% had a client ask about the use of a nutrition or food application, and 41% recommended one to their clients [3]; these findings reflect the increasing use of dietary tracking applications in the general population and in practice.

A sub‐population that may particularly benefit from these nutrition tracking applications is endurance athletes that require increased caloric intake (energy expenditures can be up to 2‐3x higher for endurance athletes compared to non‐athletes [4]) and higher amounts of key micronutrients (e.g., fat‐ and water‐soluble vitamins, iron, calcium, sodium, potassium, etc.) [5]. Knowledge of nutrition intake can play a key role in optimising athletic performance [5]. Indeed, a cross‐sectional online survey across five countries (Australia, Canada, New Zealand, the United Kingdom, and the United States) found that 32% of sports dietitians reported using a mobile application to track their athletes' dietary intake [6]. One of the most popular applications used for these purposes is MyFitnessPal (MFP); not only does it have over 200 million users globally with over 19 million items in its database, but 56% of sports dietitians opt to use MFP [6], and it was one of the most commonly recommended amongst Canadian dietitians [3]. Another popular application, Cronometer (CRO), has over 10 million users and supports the tracking of up to 84 nutrients from food databases representing over 1.2 million dietary items [7].

Although the potential of nutrition tracking applications for supporting athletic success and overall health is high, the reliability and validity of these applications have been cited as barriers to their use [6, 8]. A recent systematic review of dietary assessment app validity studies found that all 12 apps [including MFP] underestimated energy intake when compared to their reference standard [9]. One key factor that may affect the validity of these apps is the databases from which nutrition information is drawn. For example, items in MFP are either uploaded by users around the world or derived from the United States Department of Agriculture National Nutrient Database (USDA) as the standard reference database [10], and items in CRO are sourced from verified databases, including the Canadian Nutrient File (CNF) and the USDA [7]. In addition, different food industry regulations (e.g., fortification, enrichment) impact the amount of micronutrients in food products from a particular country [11]. Poor validity of MFP has been reported from studies in the United States [10, 12], Australia [13], Brazil [14], Belgium [15], Japan [16], and the United Kingdom [17]. Only one study has evaluated the validity of MFP in a Canadian population and found it was acceptable to assess energy, macronutrient, and selected micronutrient intakes [18], however this study was in military personnel who only consumed standard military rations. Limited literature exists on the validity of CRO; one study in the United States compared a researcher‐developed 3‐day food log to the Nutrition Data System for Research database and found that it underestimated caloric and macronutrient intake [19]. Another study input three sample diets developed by Registered Dieticians in Australia into Foodworks.online as the reference standard and found that Cronometer underestimated fat and sugars and overestimated carbohydrates [20]. Of note, none of the studies assessing the validity of MFP or CRO assessed reliability as well.

The purpose of the present study was to assess the inter‐rater reliability and validity of MFP and CRO, in comparison to the CNF, in a free‐living setting among Canadian endurance athletes. It was hypothesised that inter‐rater reliability would be better for CRO compared to MFP [due to the use of verified databases vs. user uploads], and that macronutrient outputs would be similar across MFP, CRO, and CNF, but not micronutrient outputs [due to varied fortification practices]. Findings of good reliability and validity can improve clinicians' confidence in the use and recommendation of these apps, as well as increase client engagement and adherence.

2. Materials and Methods

2.1. Participants

Participant recruitment and data collection for the present observational study took place from January to August 2020. Sample size was determined based on calculations for agreement studies: at an alpha of 0.05 and a beta for power of 0.8, with the assumption of no discordant pairs of measurements and accounting for 10% attrition, a minimum of 50 participants was required [21]. Canadian middle‐aged (40–65 years) endurance athletes (≥8 h of moderate‐vigorous aerobic training per week, for example, marathon runners, ultra cycling, triathletes, etc.) were recruited through various channels, including targeted outreach to Canadian endurance sport communities on Facebook, direct email invitations sent to acquaintances and professional contacts, and via snowball sampling. Informed consent and eligibility criteria were obtained using an online survey platform [Qualtrics] explaining the study with the following statement: ‘By clicking “I agree” below, you acknowledge that you have read and understand the description provided, and as such consent to participate in this study study’. The study procedures were approved by the Research Ethics Board at the University of Regina (REB 2019‐212).

2.2. FIR and Nutrient Analyses

Three‐day FIR were obtained from each athlete. Participants were encouraged to maintain their typical eating habits and instructed to record all food and beverage intake over the course of three nonconsecutive days, including 2 weekdays and 1 weekend day. A sample FIR was provided to participants that outlined the level of detail required, as well as a link to an unlisted YouTube video that explained how to fill out the FIR. Prompts were given for the participants to record their meals, including meal and time consumed (e.g., breakfast, morning snack, etc.), food description, and the amount consumed. FIR were reviewed by two trained research assistants for clarity and detail, and participants were contacted if additional information was needed (e.g., lack of brand, recipe information, portion sizing). All questions asked of participants and their corresponding responses were recorded in a master document.

All food and beverage items were entered into the free app versions of MFP (v20.19.1) [22] and CRO (v2.18.6) [7] by two independent raters (O.M. and L.M.) in 2020. Automatic updates of the software were disabled during the data entry period, and barcode scans were not used. Raters were blinded to each other's inputs to minimise bias, and raters were trained and calibrated using a shared standard operating procedure. When inputting food items into MFP, nutrient data from verified sources was selected when possible; in all cases, it was ensured that the food item selected had MFP's green check mark [to indicate completeness of the food item's nutrition information by the MFP team] [23]. When inputting food items into CRO, nutrient data from the 2015 CNF was selected when possible; otherwise, items were selected from ‘Lab Analysed’ sources, including CRO's Nutrition Coordinating Centre Food and Nutrient Database (NCCDB) and the USDA. When an item was not found in either application, raters manually entered information according to the nutrition facts table or an online search. Likewise, custom recipes provided by participants were input manually by raters into both MFP and CRO. Outputs from MFP and CRO from both raters were reviewed to identify systemic errors between raters. Cut‐offs of 15% for total energy, 20% for macronutrients, and 50% for micronutrients were established by a registered dietitian (M.K.) to identify meaningful discrepancies caused by possible input errors or outliers between raters, while remaining sufficiently lenient to align with the study objectives. When these cutoffs were exceeded, individual food entries were reviewed to determine the origin of the differences. In cases where there was a rater error (e.g., an item was missed or the quantity was incorrectly entered), the entry was fixed, and a new average output was calculated. If the discrepancy was due to raters selecting different food options (e.g., different green check mark options in MFP), the entry was not changed as it was believed to reflect a limitation of dietary assessment app data entry (i.e., aligning with the objectives of the study).

All food record data was also input into the Food Processor software (ESHA Food Processor® 11.9.X series, Cloud Services, Salem, OR) by a trained researcher (O.M.) in 2020. Food items from the 2015 CNF were selected whenever possible. In circumstances when CNF data was not available, nutrition information was found online and entered manually into ESHA. The manually entered data was then compared to other available ESHA databases (e.g., USDA); when little‐to‐no differences were found between the manually entered and ‘other’ database nutrients, nutrient data from the ‘other' database was used. For example, if a participant specified French's mustard, the Canadian online nutrition facts label was compared to the French's product information available in ESHA; if the nutrient profiles matched, the ESHA database was used. In cases where a brand was not available, a generic version of the product was used; for example, stout was not available in the CNF, so a generic beer was input in its place.

Data from each software were exported to Excel, where the 3‐day average nutrient intake for each participant was calculated for total energy (kcal), carbohydrates (g), fat (g), protein (g), cholesterol (mg), sodium (mg), sugar (g), and fibre (g). Across all three software, sugar was defined as ‘total sugar, g’, and fibre was defined as ‘total dietary fibre, g’. Data on an additional 41 nutrients were calculated from CRO: alcohol, caffeine, water, 11 amino acids, monounsaturated fatty acids, polyunsaturated fatty acids, omega‐3 and omega‐6 fatty acids, saturated fatty acids, trans fats, water‐ and fat‐soluble vitamins, and nine additional minerals.

2.3. Statistical Analysis

2.3.1. Inter‐Reliability

Absolute inter‐rater reliability for both CRO and MFP was assessed via paired samples t‐tests [using a p‐value of p < 0.05 to determine statistical significance] and 95% confidence intervals (CI) between raters, standard error of measurement (SEM), minimal detectable change (MDC), and Bland‐Altman plots with 95% limits of agreement (LOA) [24]. SEM quantifies the amount of measurement error between observed and true differences [between raters] and was calculated as follows, where ICC is the intraclass correlation coefficient [25]:

$$\mathrm{SEM}=\sqrt{(1-\mathrm{ICC})}.$$

MDC, used to determine the minimum ‘real’ difference between raters, was calculated using the following Equation (25):

$$MDC=\mathrm{SEM}\times 1.96\times \sqrt{2}.$$

Smaller SEM and MDC suggest higher reliability and greater sensitivity to detecting true differences, respectively. Relative reliability was quantified via intraclass correlation coefficients (ICC_3,1), calculated for each nutrient from the fixed raters [25, 26, 27]. ICC values of <0.50 indicate poor reliability, 0.50–0.75 moderate reliability, 0.75–0.90 good reliability, and >0.90 excellent reliability [28]. Reliability statistics were also conducted for men and women separately.

2.3.2. Validity

Validity for both dietary assessment apps was determined by comparing average nutrient values generated from CRO and MFP to a reference standard, the 2015 CNF. One‐way Repeated Measures Analysis of Variance (RM‐ANOVA) was used to assess differences between the three dietary assessment apps (MFP, CRO, CNF). If differences were observed (i.e., p < 0.05), post hoc comparisons were performed using Bonferroni‐adjusted paired t‐tests [using a p‐value of p < 0.025 to determine statistical significance]. Effect sizes using partial eta‐squared [RM‐ANOVA] were categorised as small (0.01), medium (0.06), and large (0.14), and effect sizes using Cohen's d [paired t‐tests] were categorised as small (0.2), medium (0.5), and large (0.8) [29]. Validity statistics were also conducted for men and women separately. Individual agreement between the dietary assessment tools (CRO and MFP) and the 2015 CNF was assessed using Bland‐Altman plots with 95% LOA [24]. Lastly, paired t‐tests were used to assess differences between CRO and CNF for additional micronutrients of vitamins A, D, B3, B6, B12, as well as calcium, potassium, iron, magnesium, and selenium [using a p‐value of p < 0.025 to determine statistical significance and Cohen's d to determine effect size]. These additional micronutrients were selected because of their known differences in Canadian and United States fortification practices, as well as their importance to athletes. All data were analysed using SPSS version 28.0.1.0 (IBM, Inc., Armonk, NY), and figures were generated in GraphPad Prism version 9.0 (GraphPad Software, La Jolla, CA).

3. Results

3.1. Participant Characteristics

Of the 50 athletes that completed the 3‐day FIR, n = 7 were removed due to incomplete data. FIR of n = 43 athletes (27 M/16 F) were input into MFP, CRO, and CNF (i.e., ESHA). The sample size was similar to previous dietary app validation studies (from 18 to 81 participants) [9]. Participant characteristics were self‐reported: age 51.2 ± 6.8 years, body mass: 72.7 ± 13.9 kg, height: 1.7 ± 0.1 m, and BMI: 23.8 ± 2.5 kg/m². The athletes were endurance runners, adventure racers, triathletes, ultra‐cyclists, etc., and exercised for an average of 11.2 ± 3.6 h per week.

3.2. Reliability

Table 1 shows inter‐rater reliability [between O.M. and L.M.] expressed in both absolute and relative terms for both MFP and CRO for eight select nutrients that the MFP software provides. Table 2 shows these same results separated by gender. When using MFP, there was a consistent difference in measurements between raters for total energy and carbohydrates and a trend towards a difference for fat (absolute reliability); these mean differences, while exceeding expected measurement error (SEM), were smaller than the MDC, indicating they are likely not practically or clinically meaningful. Among the men, a difference between raters was observed for total energy, and among the women, differences between raters were observed for total energy, carbohydrates, protein, and sugar; again, these mean differences exceeded SEM but were smaller than MDC. Regarding relative reliability, the raters were not good at consistently scoring for sodium and sugar, particularly among the men (Tables 1 and 2). Both absolute and relative inter‐rater reliability were excellent when using CRO for these eight select nutrients, with no differences between genders (Tables 1 and 2). Visual inspection of Bland–Altman plots of inter‐rater reliability shows wider variability, larger bias, and/or poorer data spread for all nutrients when using MFP (Figures 1 and 2, left graphs) compared to CRO (Figures 1 and 2, right graphs).

TABLE 1.

Absolute and relative inter‐rater reliability of MyFitnessPal and Cronometer.

	Absolute reliability				Relative reliability
	Mean difference L.M.‐O.M. [95% CI]	p value	SEM	MDC	ICC3,1 [95% CI]
	MYFITNESSPAL
Total energy, kcal*	225.49 [138.94, 312.03]	<0.001	105.60	292.70	0.859 [0.524, 0.944]
Carbohydrates, g*	23.74 [9.54, 37.93]	0.002	16.57	45.93	0.871 [0.723, 0.936]
Fat, g	5.64 [−0.35, 11.64]	0.064	7.79	21.61	0.840 [0.720, 0.910]
Protein, g	3.82 [−1.04, 8.68]	0.120	4.97	13.76	0.901 [0.825, 0.945]
Cholesterol, mg	−1.49 [−37.77, 34.80]	0.934	50.44	139.81	0.817 [0.686, 0.897]
Sodium, mg†	402.01 [−201.75, 1005.77]	0.186	1495.36	4144.93	0.419 [0.144, 0.635]
Sugar, g†	−11.67 [−50.06, 26.73]	0.543	100.74	279.24	0.348 [0.054, 0.585]
Fibre, g	1.71 [−1.71, 5.12]	0.319	5.96	16.53	0.711 [0.527, 0.832]
	CRONOMETER
Total energy, kcal	17.50 [−33.90, 68.90]	0.496	30.80	85.37	0.966 [0.939, 0.982]
Carbohydrates, g	4.37 [−2.44, 11.18]	0.203	3.36	9.31	0.977 [0.958, 0.987]
Fat, g	−1.22 [−4.99, 2.55]	0.517	2.87	7.96	0.945 [0.901, 0.970]
Protein, g	1.27 [−2.55, 5.09]	0.507	2.94	8.14	0.944 [0.898, 0.969]
Cholesterol, mg	4.98 [−9.15, 19.11]	0.481	7.11	19.72	0.976 [0.956, 0.987]
Sodium, mg	−81.45 [−244.81, 81.90]	0.320	151.07	418.74	0.919 [0.857, 0.955]
Sugar, g	1.94 [−1.41, 5.29]	0.249	1.54	4.27	0.980 [0.964, 0.989]
Fibre, g	−0.42 [−1.36, 0.53]	0.379	0.53	1.48	0.970 [0.945, 0.983]

Note: Paired samples t‐tests assessed mean differences between raters (O.M. and L.M.), *p‐value < 0.05.

ICC3,1 (intraclass correlations coefficients) assessed consistency in raters' assessments relative to the variability of the food intake records, †value <0.5.

Abbreviations: MDC, minimal detectable change; SEM, standard error mean.

TABLE 2.

Absolute and relative inter‐rater reliability of MyFitnessPal and Cronometer for men and women separately.

	Men (n = 27)				Women (n = 16)
	Absolute reliability			Relative reliability	Absolute reliability			Relative reliability
	Mean difference L.M.‐O.M. [95% CI]	SEM	MDC	ICC3,1	Mean difference L.M.‐O.M. [95% CI]	SEM	MDC	ICC3,1
				MYFITNESPAL
Total energy, kcal*	215.70 [120.1, 311.3]	137.11	380.05	0.805	242.00 [170.57, 313.43]	111.07	307.86	0.771
Carbohydrates, g*	16.99 [0.76, 33.2]	18.42	51.06	0.878	35.13 [25.81, 44.44]	17.73	49.16	0.657
Fat, g	6.70 [0.11, 13.30]	10.16	28.16	0.775	3.85 [−1.13, 8.84]	6.11	16.92	0.858
Protein, g*	1.36 [−3.77, 6.48]	6.06	16.78	0.868	7.98 [3.78, 12.18]	4.87	13.49	0.873
Cholesterol, mg	−7.28 [−50.31, 35.74]	71.42	197.96	0.739	8.29 [−13.23, 29.81]	18.50	51.29	0.930
Sodium, mg†	527.64 [−223.30, 1278.58]	2117.37	5869.05	0.247	190.00 [0.28, 379.72]	198.80	551.1	0.896
Sugar, g*,†	−26.81 [−74.73, 21.10]	130.62	362.06	0.296	13.90 [7.27, 20.52]	10.6	29.2	0.760
Fibre, g	0.15 [−3.12, 3.41]	5.13	14.22	0.766	4.33 [0.72, 7.95]	8.1	22.4	0.529
				CRONOMETER
Total energy, kcal	‐0.24 [−52.14, 51.67]	36.95	102.42	0.952	47.44 [−3.40, 98.28]	46.1	127.9	0.922
Carbohydrates, g	4.98 [−2.12, 12.09]	3.500	9.70	0.977	3.34 [−3.16, 9.84]	6.1	16.8	0.918
Fat, g	−2.18 [−6.09, 1.72]	3.31	9.18	0.932	0.40 [−3.19, 3.99]	3.0	8.2	0.935
Protein, g	−0.20 [−3.48, 3.08]	2.41	6.67	0.949	3.74 [−0.087, 8.34]	5.7	15.7	0.857
Cholesterol, mg	3.39 [−9.51, 16.28]	6.21	17.23	0.978	7.67 [−8.76, 24.09]	9.1	25.2	0.971
Sodium, mg	−117.23 [−310.30, 76.30]	203.82	564.95	0.895	−21.00 [−116.82, 74.82]	62.3	172.6	0.960
Sugar, g	2.68 [−0.02, 6.37]	1.66	4.59	0.981	0.70 [−2.03, 3.43]	2.1	5.8	0.944
Fibre, g	−0.62 [−1.61, 0.37]	0.57	1.59	0.968	−0.07 [−0.97, 0.82]	0.7	2.0	0.938

Note: Paired samples t‐tests assessed mean differences between raters (O.M. and L.M.) for men (n = 27) and women (n = 16) separately); *p‐value < 0.05.

ICC_3,1 (intraclass correlations coefficients) assessed consistency in raters' assessments relative to the variability of the food intake records, ^†value <0.5.

Abbreviations: MDC, minimal detectable change; SEM, standard error mean.

FIGURE 1.

Bland–Altman plots of MyFitnessPal (left) and Cronometer (right) inter‐rater reliability. Bias between the two software's (long dashed lines) as well as 95% limits of agreement (dotted lines; UL, upper limit, LL, lower limit) are presented for total energy (kcal), carbohydrates (g), fat (g), and protein (g).

FIGURE 2.

Bland–Altman plots of MyFitnessPal (left) and Cronometer (right) inter‐rater reliability. Bias between the two software's (long dashed lines) as well as 95% limits of agreement (dotted lines; UL, upper limit, LL, lower limit) are presented for cholesterol (mg), sodium (mg), sugar (g), and fibre (g).

Table 3 shows inter‐rater reliability (i.e., between O.M. and L.M.) expressed in both absolute and relative terms for 41 additional nutrients that the CRO software provides. There was a difference in measurement between raters for iron; while the difference the exceeded the SEM it was smaller than the MDC and therefore likely not practically meaningful. Further, raters were good at consistently scoring for iron (relative reliability).

TABLE 3.

Absolute and relative inter‐rater reliability of Cronometer [additional nutrients].

	Absolute reliability				Relative reliability
	Mean difference L.M.‐O.M. [95% CI]	p value	SEM	MDC	ICC3,1 [95% CI]
Alcohol, g	0.10 [−0.45, 0.64]	0.721	0.22	0.60	0.985 [0.973–0.992]
Caffeine, mg	−1.88 [−7.04, 3.28]	0.466	1.06	2.94	0.996 [0.993–0.998]
Water, g	−9.79 [−115.55, 95.97]	0.853	57.51	159.40	0.972 [0.948–0.984]
Fat
MUFA, g	0.42 [−1.80, 2.64]	0.703	2.30	6.39	0.898 [0.820–0.944]
PUFA, g	−0.04 [−1.17, 1.10]	0.950	1.05	2.92	0.919 [0.856–0.955]
Omega‐3, g	−0.05 [−0.26, 0.16]	0.603	0.18	0.49	0.934 [0.882–0.964]
Omega‐6, g	−0.41 [−1.52, 0.69]	0.451	1.25	3.48	0.877 [0.785–0.931]
SFA, g	−0.07 [−1.24, 1.10]	0.901	0.72	2.00	0.964 [0.934–0.980]
Trans, g	0.04 [−0.18, 0.26]	0.748	0.30	0.83	0.825 [0.699–0.901]
Amino Acids
Cystine, g	0.04 [−0.04, 0.11]	0.315	0.10	0.27	0.846 [0.735–0.914]
Histidine, g	0.07 [−0.11, 0.25]	0.443	0.22	0.60	0.859 [0.755–0.921]
Isoleucine, g	0.14 [−0.14, 0.42]	0.304	0.36	0.98	0.845 [0.733–0.913]
Leucine, g	0.25 [−0.22, 0.71]	0.291	0.59	1.64	0.848 [0.737–0.914]
Lysine, g	0.21 [−0.28, 0.70]	0.387	0.64	1.77	0.837 [0.719–0.908]
Methionine, g	0.06 [−0.09, 0.21]	0.410	0.19	0.54	0.841 [0.725–0.910]
Phenylalanine, g	0.12 [−0.12, 0.36]	0.326	0.29	0.81	0.859 [0.755–0.921]
Threonine, g	0.11 [−0.14, 0.36]	0.364	0.31	0.87	0.849 [0.739–0.915]
Tryptophan, g	0.03 [−0.04, 0.11]	0.359	0.09	0.26	0.840 [0.725–0.910]
Tyrosine, g	0.08 [−0.11, 0.28]	0.395	0.24	0.66	0.861 [0.759–0.922]
Valine, g	0.16 [−0.14, 0.45]	0.289	0.35	0.98	0.861 [0.759–0.922]
Vitamins
Vitamin A, ug REA	77.12[−33.57, 187.80]	0.167	87.36	242.15	0.941 [0.894–0.967]
Vitamin D, IU	16.21 [−22.92, 55.34]	0.408	69.87	193.68	0.698 [0.507–0.824]
Vitamin E, mg	0.80 [−0.63, 2.22]	0.267	2.21	6.12	0.774 [0.621–0.871]
Vitamin K, ug	16.90 [−8.03, 41.83]	0.179	25.87	71.71	0.898 [0.820–0.943]
Vitamin B1, mg	−0.06 [−0.14, 0.02]	0.160	0.08	0.22	0.912 [0.843–0.951]
Vitamin B2, mg	−0.06 [−0.17, 0.05]	0.288	0.10	0.28	0.923 [0.863–0.958]
Vitamin B3, mg	0.41 [−1.14, 1.96]	0.598	1.63	4.53	0.895 [0.814–0.941]
Vitamin B5, mg	0.06 [−0.22, 0.34]	0.662	0.23	0.63	0.939 [0.890–0.966]
Vitamin B6, mg	−0.06 [‐0.23, 0.12]	0.512	0.22	0.60	0.855 [0.748–0.919]
Vitamin B9, ug	−35.66 [−85.92, 14.60]	0.160	88.25	244.63	0.708 [0.522–0.830]
Vitamin B12, ug	−0.17 [−0.94, 0.61]	0.663	1.51	4.18	0.643 [0.426–0.789]
Vitamin C, mg	1.31 [−5.18, 7.79]	0.687	3.89	10.77	0.966 [0.938–0.981]
Minerals
Calcium, mg	20.17 [−18.27, 58.60]	0.296	24.98	69.24	0.960 [0.928–0.978]
Magnesium, mg	−4.28 [−21.62, 13.07]	0.621	14.59	40.43	0.933 [0.880–0.963]
Phosphorous, mg	36.95 [−35.73, 109.62]	0.311	82.82	229.56	0.877 [0.785–0.931]
Potassium, mg	58.59 [−82.67, 199.85]	0.407	144.42	400.32	0.901 [0.826–0.945]
Copper, mg	0.00 [−0.08, 0.07]	0.908	0.07	0.20	0.921 [0.859–0.956]
Iron, mg*	−0.74 [−1.34, −0.13]	0.018	0.59	1.64	0.910 [0.830–0.952]
Manganese, mg	−0.07 [−0.39, 0.26]	0.677	0.35	0.98	0.889 [0.804–0.938]
Selenium, ug	−2.73 [−17.03, 11.57]	0.702	12.47	34.56	0.928 [0.871–0.960]
Zinc, mg	0.03 [−0.74, 0.81	0.929	0.81	2.26	0.896 [0.816–0.942]

Note: Paired samples t‐tests assessed mean differences between raters (O.M. and L.M.); *p‐value < 0.05.

Abbreviations: MDC, minimal detectable change; MUFA, monounsaturated fatty acids; PUFA, polyunsaturated fatty acids; SEM, standard error mean; SFA, saturated fatty acids.

3.3. Validity

Tables 4 and 5 present validity findings for the entire cohort, and for men and women separately, respectively. There were no outliers and the data were normally distributed, as assessed by boxplot and Shapiro–Wilk test (p > 0.05). The assumption of sphericity was violated as assessed by Mauchly's test of sphericity for all nutrients except for protein: total energy (χ ²(2) = 15.80, p < 0.001), carbohydrates (χ ²(2) = 13.899, p < 0.001), fat (χ ²(2) = 8.862, p = 0.012), cholesterol (χ ²(2) = 9.346, p = 0.009), sodium (χ ²(2) = 57.184, p < 0.001), sugar (χ ²(2) = 8.490, p = 0.014), and fibre (χ ²(2) = 25.027, p < 0.001). Therefore, Greenhouse–Geisser corrections were applied for these nutrients. With the total cohort, there were statistical differences among the three software programs for total energy (F(1.515,63.650) = 5.247, p = 0.014, partial η ² = 0.110), protein (F(2,84) = 5.776, p = 0.004, partial η ² = 0.121), cholesterol (F(1.661,69.777) = 19.780, p < 0.001, partial η ² = 0.320), sugar (F(1.685,70.765) = 6.022, p = 0.06, partial η ² = 0.125), and fibre (F(1.373,57.658) = 10.095, p < 0.001, partial η ² = 0.194), and trended towards a difference for carbohydrates (F(1.553,65.242) = 3.358, p = 0.053, partial η ² = 0.070). Importantly, the differences observed for cholesterol and fibre were also clinically meaningful (i.e., large effect size, partial η ² > 0.14). No differences were observed between the three softwares for fat (F(1.675,70.330) = 0.776, p = 0.443, partial η ² = 0.018) or sodium (F(1.141, 47.942) = 0.764, p = 0.403, partial η ² = 0.018).

TABLE 4.

Validity of MyFitnessPal and Cronometer [compared to a reference standard Canadian Nutrient File].

	Analysis software	Mean ± SD	Mean difference CNF‐CRO/MFP [95% CI]	p value	η 2/d
*Total energy, kcal	CNF [ref standard]	2513.4 ± 663.3		0.014	0.110
	MFP	2607.4 ± 679.2	94.0 [−222.9, 411.0]	0.027	0.140
	CRO	2502.4 ± 627.2	−11.0 [−327.9, 306.0]	0.644	0.017
Carbohydrates, g	CNF [ref standard]	272.9 ± 106.7		0.053	0.070
	MFP	285.5 ± 99.1	12.6 [−37.4, 62.6]		0.122
	CRO	273.4 ± 105.1	0.45 [−49.6, 50.5]		0.004
Fat, g	CNF [ref standard]	105.6 ± 37.7		0.443	0.018
	MFP	106.7 ± 37.1	1.0 [−16.8, 18.9]		0.028
	CRO	104.0 ± 35.9	−1.7 [−19.5, 16.2]		0.045
*Protein, g	CNF [ref standard]	119.9 ± 37.1		0.004	0.121
	MFP	113.3 ± 37.3	−6.6 [−24.4, 11.2]	0.027	0.178
	CRO	121.5 ± 36.3	1.6 [−16.2, 19.4]	0.462	0.044
*†Cholesterol, mg	CNF [ref standard]	402.7 ± 207.3		<0.001	0.320
	*MFP	342.4 ± 194.2	−60.3 [−159.5, 39.0)	<0.001	0.300
	CRO	395.2 ± 214.8	−7.5 [−106.7, 91.8]	0.347	0.035
Sodium, mg	CNF [ref standard]	3120.2 ± 1441.0		0.403	0.018
	MFP	3339.1 ± 2151.1	218.9 [−588.6, 1026.4]		0.120
	CRO	3050.6 ± 1303.9	−69.6 [−877.1, 738.0]		0.051
*Sugar, g	CNF [ref standard]	98.7 ± 52.6		0.006	0.125
	*MFP	108.9 ± 54.9	10.2 [−16.2, 36.5]	0.004	0.190
	*CRO	104.8 ± 56.8	6.0 [−20.3, 32.4]	0.010	0.110
*†Fibre, g	CNF [ref standard]	32.1 ± 11.5		<0.001	0.194
	*MFP	34.8 ± 12.3	3.6 [−2.3, 9.5]	0.003	0.295
	*CRO	36.7 ± 12.4	4.6 [−1.3, 10.5]	<0.001	0.385

Note: Data is presented as mean ± standard deviation. Across all three softwares, sugar was defined as the ‘total sugar, g’, and fibre was defined as the ‘total dietary fibre, g’.

One‐way Repeated Measures Analysis of Variance (RM‐ANOVA) assessed mean differences between MyFitnessPal (MFP), Cronometer (CRO), and the reference standard Canadian Nutrient File (CNF).

p‐values beside CNF [ref standard] indicate the RM‐ANOVA result (*p‐value < 0.05) with effect sizes using partial eta‐squared (η ²): values 0.01 (small), 0.06 (medium) and ^† 0.14 (large); p‐values beside MFP and CRO indicate the post hoc Bonferroni‐adjusted comparisons (*p‐value < 0.025) with effect sizes using Cohen's d, values 0.2 (small), ^† 0.5 (medium), ^†† 0.8 (large).

TABLE 5.

Validity of MyFitnessPal and Cronometer [compared to a reference standard Canadian Nutrient File] for Men and Women Separately.

	Analysis software	Men (n = 27)			Women (n = 16)
		Mean ± SD	p value	η 2/d	Mean ± SD	p value	η 2/d
*†Total energy, kcal	CNF [ref standard]	2833.8 ± 567.6	0.189	0.064	1972.7 ± 420.7	0.021	0.254
	MFP	2895.0 ± 609.6		0.104	2122.0 ± 497.3	0.027	0.324
	CRO	2806.6 ± 533.3		0.050	1989.0 ± 401.0	0.660	0.040
*†Carbohydrates, g	CNF [ref standard]	309.2 ± 115.6	0.807	0.005	211.7 ± 48.5	0.001	0.373
	* † MFP	312.9 ± 109.6		0.033	239.2 ± 55.4	0.008	0.529
	CRO	311.2 ± 110.7		0.018	209.5 ± 25.8	0.748	0.044
Fat, g	CNF [ref standard]	147.8 ± 35.3	0.462	0.028	85.0 ± 33.1	0.801	0.007
	MFP	119.3 ± 35.5		0.043	85.3 ± 30.0		0.008
	CRO	116.0 ± 34.7		0.052	83.7 ± 28.7		0.044
*†Protein, g	CNF [ref standard]	138.2 ± 29.7	0.002	0.239	89.0 ± 26.4	0.134	0.130
	*MFP	126.1 ± 32.9	< 0.001	0.387	91.8 ± 34.9		0.088
	CRO	136.1 ± 33.0	0.391	0.068	97.0 ± 28.0		0.292
*†Cholesterol, mg	CNF [ref standard]	444.3 ± 197.0	< 0.001	0.346	332.4 ± 211.6	0.009	0.278
	*MFP	379.3 ± 192.0	< 0.001	0.334	280.2 ± 187.6	0.008	0.261
	CRO	437.8 ± 200.4	0.490	0.033	323.3 ± 225.3	0.541	0.041
Sodium, mg	CNF [ref standard]	3438.7 ± 1418.8	0.460	0.023	2583.7 ± 1354.1	0.77	0.013
	MFP	2744.7 ± 2383.2		0.156	2654.6 ± 1521.8		0.050
	CRO	3352.6 ± 1338.9		0.062	2541.1 ± 1100.9		0.034
*†Sugar, g	CNF [ref standard]	114.6 ± 58.0	0.112	0.082	71.9 ± 26.1	0.027	0.216
	* † MFP	121.5 ± 59.7		0.117	87.6 ± 37.3	0.017	0.488
	*CRO	121.1 ± 63.6		0.107	77.1 ± 27.1	0.060	0.195
*†Fibre, g	CNF [ref standard]	36.2 ± 11.5	0.008	0.201	25.1 ± 7.7	0.002	0.430
	* † MFP	37.6 ± 12.8	0.301	0.118	32.3 ± 12.3	< 0.001	0.701
	* † CRO	41.0 ± 12.6	<0.0001	0.395	29.4 ± 7.8	< 0.001	0.556

Note: is presented as mean ± standard deviation. Across all three softwares, sugar was defined as the ‘total sugar, g’, and fibre was defined as the ‘total dietary fibre, g’.

One‐way Repeated Measures Analysis of Variance (RM‐ANOVA) assessed mean differences between MyFitnessPal (MFP), Cronometer (CRO), and the reference standard Canadian Nutrient File (CNF) for men (n = 27) and women (n = 16) separately.

p‐values beside CNF [ref standard] indicate the RM‐ANOVA result (*p‐value < 0.05) with effect sizes using partial eta‐squared (η ²): values 0.01 (small), 0.06 (medium) and 0.14 (large); p‐values beside MFP and CRO indicate the post hoc Bonferroni‐adjusted comparisons (*p‐value < 0.025) with effect sizes using Cohen's d, values 0.2 (small), ^† 0.5 (medium), ^†† > 0.8 (large).

Post hoc analyses with a Bonferroni‐adjustment revealed that MFP was different from CNF for cholesterol (p < 0.001, d = 0.300), sugar (p = 0.004, d = 0.190), and fibre (p = 0.003, d = 0.295), and trended towards being different for total energy (p = 0.027, d = 0.140) and protein (p = 0.027, d = 0.178). The differences observed for cholesterol and fibre were also clinically meaningful (i.e., small to medium effect sizes). It appears that differences in total energy, carbohydrates, and sugar were driven by women, and differences in protein were driven by men (Table 5).

Post hoc analyses revealed that CRO was different from CNF for sugar (p = 0.010, d = 0.110) and fibre (p < 0.001, d = 0.385), with a clinically meaningful difference for fibre (i.e., small to medium effect size). Additional analyses [for nutrients provided by CRO] showed statistical and clinically meaningful differences between CRO and the reference standard CNF for vitamins A (p = 0.013, d = 0.208) and D (p = 0.013, d = 0.318) (Table 6).

TABLE 6.

Validity of Cronometer for additional nutrients [compared to a reference standard Canadian Nutrient File].

	Analysis software	Mean ± SD	Mean difference CNF‐CRO [95% CI]	p value	d
*Vitamin A, ug REA	CNF [ref standard]	1309.9 ± 927.5	−209.8 [−372.6, −47.0]	0.013	0.208
	CRO	1519.7 ± 1086.2
*Vitamin D, IU	CNF [ref standard]	178.1 ± 136.5	−51.4 [−91.6, −11.2]	0.013	0.318
	CRO	229.6 ± 183.8
Vitamin B3, mg	CNF [ref standard]	27.9 ± 9.1	−1.0 [−2.8, 0.8]	0.278	0.098
	CRO	28.8 ± 10.8
Vitamin B6, mg	CNF [ref standard]	2.3 ± 0.9	−0.1 [−0.2, 0.0]	0.150	0.090
	CRO	2.4 ± 0.9
Vitamin B12, ug	CNF [ref standard]	5.1 ± 2.3	0.3 [−0.4, 1.0]	0.353	0.125
	CRO	4.8 ± 2.7
Calcium, mg	CNF [ref standard]	1162.8 ± 459.2	−18.0 [−81.0, 45.0]	0.567	0.040
	CRO	1180.0 ± 434.0
Potassium, mg	CNF [ref standard]	3918.3 ± 1166.9	206.8 [14.7, 398.9]	0.036	0.193
	CRO	3711.5 ± 967.7
Iron, mg	CNF [ref standard]	17.4 ± 5.3	−0.3 [−1.2, 0.6]	0.506	0.059
	CRO	17.7 ± 4.7
Magnesium, mg	CNF [ref standard]	471.2 ± 172.2	17.3 [−14.6, 49.2]	0.280	0.105
	CRO	453.9 ± 157.0
Selenium, ug	CNF [ref standard]	151.4 ± 96.0	0.9 [−14.8, 16.7]	0.906	0.009
	CRO	150.4 ± 114.7

Note: Data is presented as mean ± standard deviation.

Paired samples t‐tests assessed mean differences between Cronometer (CRO) and the reference standard Canadian Nutrient File (CNF); *p‐value < 0.025 with effect sizes using Cohen's d, values 0.2 (small), ^† 0.5 (medium), ^†† 0.8 (large).

Figures 3 and 4 show Bland–Altman plots of validity for MFP (left graphs) and CRO (right graphs), compared to CNF [ref standard]. Wide variability and poor data spread with a tendency to overestimate total energy (kcal), carbohydrates (g), sodium (mg), sugar (g), and fibre (g), and underestimate protein (g) and cholesterol (mg) can be seen when using MFP. Poor data spread with a tendency to overestimate sugar (g) can be seen when using CRO, as well as a tendency to overestimate fibre (g).

FIGURE 3.

Bland–Altman plots of MyFitnessPal (left) and Cronometer (right) validity, compared to the reference standard Canadian Nutrient File. Bias between the two softwares (long dashed lines) as well as 95% limits of agreement (dotted lines; UL, upper limit, LL, lower limit) are presented for total energy (kcal), carbohydrates (g), fat (g), and protein (g).

FIGURE 4.

Bland–Altman plots of MyFitnessPal (left) and Cronometer (right) validity, compared to the reference standard Canadian Nutrient File. Bias between the two softwares (long dashed lines) as well as 95% limits of agreement (dotted lines; UL, upper limit, LL, lower limit) are presented for cholesterol (mg), sodium (mg), sugar (g), and fibre (g).

4. Discussion

Overall findings from the present study indicate that (1): MFP has low inter‐rater reliability and validity for most nutrients, and differences exist between men and women; and (2) CRO offers excellent inter‐rater reliability for all nutrients except for iron, and good validity for all nutrients except for fibre and vitamins A and D, with no differences between genders.

4.1. Reliability

When using MFP, the present study observed low inter‐rater reliability for many of the nutrients. While differences were found between raters for total energy, carbohydrates, protein, and sugar, particularly among women, the differences were too small to be considered practically or clinically important. These small absolute differences observed could be due to how the raters interpret food descriptions, estimate portion sizes, or navigate the app's food database. On the other hand, poor relative reliability (ICC_3,1 < 0.5) for sodium and sugar, particularly among men, indicates raters are inconsistent when inputting these nutrient values. This is particularly problematic since sodium and sugar are often ‘hidden' in foods and their values depend on the specific brand or recipe, which the rater may handle inconsistently [30]. Of note, observed gender differences in inter‐rater reliability are not due to inherent differences in participants, but rather how raters used the dietary software. The better ICC_3,1 for women compared to men [for sodium and sugar] may be due to women providing more descriptive dietary data, resulting in raters choosing more similar items in MFP. That is, higher social desirability (i.e., tendency to avoid criticism) among women [31] may contribute to a more detailed and descriptive FIR.

In contrast to the low inter‐rater reliability observed when using MFP, excellent absolute and relative inter‐rater reliability for the same eight nutrients that the MFP software provides was observed for CRO. It is likely that the inferior inter‐rater reliability performance of MFP compared to CRO is related to its abundant user‐uploaded options for each food entry. That is, the MFP database contains copious non‐verified consumer entries; for participant records where an exact food item was not specified in enough detail or found through an online search [thus necessitating the use of a close alternative], there was an increased opportunity that the two raters would select different food options. Suggestions to improve inter‐rater reliability of MFP include calibrating raters to reduce systemic bias. For example, improve standardisation of how raters interpret and input data to minimise systemic bias, including training and consensus sessions, creating a shared glossary or food mapping reference (e.g., how to interpret cooking methods [baked, fried, raw], when to default to high‐fat vs. low‐fat variant, standardise brand selection rules), and periodically double‐checking entries.

4.2. Validity

Importantly, reliability is a prerequisite for validity [32]. Therefore, the low inter‐rater reliability of MFP for total energy, carbohydrates, fat, sodium, and sugar prevents it from providing valid measures of these nutrients when different raters are inputting food items. Findings from the present study regarding validity highlight issues with both MPF and CRO, especially for sugar and fibre, i.e., statistical and/or clinically meaningful differences observed between MFP/CRO and the reference standard CNF. While the reference standard itself may have limitations, these findings indicate that users relying solely on dietary assessment apps may be getting inaccurate nutritional information for certain nutrients. Clinically meaningful differences between MFP and CNF were observed for cholesterol and fibre, and between CRO and CNF for fibre [as well as for vitamins A and D]; the small to medium effect sizes suggest the differences are large enough to be practically important for the individual's health. Visual inspection of the Bland Altman plots supports these findings.

Previous work has reported similar poor validity of MFP against reference databases (e.g., Nutrition Data System for Research [NDSR], USDA) [10, 12, 13, 14, 15, 16, 17, 33, 34, 35]. A recent systematic review assessing 14 validity studies of 12 dietary assessment apps reported MFP to have the highest mean difference for energy, fat, and protein [9]. Unlike the present findings, many of these previous studies reported an underestimation of nutrients when using MFP. It should be noted that methodologies of these validation studies differ from the present study; for example, various reference databases and different comparator methods of dietary assessment (24 h recall, food frequency questionnaire) were used, and participants logged their own food items into MFP (i.e., higher likelihood of selecting non‐verified sources) [9]. The low validity of MFP in the present study may be due to its user‐upload options. While raters prioritised selection of verified foods (i.e., the green check mark), these verified foods may still contain information that is incomplete and/or incorrect [23]. Further, it has been observed that higher discrepancies between MFP and other assessment tools are more prevalent among participants with higher caloric intake [12]. According to the 2015 Canadian Community Health Survey, the average total energy intake for Canadian adults (31–70 years old) is 1861 kcal/day [36]; the endurance athletes in the present cohort consumed an average total intake of >2500 kcal per day, possibly contributing to the overestimation observed for some nutrients (i.e., total kcal, carbohydrates, sodium, and sugar).

Interestingly, gender differences emerged when assessing the validity of the MFP app; poor validity was found among women for total energy, carbohydrates, and sugar, and poor validity was found among men for protein. Possible rationale may include the generally more varied dietary patterns in women compared to men. That is, a recent study from Italy including 2198 participants reported men tend to consume higher rates of red and processed meat while women tend to consume more vegetables, whole grains, tofu, and high‐cocoa content chocolate [37].

CRO's statistical and clinically meaningful differences were limited to fibre, vitamin A, and vitamin D, suggesting the CRO database may be more closely aligned with CNF than MFP for most nutrients. Of note, raters documented overall usability of MFP and CRO while inputting food items: verified items in MFP (i.e., those with a green check mark) were not necessarily sorted to the top of a search necessitating more time spent looking for the verified item, whereas CRO sorted items from verified databases to the top of the search (e.g., CNF, NCCDB, etc.). In addition, although both applications allowed manual input of an item [if it was not found in the database], CRO enabled input of more micronutrients compared to MFP, likely contributing to a more accurate output overall. Possible rationale for low validity of fibre likely includes how this nutrient is represented in the software; that is, CRO provides ‘fibre, g’ only whereas CNF provides data on ‘total dietary fibre, g’, ‘total soluble fibre, g’, ‘dietary fibre (2016), g’, and ‘soluble fibre (2016), g’ [*note: in the present study CNF ‘total dietary fibre, g’ was used for comparison]. Further, fibre values can be highly dependent on the food processing method (e.g., whole vs. refined grains), which the app may handle inconsistently. Of note, visual inspection of the Bland‐Altman plots showed reasonable LOAs and good horizonal spread of the data for fibre. Low validity for vitamins A and D is likely due to fortification practices differing between brands and countries [11], as vitamins A and D are commonly added to a wide range of foods (e.g., cereals, breads, pastas, etc.) [38].

4.3. Strengths and Limitations

This study has several strengths, including the fact that it assessed both inter‐rater reliability and the validity of two popular free nutrition tracking applications. Previous research, particularly that which evaluated MFP, only considered validity. It is important to note that while a valid measure must be reliable, the reverse isn't true; a measure can be reliable without being valid [32]. Another strength of the present study was the collection of three nonconsecutive days of food intake in a free‐living setting, allowing for a more representative assessment of participants' true intake [39]. A limitation could be the reference standard itself; while the CNF is a robust and widely used reference tool, it is infrequency updated and in some cases outdated, as well as in some cases provides generic average nutrient values instead of brand‐specific data [40, 41]. Another limitation was the lack of randomisation of entry order across softwares; however, raters were trained and calibrated using a shared standard operating procedure to minimise a learning curve. Lastly, the lack of an objective measure for usability of the applications can be cited as a limitation. Previous studies have used the System Usability Scale [2, 12, 13] or Mobile App Rating Scale [42, 43] to measure usability of different applications; in the present study, a usability score may have provided additional insight into rationale for reliability and validity findings.

5. Conclusion

To the author's knowledge, this is the first study assessing inter‐rater reliability and validity [comparing to the reference standard CNF database] of popular dietary tracking applications MFP and CRO, in a Canadian endurance athlete population. MFP provided data on eight nutrients and revealed low reliability and/or validity for all nutrients, with differences between men and women. Rationale for low reliability and validity may be due to the copious options for each food entry, including non‐verified consumer entries. Possible rationale for gender differences may be more detailed and descriptive FIRs (reliability) and generally more varied dietary patterns (validity) in women compared to men. CRO is a free Canadian‐based nutrition tracking application that provides data for up to 84 nutrients; excellent reliability for all nutrients except for iron, and good validity for all nutrients except for fibre and vitamins A and D was found in the present study, with no differences between genders. Rationale for the low validity of fibre may be due to how it is represented in the software (i.e., total vs. soluble), and rationale for vitamins A and D may be due to fortification/enrichment practices differing between brands and countries. Given the unique energy and nutrient needs of athletes, they [and their health care providers] should be aware of the potential for MFP to provide dietary information that is not representative of true consumption, and that CRO is a promising alternative. Although the current research was focused on an athletic population, future research should evaluate the use of MFP and CRO [and other nutrient assessment applications] from a healthcare/medical perspective.

Author Contributions

Olivia Morello and Lucas McPhee: data curation (supporting), formal analysis (leads), writing – review and editing (supporting). Michaela Kucab: formal analysis (supporting), writing – review and editing (supporting). Nick Bellissimo: supervision (supporting), writing – review and editing (supporting). Julia O. Totosy de Zepetnek: conceptualisation (lead), data curation (lead), formal analysis (supporting), methodology (supporting), project administration (lead), resources (lead), supervision (lead), visualisation (lead), writing – original draft (lead).

Ethics Statement

The study procedures were approved by the Research Ethics Board at the University of Regina (REB #2019‐212).

Conflicts of Interest

The authors declare no conflicts of interest.

Peer Review

The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1111/jhn.70148.

Morello O., McPhee L., Kucab M., Bellissimo N., and Totosy de Zepetnek J. O., “Reliability and Validity of Nutrient Assessment Applications for Canadian Endurance Athletes: MyFitnessPal and Cronometer,” Journal of Human Nutrition and Dietetics 38 (2025): 1‐16, 10.1111/jhn.70148.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.