Evaluation of the Chrononutrition Profile – Questionnaire in an online community sample of adults

Abstract

Chrononutrition (i.e., circadian timing of food intake) has been linked to indicators of health status such as body weight and insulin resistance. A measure of general chrononutrition patterns, the Chrononutrition Profile – Questionnaire, has been developed and preliminary evidence of validity and reliability of the measure has been documented in a homogenous group of undergraduates. However, this measure has not yet been validated in an online, community-based sample. The present study therefore aimed to evaluate the validity of the Chrononutrition Profile – Questionnaire in a web-based community sample. Analyses suggested that the Chrononutrition Profile – Questionnaire displays acceptable validity for use in diverse community samples of adults, with moderate to strong correlations (r = 0.39–0.91) between the Chrononutrition Profile – Questionnaire and measures of dietary intake and sleep. This measure is suitable for use in a variety of settings, by stakeholders and scientists, and may contribute to future development of health behavior interventions and research programs centered around chrononutrition.

1. Introduction

Chrononutrition, defined as the circadian timing of food intake (Arble et al., 2009), is significant to public health due to its associations with overweight and obesity (Arble et al., 2009), which are common (Hales et al., 2020) and costly health conditions in the United States (Lega & Lipscombe, 2020). In particular, a foundational rodent study by Arble et al. (2009) showed that eating at the “wrong” time of day (i.e., during the typical resting period) led to greater weight gain than did eating at the “right” time of day, even when caloric intake was equal across study conditions; research has begun to examine these relationships in humans (Allison et al., 2021; Gill & Panda, 2015). Chrononutrition may also be important to consider because of its ties to other markers of health. For instance, unhealthy chrononutrition has been linked to health indicators such as waist circumference (Aqeel et al., 2020; Jakubowicz et al., 2013; Kahleova et al., 2017), impaired blood glucose tolerance (Lopez-Minguez et al., 2018), inflammation (Chaix et al., 2014), increased blood pressure (Wilkinson et al., 2020). The majority of research on chrononutrition and its health impacts has consisted of experimental work, such as prescribed eating windows (Allison et al., 2021; Chaix et al., 2014; Gill & Panda, 2015), and until recently, no comprehensive, valid and reliable measure of naturalistic chrononutrition patterns existed.

The Chrononutrition Profile – Questionnaire (CP-Q; Veronda et al., 2020) was developed to serve as a questionnaire that captures general chrononutrition patterns. Veronda et al. (2020) demonstrated preliminary evidence of the validity and reliability of the CP-Q in a sample of undergraduate university students. The CP-Q improves upon existing meal timing assessment methods by providing a comprehensive assessment of chrononutrition patterns. Further, the CP-Q is brief and takes only a few moments to administer and complete. Thus, this questionnaire invokes less participant and administrator burden compared to other meal timing assessments.

A primary limitation of the initial study by Veronda et al. (2020) was conducted in a sample of college students; the CP-Q has not yet been validated in a diverse community sample of adults. Evidence of validity of this measure in an adult, community-based sample would provide researchers with a brief but comprehensive method of assessing chrononutrition that can be widely used in community samples. Therefore, the primary aim of the present study was to evaluate the validity of four primary chrononutrition behaviors derived from the CP-Q in a community-based sample of adults. We hypothesized that these behaviors measured by the CP-Q would be strongly and positively associated with the corresponding values as assessed by the criterion measure.

2. Method

2.1. Participants and procedure

This study represents one aim of a larger online longitudinal study examining chrononutrition, sleep, body weight, and other health behaviors over time. All study procedures and materials were approved by the North Dakota State University Institutional Review Board.

2.1.1. Recruitment and screening

Participants were recruited and screened for eligibility through Prolific (www.prolific.co), a web-based platform created for online survey research (Palan & Schitter, 2018). Individuals were eligible for the present study if they read and spoke English; were current residents of the United States; were 18 to 65 years of age; did not report a diagnosis of any chronic diseases such as diabetes, heart disease, stroke, etc.; and did not report shift work. These criteria were selected to reduce potential confounding age- and lifestyle-related factors. Individuals who were eligible to participate received an invitation to enroll in the study via Prolific and could then sign up for the study on Prolific. All participants completed the informed consent process online via Prolific prior to enrolling. Enrolled participants (N = 258) completed a series of questionnaires online. Data collection occurred in November 2020. Participants received monetary compensation upon the conclusion of the survey session.

Of the 258 enrolled participants, 13 (5.04%) were excluded from analyses due to excessive missing data (e.g., not completing any of the food intake log). Further, 48 of the 258 enrolled participants (18.60%) were excluded from analyses due to reporting an eating window of 0 min (i.e., reporting only a single eating event per day). The CP-Q is neither designed nor validated to assess the chrononutrition of individuals engaging in one eating event per day, and it is difficult to ascertain if these are genuine responses or measurement error (e.g., boredom, fatigue). The study team made significant efforts to avoid potential biases in characteristics of the sample, including recruiting approximately equivalent numbers of men and women. Additionally, the survey was released at two different times of day (i.e., morning and evening) to minimize potential time of day effects and to avoid favoring recruitment of one chronotype over the other. Attrition analyses revealed that participants with an eating window of 0 min did not differ in age, gender, annual household income, race, education level, or employment status from those with an eating window of greater than 0 min. This resulted in a final sample size of 197 participants.

2.2. Measures

The following measures were used in the present study.

2.2.1. Chrononutrition Profile – Questionnaire

The Chrononutrition Profile – Questionnaire (CP-Q) (Veronda et al., 2020) consists of 18 items which are designed to measure key chrononutrition behaviors on both work/school days and free days, as sleep/wake and feeding/fasting patterns can vary greatly between free days and work/school days (Veronda et al., 2020; Wittmann et al., 2006). The present study examined evening latency, evening eating, eating window, and eating midpoint. For detailed discussion of each chrononutrition behavior, see Veronda et al. (2020). Briefly, “evening latency” is the duration of time in minutes between the last eating event of the day and sleep onset. “Evening eating” is defined as the time (represented as clock time, i.e., HH:MM) of one’s last eating event before bed. “Eating window” refers to the duration of time (in minutes) between the first eating event of the day and the last eating event of the day. Another value that can be derived from the CP-Q is the “eating midpoint”, which represents the clock time (HH:MM) halfway between the first and last eating events. Although empirical guidelines for healthy and unhealthy chrononutrition behaviors have not yet been established, research suggests that both longer eating windows and eating later in the day (i.e., later eating midpoint, later evening eating) are associated with markers of poor health and increased body weight (for review, see Veronda et al., 2021). Evidence of test-retest reliability and validity of the CP-Q has been reported (Veronda et al., 2020). Specifically, convergent validity of the CP-Q has been shown to be acceptable, and especially strong for evening eating and eating window, with significant and moderate correlations between the CP-Q and the criterion measure, the ASA24 (Subar et al., 2012), i.e., Automated Self-Administered 24-hour dietary recall tool. In addition, evidence of moderately strong test-retest reliability over a two-week time frame was reported.

2.2.2. Pittsburgh Sleep Quality Index

The Pittsburgh Sleep Quality Index (PSQI) (Buysse et al., 1989) is a 19-item measure designed to assess seven components of sleep quality complaints over the past month, including subjective sleep quality and sleep disturbances. The present study utilized bedtime and wake time as reported in the PSQI. This measure has displayed evidence of internal consistency reliability and construct validity, with Cronbach’s alphas of 0.80 for its seven components and moderate to strong correlations with criterion measures, as described by Carpenter and Andrykowski (1998). For the seven PSQI components, Cronbach’s alpha was 0.74 in this study.

2.2.3. Food intake log

A food intake log was created to provide additional information regarding participants’ dietary intake patterns. Participants were asked to report the times at which they eat or drink something on a typical weekend day or free day, and on a typical work day or school day. They were also asked to label each item as either a drink, a snack, breakfast, lunch, or dinner/supper. This food log was modeled after the ASA24 (Subar et al., 2012), such that questions were phrased as they are provided in the ASA24. We chose not to utilize the full ASA24 as this tool can be quite burdensome for participants and may take as long as 30 min to complete; thus in contrast to the ASA24 and to reduce this participant burden, participants were not asked questions regarding what items they consumed and portion size of those items. The food intake log used in the present study can be viewed as a supplement.

2.2.4. Body mass index

Participants were asked to report their height in feet and inches and their weight in pounds to allow for calculation of body mass index (BMI). BMI was calculated as kg/m². In addition, BMI was categorized into ranges based on Centers for Disease Control recommendations (Defining Adult Overweight and Obesity). A BMI of less than 18.5 was categorized as underweight, BMI from 18.5 to <25 was categorized as healthy weight, BMI from 25.0 to <30 was categorized as overweight, and BMI of 30.0 or higher was categorized as obese.

2.2.5. Rapid eating assessment for patients

The Rapid Eating Assessment for Patients (REAP) (Gans et al., 2003) consists of 31 items designed to evaluate dietary behaviors. Twenty-seven REAP items are designed to assess diet quality. Example items are: “In an average week, how often do you eat 4 or more meals from sit-down or take out restaurants” and “In an average week, how often do you eat beef, pork, or dark meat chicken more than 2 times a week?”. The measure also includes items to capture dietary behaviors (e.g., “Do you usually shop and prepare your own food?”). Values of the 27 diet quality items are summed to compute a total score, which may range from 27 to 81. Higher scores on the measure are indicative of a healthier-quality diet. In the present study, Cronbach’s alpha was 0.79.

2.2.6. Composite scale of morningness

The 13-item composite scale of morningness (CSM; Smith et al., 1989) was used to examine participants’ chronotype (i.e., sleep/wake timing preference). Example CSM items include: “Assuming normal circumstances, how easy do you find getting up in the morning” and “At what time in the evening do you feel tired and as a result, in need of sleep?”, and response options for each item range from 1 (indicating extreme evening chronotype) to either 4 or 5 (indicating extreme morning chronotype). Scores for all items are summed to compute a total score. Total scores may range from 13 to 55, and higher total scores reflect a greater tendency toward a morning chronotype (Smith et al., 1989). Specifically, a score of 22 or below is indicative of an evening chronotype, a score of 23 to 44 indicates an intermediate chronotype, and a score of above 44 indicates a morning chronotype. In this study, Cronbach’s alpha was 0.92.

Participants were also asked to report health and sociodemographic information (e.g., age, gender, race) to allow for assessment of potential covariates.

3. Data analysis

3.1. Sample size determination

While the present study was one aim of a larger longitudinal study, our sample size determination (recruiting n = 258 at Time 1) was based on statistical power estimates for another study aim centered around evaluating a mediation model, and to allow for attrition as part of the parent longitudinal study. Our sample size was also adequate for the present study’s planned statistical tests based on guidelines provided by Cohen (1992), suggesting a minimum sample size of n = 85 for a bivariate correlational analysis with minimum power = 0.80, alpha = 0.05, and a medium effect size (r = 0.30).

3.2. Initial examination of data

For this study, we first aimed to explore the characteristics of individual items of the CP-Q using descriptive statistics. We examined ranges of reported items, along with means, standard deviations, skewness and kurtosis values. Our examination of CP-Q items also involved supplementary, exploratory analyses to examine associations between chrononutrition behaviors and demographic characteristics and BMI in this community-based sample.

3.3. Convergent validity

We also aimed to assess the convergent validity of four key chrononutrition behaviors derived from the CP-Q in our community-based sample. Fig. 1 provides a list of each CP-Q value and the corresponding PSQI and food intake log items that were paired to assess convergent validity. First, to explore whether there were any systematic biases in the CP-Q compared to the criterion measures, paired samples t-tests and effect size estimates (i.e., Cohen’s d) were conducted for the pairs of variables listed in Fig. 1. As described by Cohen (1988), effect sizes for Cohen’s d may be interpreted as follows: d = 0.2 (small), d = 0.5 (medium), and d = 0.8 (large).

Fig. 1.

Chrononutrition Profile-Questionnaire (CP-Q), Pittsburgh Sleep Quality Index (PSQI), and food intake log items used for validity estimates.

Next, Pearson product-moment coefficients were calculated between those key chrononutrition behaviors and corresponding items from the PSQI and food intake log. Pearson product-moment coefficients allowed us to examine the strength and direction of the relationships, with high correlation coefficients indicating stronger convergent validity between the CP-Q and established measures. Lastly, to descriptively evaluate the magnitude of similarity between established measures and the CP-Q we estimated the discrepancy between values reported in the CP-Q and in the established measures. To examine the degree of discrepancy between the two measures, 60 min was used as a predetermined cutoff value based on past research (Veronda et al., 2020).

4. Results

4.1. Initial examination of data

This sample was 50.8% female and 74.9% white, with ages ranging from 18 to 65 years (M = 33.57, SD = 8.86). Table 1 details participants’ demographic information.

Table 1.

Sample demographics (N = 197).

Demographic variable	N (%)
Gender
Male	96 (48.7)
Female	100 (50.8)
Prefer not to say	1 (0.5)
Race
American Indian or Alaskan Native	1 (1.0)
Asian or Pacific Islander	31 (15.7)
Black or African American	9 (4.6)
White or Caucasian	146 (74.1)
Other/mixed	7 (3.6)
Did not disclose	2 (1.0)
Marital status
Single, never married	80 (40.6)
Married or domestic partnership	108 (54.8)
Widowed	1 (0.5)
Divorced	8 (4.1)
Annual household income
Less than $25,000/year	10 (5.1)
$25,000–$49,999/year	33 (16.8)
$50,000–$74,999/year	38 (9.3)
$75,000–$99,999/year	43 (21.8)
$100,000–$124,999/year	26 (13.2)
$125,000–$149,999/year	24 (12.2)
$150,000–$174,999/year	9 (4.6)
$175,000–$199,999/year	2 (1.0)
$200,000/year or more	10 (5.1)
Prefer not to say	2 (1.0)
Highest educational level
High school degree or equivalent (e.g., GED)	10 (5.1)
Some college but no degree	16 (8.1)
Associate degree	14 (7.1)
Bachelor’s degree	101 (51.3)
Graduate degree	56 (28.4)
Employment status
Employed, part time	29 (14.7)
Employed, full time	160 (81.2)
Not employed, looking for work	4 (2.0)
Student	4 (2.0)
Composite scale of morningness
Evening type	18 (9.1)
Intermediate type	146 (74.1)
Morning type	27 (13.7)
Missing data	6 (3.0)

Descriptive statistics were used to examine characteristics of the CP-Q. Initial descriptive analyses indicated that responses provided in the CP-Q were normally distributed based on acceptable skewness and kurtosis. Further, a broad range of values were reported for each item with a low frequency of missing or impossible values. Descriptive statistics of these items are shown in Table 2. Additional exploratory analyses indicated that CP-Q-assessed chrononutrition behaviors did not differ by gender, race, or annual household income (all ps > .05). Correlational analyses did suggest that chrononutrition may vary by age. Age was negatively correlated with workday first eating event (r = −0.24, p < .001), free day first eating event (r = −0.17, p = .019), and free day last eating event (r = −0.14, p < .048), suggesting that an older age was related to earlier timing of eating events. Age was not significantly related to any other chrononutrition behaviors.

Table 2.

Descriptive statistics of BMI, REAP, and CP-Q-, food intake log-, and PSQI-assessed chrononutrition behaviors (n = 197).

Item	Range	Mean	SD	Skewness	Kurtosis
BMI	16.94–52.48	25.43	5.24	1.84	5.36
REAP	33–73	54.66	7.38	−0.24	−0.07
CP-Q
Wake time – WKDY	4:00–11:00	6:52	1:11	0.24	0.53
Wake time – FREE	5:00–12:00	8:22	1:35	0.24	−0.51
First eating event – WKDY	4:55–18:00	9:05	2:07	0.69	0.42
First eating event – FREE	5:30–16:00	10:06	1:57	0.29	−0.40
Evening eating – WKDY	15:00–1:00	19:53	1:33	0.35	0.30
Evening eating – FREE	14:00–1:00	20:10	1:44	0.18	0.27
Evening latency – WKDY	30.00–540.00	193.15	84.24	0.49	0.61
Evening latency – FREE	30.00–660.00	222.78	100.90	0.73	1.57
Eating window – WKDY	210.00–1020.00	646.92	142.30	−0.32	−0.20
Eating window – FREE	240.00–900.00	606.20	126.35	−0.20	−0.33
Eating midpoint – WKDY	11:15–19:45	14:28	1:26	0.69	0.95
Eating midpoint – FREE	11:15–19:30	15:09	1:30	0.43	0.04
Bedtime – WKDY	20:00–2:30	23:05	1:07	0.19	0.36
Bedtime_FREE	20:00–4:00	23:53	1:25	0.15	−0.19
Food intake log
First eating event - WKDY	4:55–18:30	9:18	2:21	0.89	0.90
First eating event - FREE	5:00–16:00	9:55	2:04	0.41	−0.02
Evening eating - WKDY	12:00–1:00	19:37	1:48	−0.20	1.74
Evening eating - FREE	15:00–2:00	20:02	1:50	0.25	0.25
Evening latency - WKDY	30.00–720.00	211.11	104.00	1.19	3.51
Evening latency - FREE	0.00–540.00	187.98	99.00	0.56	0.36
Eating window - WKDY	30.00–1020.00	619.48	165.08	−0.73	1.09
Eating window - FREE	60.00–1020.00	606.40	148.01	−0.41	0.69
Eating midpoint - WKDY	11:00–20:00	14:27	1:36	0.66	0.54
Eating midpoint - FREE	11:15–19:00	14:59	1:31	0.39	0.09
PSQI
Wake time	4:00–12:00	7:02	1:20	0.47	0.92
Bedtime	20:00–3:00	23:08	1:14	0.40	0.34

Note. BMI = body mass index; REAP = Rapid Eating Assessment for Patients; CP-Q = Chrononutrition Profile – Questionnaire; FREE = weekend/free day; WKDY = work/school day; PSQI = Pittsburgh Sleep Quality Index.

Note. First eating event and evening eating represented as HH:MM, evening latency and eating window expressed as minutes, and breakfast skipping expressed as percentage of days per week.

Note. Due to small number of missing data points, all available data points were used in analyses. Sample sizes for each analysis ranged from N = 186 to N = 197.

We also chose to explore descriptive statistics of participants’ chronotype. Scores on the CSM ranged from 14 to 55 (M = 35.22, SD = 8.61). We discovered that the majority (74.1%) of the sample (n = 146 out of 197) was classified as having an intermediate chronotype (Table 1).

Next, because prior literature suggests a relationship between chrononutrition and body weight, we chose to explore relationships between self-reported BMI and CP-Q-assessed chrononutrition behaviors. Five participants (2.5%) were categorized as underweight, 101 participants (51.3%) were categorized as healthy weight, 57 participants were categorized as overweight (28.9%), 27 participants were categorized as having obesity (13.7%), and 7 participants (3.6%) did not disclose their height and weight. One-way between-subjects ANOVAs were conducted to examine mean differences in CP-Q assessed chrononutrition behaviors between BMI categories. Analyses revealed significant differences in chrononutrition behaviors for free day eating window [F(3, 185) = 3.63, p = .014, η² = 0.056] and free day evening latency [F(3, 185) = 2.94, p = .035, η² = 0.045]. For all other CP-Q assessed chrononutrition behaviors, all ps > .05. Table 3 displays means and standard deviations for CP-Q-assessed chrononutrition behaviors by BMI category. In addition, to further explore the relationship between chrononutrition and BMI, we conducted these analyses again while controlling for REAP scores (i.e., diet quality). Analyses of covariance (ANCOVAs) were used to evaluate the relationship between CPQ-assessed chrononutrition behaviors and BMI, after controlling for diet quality. An ANCOVA revealed that the significant difference in free day eating window remained after controlling for diet quality [F(3, 173) = 3.18, p = .025, η² = 0.052], but no other statistically significant relationships were found (ps > .05).

Table 3.

Means and standard deviations of CP-Q-assessed chrononutrition behaviors by BMI category.

Chrononutrition behavior	M (SD)
	Underweight BMI	Healthy weight BMI	Overweight BMI	Obese BMI
	N = 5	N = 101	N = 57	N = 27
Wake time
Work/school day	6:40 (1:39)	6:57 (1:04)	6:53 (1:19)	6:46 (1:17)
First eating event	9:30 (2:26)	8:52 (2:01)	9:10 (1:56)	9:05 (2:08)
Evening eating	21:06 (1:30)	19:59 (1:38)	19:50 (1:26)	19:30 (1:37)
Evening latency	120.00 (21.21)	184.65 (79.56)	206.05 (82.28)	205.00 (105.82)
Eating window	696.00 (109.00)	665.13 (134.47)	628.39 (155.91)	619.26 (149.28)
Eating midpoint	15:18 (1:49)	14:26 (1:27)	14:34 (1:28)	14:20 (1:16)
Bedtime	23:06 (1:25)	23:03 (1:02)	23:16 (1:12)	22:55 (1:07)
Free day
Wake time	8:18 (1:47)	8:28 (1:37)	8:23 (1:33)	8:22 (1:29)
First eating event	10:18 (2:56)	9:53 (1:58)	10:34 (1:59)	10:09 (1:39)
Evening eating	21:12 (1:31)	20:19 (1:53)	20:11 (1:37)	19:31 (1:30)
Evening latency	138.00 (75.30)	211.49 (95.16)	237.86 (107.50)	253.89 (103.16)
Eating window	654.00 (113.05)	629.75 (123.10)	576.75 (131.29)	561.67 (117.84)
Eating midpoint	15:45 (2:09)	15:08 (1:35)	15:22 (1:26)	14:50 (1:14)
Bedtime	23:30 (1:52)	23:50 (1:20)	00:09 (1:32)	23:45 (1:21)

Note. CP-Q = Chrononutrition Profile – Questionnaire; BMI = body mass index.

Note. Analyses were based on all available data due to missing items; actual sample sizes ranged from N = 188 to N = 190.

Note. 7 participants did not disclose their height and weight.

4.2. Convergent validity

To first explore whether the chrononutrition estimates derived from the CP-Q and the criterion measures differed significantly, paired samples t-tests were conducted for each pair of variables presented in Fig. 1. Table 4 displays mean differences and effect sizes for each pair of variables.

Table 4.

Paired samples t-tests between CP-Q-assessed chrononutrition behaviors and criterion measures (N = 197).

Variable	Mean difference	t	n	Two-sided p	Effect size (Cohen’s d)
Work/school day
Wake time	0:10	3.87	195	<0.001	0.28
First eating event	0:12	1.64	197	0.101	0.12
Evening eating	0:15	2.29	196	0.023	0.16
Evening latency	15.26	2.21	192	0.029	0.16
Eating window	24.20	2.34	194	0.020	0.17
Eating midpoint	0:01	0.34	194	0.734	0.02
Bedtime	0:01	0.838	193	0.403	0.06
Free day
Wake time	1:20	15.70	196	<0.001	1.12
First eating event	0:08	1.46	195	0.146	0.10
Evening eating	0:08	1.09	196	0.278	0.08
Evening latency	35.63	4.46	190	<0.001	0.32
Eating window	1.33	0.14	195	0.892	0.01

Note. CP-Q = Chrononutrition Profile – Questionnaire.

Note. Wake time, first eating event, evening eating, eating midpoint, and bedtime represented as HH:MM, and evening latency and eating window expressed as minutes.

Note. Due to small number of missing data points, all available data points were used in analyses.

Note. Cohen’s d calculated as d = t/√n.

To evaluate convergent validity of chrononutrition behaviors derived from the CP-Q, Pearson product-moment correlations were calculated between CP-Q and corresponding values from the PSQI and the food intake log. These ranged from moderate to strong positive associations (see Table 5).

Table 5.

Pearson correlation coefficients (r) for CP-Q items and corresponding PSQI and food intake log items (N = 197) and Pearson Correlation coefficients (r) for those items with outliers (top 5% of cases removed) (N = 187).

CP-Q item	Pearson correlation coefficient (r)	Pearson correlation coefficient (r) with top 5% of cases removed
Work/school day
Wake time	0.89	0.98
First eating event	0.69	0.89
Evening eating	0.56	0.71
Evening latency	0.48	0.68
Eating window	0.53	0.75
Eating midpoint	0.72	0.85
Bedtime	0.91	0.94
Free day
Wake time	0.69	0.79
First eating event	0.77	0.94
Evening eating	0.53	0.72
Evening latency	0.39	0.59
Eating window	0.51	0.70
Eating midpoint	0.77	0.86
Bedtime	0.81	0.86

Note. CP-Q = Chrononutrition Profile – Questionnaire; PSQI = Pittsburgh Sleep Quality Index.

Note. Analyses were based on all available data due to missing items; actual sample sizes ranged from N = 190 to N = 197 and N = 180 to N = 187.

Note. All correlations significant at the p < .001 level.

To further assess convergent validity of CP-Q-assessed chrononutrition behaviors, we then estimated the level of agreement or discrepancy between evening latency, evening eating, eating window, and eating midpoint on work/school days and free days as reported in the CP-Q and the food intake log, using descriptive statistics. Descriptive analyses showed that approximately 65% of the sample had no more than 60 min of discrepancy on these chrononutrition behaviors between the CP-Q and criterion measures; notably, approximately 80% of the sample had no more than 60 min of discrepancy on eating midpoint (see Table 6). Cutoffs for these minutes of discrepancy were based on those provided by Veronda et al. (2020), although for the present study we chose to descriptively assess more stringent tests of agreement as well, by examining the percentage of participants within 15 min of agreement and within 30 min of agreement. These more stringent tests showed a fair level of similarity, with over half of the sample having no more than 30 min of discrepancy in chrononutrition behaviors on work/school days, and the largest percentage of participants displaying agreement on eating midpoint in particular, both for the 15-minute and 30-minute ranges (Table 6).

Table 6.

Descriptive similarity between Chrononutrition Profile-Questionnaire and food intake log-assessed chrononutrition behaviors (N = 197).

		% of sample within range of agreement
Chrononutrition behavior		<−120 min	−120 to −61 min	−60 to 60 min	−30 to 30 min	−15 to 15 min	61 to 120 min	>120 min
Work/school day	Evening latency	6.8	9.9	72.9	56.3	41.6	6.8	3.6
	Evening eating	3.6	7.7	74.4	57.9	47.2	6.7	7.7
	Eating window	6.7	8.2	64.4	51.3	37.1	7.7	12.9
	Eating midpoint	4.1	4.1	79.9	61.9	49.7	7.7	4.1
Free day	Evening latency	5.8	3.7	62.1	39.1	25.4	14.7	13.7
	Evening eating	5.6	7.1	74.0	54.8	46.2	4.1	9.2
	Eating window	11.3	9.2	64.6	45.2	37.1	4.1	10.8
	Eating midpoint	2.6	3.1	83.6	67.5	46.7	6.2	4.6

Note. CP-Q = Chrononutrition Profile – Questionnaire.

Note. Evening eating expressed as HH:MM, evening latency and eating window expressed as minutes.

Note. Analyses were based on all available data due to missing items; actual sample sizes ranged from N = 190 to N = 196.

In descriptively examining the level of discrepancy between the CP-Q and criterion measures, we discovered the presence of some extreme discrepancies in the values reported in the CP-Q and the food intake log. It may be plausible that the moderate correlation coefficients could be due to the presence of a few extreme discrepancies in the values that were reported, as a result of various factors as participants completed the CP-Q and the food intake log (e.g., participant error, participant fatigue with study design). Thus, we wanted to explore whether these extreme discordant responses created an artificial high discrepancy between the two measures. To explore this, we removed the top 5% of the most discrepant values for each chrononutrition behavior and reconducted our descriptive analyses. Removing these outliers resulted in improved correlation coefficients, with values ranging from r = 0.59 (free day evening latency) to r = 0.98 (work/school day wake time) (Table 4).

5. Discussion

The present study aimed to evaluate the validity of the Chrononutrition Profile – Questionnaire (CP-Q), a measure of chrononutrition. The original study by Veronda et al. (2020) provided preliminary evidence for the validity of this measure, but the study was limited by the use of a college student sample. Thus, the goal of the current study was to build upon the original study by testing the validity of the CP-Q in a community-based online sample of adults.

Similar to the original study (Veronda et al., 2020), our online participants reported a broad range of values with normal distribution and acceptable skewness and kurtosis values. In addition, descriptive statistics showed the majority of participants reporting values with less than 60 min of discrepancy between the two assessment methods, and significant moderate correlations were displayed between the CP-Q and corresponding items in the PSQI and food intake log. These were similar in magnitude to those reported in the original study, which found approximately 40% of the sample within 60 min of discrepancy (Veronda et al., 2020). Further, exploratory analyses showed that these correlation coefficients increased when extreme values were excluded, with some correlation coefficients becoming very strong. Overall, these findings indicate acceptable validity for the CP-Q, with some variables derived from the measure appearing stronger (e.g., evening eating) and others indicating a weaker level of validity (e.g., evening latency). Thus, this measure is likely appropriate for use in community-based adult samples as well as college samples.

Because these data were collected in November 2020 during the height of the COVID-19 pandemic in the United States, it is possible that the timing of behaviors such as eating and sleep differed from those in non-pandemic times (González-Monroy et al., 2021; Robillard et al., 2021), and that the pandemic, coupled with self-reported BMI, may have contributed to these findings. To evaluate potential pandemic-related effects, we assessed the perceived impact of the COVID-19 pandemic on body weight, bedtimes and wake times, and timing of food intake in a subsample of 157 participants. Frequencies were computed to see how many individuals reported a significant change in their body weight, bedtimes and wake times, and timing of food intake; overall, many participants did not report a significant change in behavior compared to their behavior before the pandemic (e.g., February 2020). One hundred forty of 157 participants (89.2%) reported that their timing of food intake either did not change, or changed a little, compared to before the pandemic. One hundred forty-three of 157 participants (91.1%) reported their body weight was either about the same, a little heavier, or a little lighter compared to before the pandemic. Lastly, 105 of 157 individuals (66.9%) reported a bedtime of about the same compared to before the pandemic, and 83 of 157 participants (52.87%) reported a wake time of about the same compared to before the pandemic.

While the present sample was more diverse in terms of age and gender compared to the original sample in Veronda et al. (2020), future work should continue to evaluate the CP-Q in more racially and ethnically diverse samples. As compared to Veronda et al. (2020), which consisted of primarily female and white young adult participants, this study consisted of 48.7% male and 25.9% non-white participants with ages ranging from 18 to 65 years. Continued evaluation of this measure in broad and diverse samples is beneficial. For instance, the validity of this measure in some populations remains unknown (e.g., shift workers, clinical populations). As this study excluded shift workers, the generalizability of this study to shift workers is fairly limited; therefore, additional work is warranted to evaluate these chrononutrition constructs in this group. The chrononutrition of shift workers is an important topic of study as shift work has been shown to be linked to the development of overweight and obesity (Sun et al., 2018), adverse mental health outcomes (Torquati et al., 2019), and cardiovascular disease (Torquati et al., 2018).

Unlike the original study, our present analyses were unable to evaluate the validity of CP-Q derived constructs of breakfast skipping, largest meal, and night eating. Evaluation of the validity of these chrononutrition behaviors would be a valuable contribution for future research to address. It should also be noted that the PSQI assessment period is anchored to the past month, whereas the CP-Q assesses both work/school days and free days. However, the high correlation between the bedtime and wake time as reported in the PSQI and as reported on work/school days and free days in the CP-Q suggests there is consistency between the two measures.

Future research should aim to elucidate the role of meal frequency in body weight, body composition, and obesity, as current evidence in this area is mixed, and meal frequency is likely closely related to chrononutrition (for review, see Paoli et al., 2019). For instance, increased meal frequency has been linked to increases in BMI (Kahleova et al., 2017) and type 2 diabetes (Mekary et al., 2012), but older research has suggested more frequent meals may protect against weight gain (e.g., Fabry et al., 1964; Jenkins et al., 1989). In light of these contrasting findings, Paoli et al. (2019) posit that such differences may be due to a combination of factors such as the time in which meals are consumed, the time between meals, and daily fasting periods differentially affecting body weight and body composition. Consideration of meal frequency along with chrononutrition is needed in order to further advance our understanding of these complexities and their influences on body weight and health.

Our supplementary analyses revealed that chrononutrition behaviors were largely not related to demographic characteristics, self-reported BMI, and diet quality. Additional supplementary analyses also suggested that pandemic-related changes in behavior and body weight may have influenced some of the supplementary analyses regarding BMI and dietary intake, but the primary purpose of this study was to conduct a cross-sectional analysis of agreement between CP-Q-derived constructs and previously established validity criterion. Therefore, even if one behavior is altered due to the pandemic, responses on both the CP-Q and the established validity criterion should be altered similarly and both measures would likely still be in agreement if the validity of the CP-Q is indeed adequate.

The CP-Q can be utilized in diverse settings by scientists, stakeholders, and primary care practitioners to serve as a brief but thorough measurement of chrononutrition. Because it only takes a few minutes to complete, it can be completed during a short office or laboratory visit and requires minimal training to administer. The CP-Q may therefore be of high value to researchers for its utility in community samples as research works to elucidate relationships between chrononutrition and health. In sum, this study builds upon past research and serves as an important next step by providing additional evidence for the validity of the Chrononutrition Profile - Questionnaire in an online, community-based sample of adults.