Association of Employees’ Meal Skipping Patterns with Workplace Food Purchases, Dietary Quality, and Cardiometabolic Risk: A Secondary Analysis from the ChooseWell 365 Trial

Abstract

Background

Employed adults may skip meals due to time or financial constraints, challenging work schedules, or limited workplace food choices. Little is known about the relationship between employees’ meal skipping patterns and workplace dietary choices and health.

Objective

To examine whether hospital employees’ meal skipping patterns were associated with workplace food purchases, dietary quality, and cardiometabolic risk factors (ie, obesity, hypertension, and prediabetes/diabetes).

Design

This is a secondary cross-sectional analysis of baseline data from the ChooseWell 365 randomized controlled trial. Employees reported meal-skipping frequency in a baseline survey. The healthfulness of workplace food purchases was determined with a validated Healthy Purchasing Score (HPS) (range = 0 to 100 where higher scores = healthier purchases) calculated using sales data for participants’ purchases in the 3 months before study enrollment. Dietary quality was measured with the 2015 Healthy Eating Index (range = 0 to 100 where higher score = healthier diet) from two 24-hour recalls. Cardiometabolic risk factors were ascertained from clinic measurements.

Participants/setting

Participants were 602 hospital employees who regularly visited workplace cafeterias and enrolled in ChooseWell 365, a workplace health promotion study in Boston, MA, during 2016–2018.

Main outcome measures

Primary outcomes were HPS, 2015 Healthy Eating Index, and cardiometabolic risk factors.

Statistical analyses

Regression analyses examined differences in HPS, 2015 Healthy Eating Index, and cardiometabolic variables by meal skipping frequency, adjusting for demographic characteristics.

Results

Participants’ mean (standard deviation) age was 43.6 (12.2) years and 478 (79%) were women. Overall, 45.8% skipped breakfast, 36.2% skipped lunch, and 24.9% skipped dinner ≥1 day/week. Employees who skipped breakfast ≥ 3 days/week (n = 102) had lower HPS (65.1 vs 70.4; P < 0.01) and 2015 Healthy Eating Index score (55.9 vs 62.8; P < 0.001) compared with those who never skipped. Skipping lunch ≥ 3 days/week and dinner ≥ 1 day/week were associated with significantly lower HPS compared with never skipping. Employees who worked nonstandard shifts skipped more meals than those who worked standard shifts. Meal skipping was not associated with obesity or other cardiometabolic variables.

Conclusions

Skipping meals was associated with less healthy food purchases at work, and skipping breakfast was associated with lower dietary quality. Future research to understand employees’ reasons for skipping meals may inform how employers could support healthier dietary intake at work.

MANY WORKING AMERICANS SPEND A LARGE portion of their day at the workplace¹ and consume beverages, meals, and snacks acquired at work.² Structural work conditions such as shift schedules and accessibility of food on site are associated with eating behaviors, including missing or skipping meals.³ Employees who work nonstandard shifts, such as nurses and other hospital workers, engage in more frequent meal skipping during their work shifts.^4,5 Workers with family and socioeconomic burdens (eg, caretaking duties and financial stress), and those who work very long hours, may also be more likely to skip meals at work or at home.³ Many workplaces provide limited onsite meal options and limited healthy foods.² Even when a variety of foods are available, workers may not be able to access these options due to time pressure or difficulty leaving their work task.

Breakfast skipping has been associated with obesity,^6,7 type 2 diabetes,^8,9 cardiovascular diseases,^10,11 and poor dietary intake.¹² Breakfast skipping has increased in the past few decades;¹³ an estimated 17% to 30% of US adults do not eat breakfast.^13–15 Despite the proliferation of research on breakfast skipping in recent years, research about the associations of lunch or dinner skipping with cardiometabolic health is limited. Skipping either lunch or dinner was associated with lower dietary quality in the National Health and Nutrition Examination Study,¹⁶ and another study found that skipping lunch or dinner was associated with obesity¹⁷ in a large cohort of Chinese adults.

Given the time people spend at work, the workplace provides an opportunity to promote and sustain healthy dietary behaviors. Even simple, low-touch “nudge” interventions in workplace food environments, such as product placement (ie, choice architecture) and traffic light food labeling, improve the healthfulness of employees’ purchases.^18–20 Little is known about the relationship of employees’ meal-skipping patterns with food and beverage purchasing at work. The goal of this study was to determine whether breakfast, lunch, or dinner skipping by hospital employees was associated with the healthfulness of workplace food purchases, overall dietary quality, and cardiometabolic risk factors.

MATERIALS AND METHODS

Setting and Participants

Participants were employees of a large urban academic medical center in Boston, MA, who enrolled in a randomized controlled trial of a workplace health promotion intervention (ChooseWell 365; ClinicalTrials.gov identifier: NCT02660086) between September 2016 and February 2018.²¹ The current study examined baseline data that was collected before randomization. During the study, the hospital had six onsite food retail locations, including four cafeterias and two cafes (hereafter referred to as cafeterias). All employees could opt to purchase cafeteria items by payroll deduction using employee identification cards, and their purchases could be tracked by their employee identification number. Employees were eligible to participate in the trial in the case that they were between ages 20 and 75 years and used their employee badge to purchase items at hospital cafeterias at least four times per week for at least 6 weeks before recruitment. Each week during the recruitment period (September 2016 through December 2017) E-mail messages were sent to 100 to 200 randomly selected employees who met cafeteria purchasing criteria, as identified using cafeteria sales data, to ask whether or not they were interested in participating. Employees who did not respond to the first E-mail message were E-mailed up to two more times in the following 3 months. If employees were interested in participating, they completed additional screening by telephone. Employees were excluded in the case that they were pregnant, reported a desire to gain weight, were participating in a weight-loss study, had a history of an eating disorder, had weight-loss surgery during the past 6 months, were employed as cafeteria staff, or planned to leave employment in the upcoming year. A total of 3,293 employees were approached for participation based on meeting cafeteria purchasing criteria. Of those, 868 responded and completed telephone screening, 744 were eligible, 656 completed informed consent, and 602 completed baseline assessment and randomization.²² Baseline assessment included an online questionnaire, two online 24-hour dietary recalls, and an in-person clinical visit. Study procedures were reviewed and approved by the Mass General Brigham Institutional Review Board on October 2, 2015. All participants provided written informed consent.

Traffic light food labeling has been utilized in hospital food retail locations since 2010 to inform employees of the healthfulness of food and drink items (green label = healthy, yellow label = less healthy, and red label = unhealthy). The labeling algorithm was created by hospital dietitians based on the 2010 US Department of Agriculture Dietary Guidelines for Americans²³ and utilized positive and negative nutritional criteria for each item.^18,21

Measures and Outcomes

Meal Skipping Frequency.

Participants self-reported meal skipping frequency in the week prior to survey completion by responding to three items, one for each meal: “On how many days did you skip (breakfast/lunch/dinner) for any reason?” Response options were Never Skip, Skip 1 to 2 Days, Skip 3 to 4 Days, Skip 5 to 6 Days, or Skip Every Day.

Demographic Information.

Participants completed an electronic self-report questionnaire to assess demographic and psychosocial variables (eg, age, sex, race, ethnicity, marital status, education level, household size, and caregiving responsibility) and health history, including smoking status and desire to lose or maintain weight. Based on their hospital job titles, participants were categorized into four job types that roughly aligned with increasing education attainment: service workers and administrative assistants, craft workers/technicians (eg, technicians and respiratory therapists), management/professionals (eg, hospital managers, nurses, and social workers), and MDs/PhDs (eg, physicians and researchers). Standard and nonstandard shift work status was assessed by asking participants, “In the PAST MONTH, did you typically work a nonstandard shift schedule (starts before 7am or after 2pm, rotates, or regularly includes hours outside of the standard 7am to 6pm work day)?” Participants who responded “yes” were considered to have a nonstandard work schedule, and those who responded “no” were considered to have a standard work schedule.

Cafeteria Food Purchases.

Participants’ food and beverage purchases during the 3 months before enrollment were collected retrospectively via the cafeteria cash register data system. Item type, time and date of purchase, and the assigned traffic light label color (ie, red, yellow, or green) were recorded for each item. To capture typical purchasing patterns over time and dilute the influence of occasional atypical purchases, 3 months of purchasing data were analyzed for each participant. A previously validated Healthy Purchasing Score (HPS) was generated to reflect the overall healthfulness of an employee’s baseline purchases over a 3-month period.²⁴ For each participant, the percentage of red items purchased during the baseline period was multiplied by zero, the percentage of yellow items by 50, and the percentage of green items by 100. The HPS was the sum of these values, which ranged from zero (100% red items) to 100 (100% green items). For example, in the case that an employee’s 3-month baseline purchases were 40% red-labeled items, 30% yellow items, and 30% green items, the HPS would be: (0.4 red × 0) + (0.3 yellow × 50) + (0.3 green × 100) = 45.

Dietary Quality.

Participants completed the Automated Self-Administered 24-hour (ASA24) to measure dietary quality and average daily calorie intake. Created by the National Cancer Institute, the ASA24 is a free web-based dietary intake assessment tool modeled after the US Department of Agriculture interview-administered dietary recall method.²⁵ The ASA24 guides respondents through a 24-hour recall period using multilevel probes and is a valid measure of dietary intake in adults.^26,27 The majority of participants completed two ASA24 recalls on nonconsecutive days during the baseline period, and 38 participants (6.3%) completed only one recall. An estimate of daily calorie intake was calculated by averaging the daily calories reported in each ASA24. Although multiple 24-hour dietary recalls do not represent average or overall dietary intake,²⁸ this measure was used as a proxy for daily calories as a way to understand the associations between meal skipping and caloric intake. To assess overall dietary quality, Healthy Eating Index 2015 (HEI-2015) score^29,30 was generated for each participant from combined ASA24 recall scores using the National Cancer Institute bivariate scoring algorithm.³¹ For participants who completed only one ASA24, HEI-2015 score was calculated based on that recall alone. Scores range from 0 to 100, with higher scores indicating higher accordance with 2015–2020 Dietary Guidelines for Americans.³² The mean HEI-15 score for Americans is 59.³³

Cardiometabolic Risk Factors.

Baseline clinical assessments were conducted by research nurses at the hospital’s Translational and Clinical Research Center and included measurements of weight and height used to calculate body mass index (BMI) and obesity (BMI ≥30), as well as blood pressure, fasting glucose, and fasting glycated hemoglobin (HbA1c) level. Height and weight were measured with participants in light clothing with shoes removed. Weight was measured once with a Tanita BWB-800S Scale. Height was measured in triplicate using a Holtain Harpenden Stadiometer and an average height was calculated. Blood pressure was measured once with a medical grade automatic blood pressure monitor. Fasting glucose and HbA1c were processed in internal laboratories using a Cobas C702 Chemistry analyzer for glucose and a boronate affinity high-performance liquid chromatography assay and the Premier Hb0210 analyzer for HbA1c. Hypertension was defined as self-reported hypertension/high blood pressure diagnosis and/or self-reported prescription antihypertension medication and/or systolic blood pressure ≥150 mm Hg or diastolic blood pressure ≥90 mm Hg. Threshold scores for systolic and diastolic blood pressure were set slightly higher than the diagnostic criteria for hypertension³⁴ to account for possible small elevations in blood pressure due to being in the presence of a physician or other health care provider (ie, so-called white coat syndrome).³⁵ Prediabetes/diabetes was defined as self-reported diabetes or prediabetes diagnosis and/or self-reported prescription medication for diabetes and/or HbA1c ≥ 5.7.³⁶

Statistical Analyses

Descriptive characteristics were assessed by meal skipping frequency (Never skip, Skip 1 to 2 days/week, Skip ≥ 3 days/week) for each meal (breakfast, lunch, dinner). Pearson χ² tests and analysis of variance were used to assess differences in participant characteristics by skipping frequency. Linear regression analyses examined differences in HPS, HEI-2015, and average daily calories on days ASA24s were completed by meal skipping frequency for each meal, using Never Skip as the referent category, and adjusting for age, sex, race, ethnicity, education, job type, shift work status, marital status, household size, 1-year weight goal (to maintain or lose weight), and total number of cafeteria purchases. Marginal means for HPS, HEI-2015, and daily calories were estimated by meal skipping frequency for each meal. Logistic regression analyses examined differences in the prevalence of obesity, prediabetes/diabetes, and hypertension using the same main exposure measure and the same covariates. Marginal mean prevalence scores were estimated by meal skipping frequency for each meal. As a sensitivity analysis, linear regression models were conducted using the continuous outcome variables that correspond to each of these conditions (BMI, HbA1c, and systolic and diastolic blood pressure). To ensure that analyses included individuals who regularly purchased food at the workplace, the 1% lowest purchasers (individuals who purchased < 34 items in the 3-month baseline period [n = 7]) were excluded from analyses. Analyses were conducted using Stata statistical software version 15.1.³⁷ Statistical significance was set at P < 0.05.

RESULTS

Descriptive characteristics of the full sample (N = 602) by breakfast skipping frequency are presented in Table 1. Participants’ mean age (standard deviation) was 43.6 (12.2) years and 478 (79%) were women. Employees who skipped breakfast ≥3 days per week were younger and fewer women skipped breakfast than men. Employees who were Black, had lower education levels, or had lower-wage jobs (eg, administrative workers and service employees) skipped breakfast more frequently than employees who were White, had higher education, and had higher paying jobs. Descriptive statistics by lunch and dinner skipping are shown in Tables 2 and 3 (available at www.jandonline.org).

Table 1.

Baseline descriptive statistics by weekly breakfast skipping frequency^a for 602^b hospital employees participating in the ChooseWell 365 study in Massachusetts during September 2016 through February 2018

Variable	Never (n = 326)	1–2 d (n = 174)	≥ 3 d (n = 102)	P valuec
	mean (standard deviaton)
Age (y)	45.2 (12.1)	42.4 (12.7)	40.8 (11.2)	0.002
Body mass index	27.6 (6.0)	29.7 (7.3)	27.9 (6.5)	0.003
Household size	2.9 (1.3)	2.9 (1.5)	2.8 (1.2)	0.74
	n (%)
Sex				< 0.001
Female	279 (58.4)	136 (28.5)	63 (13.2)
Male	47 (37.9)	38 (30.7)	39 (31.5)
Race				0.001
White	282 (57.8)	126 (25.8)	80 (16.4)
Black	17 (31.5)	27 (50.0)	10 (18.5)
Asian	13 (48.2)	5 (18.5)	9 (33.3)
Other/not reportedd	14 (42.4)	16 (48.5)	3 (0.1)
Ethnicity				0.68
Non-Hispanic/Latino/a	301 (54.1)	160 (28.8)	95 (17.1)
Hispanic/Latino/a	21 (61.8)	8 (23.5)	5 (14.7)
Job type				0.001
Administrative/service	34 (40.5)	31 (36.9)	19 (22.6)
Craft/technicians	26 (38.8)	30 (44.8)	11 (16.4)
Management/professionals	226 (60.0)	95 (25.2)	56 (14.9)
MDs/PhDs	40 (54.1)	18 (24.3)	16 (21.6)
Education level				<0.001
High school/some college	30 (40.0)	30 (40.0)	15 (20.0)
College degree	113 (47.1)	79 (32.9)	48 (20.0)
Graduate degree	183 (64.4)	62 (21.8)	39 (13.7)
Marital status				0.013
Married/living with partner	223 (60.3)	91 (24.6)	56 (15.1)
Widowed	4 (57.1)	3 (42.9)	0 (0.0)
Single, living with others	51 (41.1)	46 (37.1)	27 (21.8)
Single, living alone	45 (49.5)	29 (31.8)	17 (18.7)
Caregiving responsibility e				0.78
Caregiver	157 (55.3)	82 (28.9)	45 (15.9)
Not a caregiver	169 (53.1)	92 (28.9)	57 (17.9)
Smoking status				0.009
Current smoker	4 (23.5)	7 (41.2)	6 (35.3)
Nonsmoker	321 (55.6)	163 (28.3)	93 (16.1)
1-y weight goal f				0.45
Lose weight	274 (54.8)	148 (29.4)	81 (16.1)
Maintain weight	52 (52.5)	26 (26.3)	21 (21.2)

^aDescriptive statistics shown by breakfast skipping as breakfast was the most frequently skipped meal. Descriptive statistics by lunch and dinner skipping are shown in Tables 2 and 3 (available at www.jandonline.org).

^bSome variables do not total to 602 because a small number of participants selected the option “Prefer not to respond” for the item on the survey.

^cP value for group difference using Pearson’s χ² tests (categorical variables) and analyses of variance (continuous variables).

^dOther = Native Hawaiian/Pacific Islander, Native American/Alaskan Native, or preferred not to answer.

^eCaregivers were participants who reported providing regular care outside of their employment for children, grandchildren, spouse, parent, or other person with a disability or illness.

^fSelf-reported goal for weight during the upcoming year.

Table 2.

Baseline descriptive statistics by weekly lunch skipping frequency for 602^a hospital employees participating in the ChooseWell 365 study in Massachusetts during September 2016 through February 2018

Variable	Never (n = 384)	1–2 d (n = 184)	≥ 3 d (n = 34)	P valueb
mean (standard deviation)
Age (y)	42.9 (12.2)	45.4 (11.9)	42.9 (13.9)	0.070
Body mass index	27.8 (6.2)	29.4 (7.3)	27.3 (4.5)	0.019
Household size	2.9 (1.4)	2.9 (1.4)	2.8 (1.2)	0.86
n (%)
Sex				0.066
Female	312 (65.3)	144 (30.1)	22 (4.6)
Male	72 (58.1)	40 (32.3)	12 (9.7)
Race				<0.001
White	327 (67.0)	132 (27.1)	29 (5.9)
Black	24 (44.4)	28 (51.9)	2 (3.7)
Asian	19 (70.4)	8 (29.6)	0 (0.0)
Other/not reportedc	14 (42.4)	16 (48.4)	3 (0.1)
Ethnicity				0.171
Non-Hispanic/Latino/a	361 (64.9)	164 (29.5)	31 (5.6)
Hispanic/Latino/a	20 (58.8)	14 (41.2)	0 (0.0)
Job type				<0.001
Administrative/service	39 (46.4)	43 (51.2)	2 (2.4)
Craft/technicians	47 (70.2)	20 (29.9)	0 (0.0)
Management/professionals	249 (66.1)	103 (27.3)	25 (6.6)
MDs/PhDs	49 (66.2)	18 (24.3)	7 (9.5)
Education level				0.001
High school/some college	42 (56.0)	32 (42.7)	1 (1.3)
College degree	143 (59.6)	85 (35.4)	12 (5.0)
Graduate degree	198 (69.7)	65 (22.9)	21 (7.4)
Marital status				0.044
Married/living with partner	248 (67.0)	100 (27.0)	22 (6.0)
Widowed	4 (57.1)	3 (42.9)	0 (0.0)
Single, living with others	69 (55.7)	48 (38.7)	7 (5.7)
Single, living alone	61 (67.0)	26 (28.6)	4 (4.4)
Caregiving responsibility d				0.926
Caregiver	179 (63.0)	89 (31.3)	16 (5.6)
Not a caregiver	205 (64.5)	95 (29.9)	18 (5.7)
Smoking status				0.286
Current smoker	7 (41.2)	9 (52.9)	1 (5.9)
Nonsmoker	372 (64.5)	173 (30.0)	32 (5.5)
1-y weight goal e				0.599
Lose weight	317 (63.0)	156 (31.0)	30 (6.0)
Maintain weight	67 (67.7)	28 (28.3)	4 (4.0)

^aSome variables do not total to 602 because a small number of participants selected the option “Prefer not to respond” for the item on the survey.

^bP value for group difference using Pearson χ² tests (categorical variables) and analyses of variance (continuous variables).

^cOther = Native Hawaiian/Pacific Islander, Native American/Alaskan Native, or preferred not to answer.

^dCaregivers were participants who reported providing regular care outside of their employment for children, grandchildren, spouse, parent, or other person with a disability or illness.

^eSelf-reported goal for weight during the upcoming year.

Table 3.

Baseline descriptive statistics by weekly dinner skipping frequency for 602^a hospital employees participating in the ChooseWell 365 study in Massachusetts during September 2016 through February 2018

Variable	Never (n = 452)	1–2 d (n = 121)	≥ 3 d (n = 29)	P valueb
mean (standard deviation)
Age (y)	43.8 (12.1)	42.4 (12.9)	46 (11.5)	0.30
Body mass index	27.9 (6.4)	28.6 (6.2)	32.2 (7.8)	0.002
Household size	2.9 (1.3)	3.0 (1.5)	2.3 (1.0)	0.05
n (%)
Sex				0.039
Female	351 (73.4)	99 (20.7)	28 (5.9)
Male	101 (81.5)	22 (17.7)	1 (0.8)
Race				< 0.001
White	384 (78.7)	85 (17.4)	19 (3.9)
Black	27 (50.0)	19 (35.2)	8 (14.8)
Asian	22 (81.5)	5 (18.5)	0 (0.0)
Other/Not reportedc	19 (57.6)	12 (36.4)	2 (0.6)
Ethnicity				0.052
Non-Hispanic/Latino/a	426 (76.6)	104 (18.7)	26 (4.7)
Hispanic/Latino/a	20 (58.8)	12 (35.3)	2 (5.9)
Job type				< 0.001
Administrative/service	51 (60.7)	22 (26.2)	11 (13.1)
Craft/technicians	44 (65.7)	21 (31.3)	2 (3.0)
Management/professionals	290 (76.9)	72 (19.1)	15 (4.0)
MDs/PhDs	67 (90.5)	6 (8.1)	1 (1.4)
Education level				0.005
High school/some college	48 (64.0)	20 (26.7)	7 (9.3)
College degree	171 (71.3)	57 (23.8)	12 (5.0)
Graduate degree	232 (81.7)	42 (14.9)	10 (3.5)
Marital status				< 0.001
Married/living with partner	304 (82.2)	58 (15.7)	8 (2.2)
Widowed	6 (85.7)	1 (14.3)	0 (0.0)
Single, living with others	77 (62.1)	35 (28.2)	12 (9.7)
Single, living alone	63 (69.2)	22 (24.2)	6 (6.6)
Caregiving responsibility d				0.428
Caregiver	220 (77.5)	51 (18.0)	13 (4.6)
Not a caregiver	232 (73.0)	70 (22.0)	16 (5.0)
Smoking status				< 0.001
Current smoker	4 (23.5)	11 (64.7)	2 (11.8)
Nonsmoker	444 (77.0)	107 (18.5)	26 (4.5)
1-year weight goal e				0.229
Lose weight	372 (74.0)	104 (20.7)	27 (5.4)
Maintain weight	80 (80.8)	17 (17.2)	2 (2.0)

^aSome variables do not total to 602 because a small number of participants selected the option “Prefer not to respond” for the item on the survey.

^bP value for group difference using Pearson χ² tests (categorical variables) and analyses of variance (continuous variables).

^cOther = Native Hawaiian/Pacific Islander, Native American/Alaskan Native, or preferred not to answer.

^dCaregivers were participants who reported providing regular care outside of their employment for children, grandchildren, spouse, parent, or other person with a disability or illness.

^eSelf-reported goal for weight during the upcoming year.

Self-reported frequency of skipping (never, 1 to 2 days per week, or ≥ 3 days per week) for each meal (breakfast, lunch, dinner) are presented in Table 4 for the full sample and by shift work status. Overall, breakfast was the most frequently reported skipped meal; 46% of the sample reported skipping breakfast ≥ 1 day per week, whereas 36% skipped lunch and 25% skipped dinner ≥ 1 day per week. More than one-third of the sample (36.2%) worked nonstandard shifts (ie, evening/night shifts, rotating shifts, or hours outside of 7 AM to 6 PM), and these employees skipped more meals than employees who worked standard shifts. For example, 24.3% of nonstandard shift workers skipped breakfast ≥ 3 times/week, compared with 12.8% of standard shift workers.

Table 4.

Baseline frequency of weekly meal skipping by shift work status for 602^a hospital employees participating in the ChooseWell 365 study in Massachusetts during September 2016 through February 2018

Meal	Frequency	Total Sample		Standard Shift Workb		Nonstandard Shift Work		P valuec
n (%)
Breakfast skipping	Never	326	54.2	225	58.9	100	45.9	0.003
	1–2 d	174	28.9	108	28.3	65	29.8
	≥ 3 d	102	16.9	49	12.8	53	24.3
Lunch skipping	Never	384	63.8	256	67.0	128	58.7	< 0.001
	1–2 d	184	30.6	114	29.8	69	31.7
	≥ 3 d	34	5.7	12	3.1	21	9.6
Dinner skipping	Never	452	75.1	291	76.2	161	73.9	0.037
	1–2 d	121	20.1	70	18.3	49	22.5
	≥ 3 d	29	4.8	21	5.5	8	3.7

^aThe total N for shift work status categories is 600 because two participants selected the option “Prefer not to respond” for this item on the survey.

^bStandard shift work = day shifts (eg, 7:00 AM to 6:00 PM); nonstandard shifts = shifts that start before 7 AM or after 2 PM, rotating shifts, or shifts that regularly include hours outside of the standard 7 AM to 6 PM work day.

^cP value for group (shift work status) difference using Pearson χ² tests.

Table 5 indicates inverse associations of meal skipping frequency with overall dietary quality (ie, HEI-2015) and workplace food purchase quality (ie, HPS). Participants who reported skipping ≥ 3 days per week of breakfast or lunch, or 1 to 2 days per week of dinner, had a lower adjusted mean HPS than participants who reported never skipping these meals. Participants who reported any breakfast skipping (1 to 2 days per week or ≥ 3 days per week) had a lower adjusted mean HEI-2015 score compared with participants who reported no breakfast skipping. Average daily calories on days measured by dietary recalls were not significantly different for participants who skipped meals. Additional linear regression model results (eg, β [standard error]) showing the associations of meal skipping frequency with HPS, HEI-2015, and daily calories are included in Table 6.

Table 5.

Adjusted mean^a Healthy Purchasing Score, Healthy Eating Index-2015, and daily calorie intake at baseline by weekly meal skipping frequency for hospital employees participating in the ChooseWell 365 study in Massachusetts during September 2016 through February 2018

Meal	Never Skip	Skip 1–2 d	Skip ≥ 3 d
	Mean (standard deviation)
Breakfast	(n = 326)	(n = 174)	(n = 102)
Healthy Purchasing Score	70.4 (0.7)	69.5 (1.0)	65.1 (1.3)**
Healthy Eating Index-2015d	62.8 (0.7)	58.5 (1.0)***	55.9 (1.3)***
Average daily calories (from ASA24e days)	1808 (32)	1787 (45)	1779 (59)
Lunch	(n = 384)	(n = 184)	(n = 34)
Healthy Purchasing Score	69.9 (0.7)	68.7(1.0)	63.4 (2.4)**
Healthy Eating Index-2015	61.2 (0.6)	59.4 (1.0)	57.6 (2.3)
Average daily calories (from ASA24 days)	1830 (29)	1743 (44)	1698 (101)
Dinner	(n = 452)	(n = 121)	(n = 29)
Healthy Purchasing Score	69.8 (0.6)	66.8 (1.2)*	69.3 (2.5)
Healthy Eating Index-2015	60.6 (0.6)	59.3 (1.2)	62.9 (2.4)
Average daily calories (from ASA24 days)	1821 (27)	1744 (54)	1651 (107)

^aMarginal means from linear regression models; analyses adjusted for age, sex, race, ethnicity, education, shift work status, marital status, household size, job type, 1-year weight goal, and number of items purchased (excluding 1% lowest purchasers [< 34 items]), and used “Never Skip” as the reference group.

^dHealthy Eating Index-2015 is a dietary quality score for the full day, calculated from Automated Self-Administered 24-hour Dietary Recalls.

^eASA24 = Automated Self-Administered 24-hour Dietary Recall.

^*P < 0.05 for comparisons between the “Skip 1–2 days” or “Skip 3+ days” groups to the “Never Skip” group.

^**P < 0.01 for comparisons between the “Skip 1–2 days” or “Skip 3+ days” groups to the “Never Skip” group.

^***P < 0.001 for comparisons between the “Skip 1–2 days” or “Skip 3+ days” groups to the “Never Skip” group.

Table 6.

Linear regression results^a for associations of baseline weekly meal skipping with baseline dietary and cardiometabolic outcomes for hospital employees participating in the ChooseWell 365 study in Massachusetts during September 2016 through February 2018^b

	Healthy Purchasing Score		Healthy Eating Index-2015c		Average Daily Caloriesd		Body mass index		Hemoglobin A1c (%)		Systolic blood pressure (mm Hg)		Diastolic blood pressure (mm Hg)
Meal skipping frequency (d/wk)	β (standard error)	P value	β (standard error)	P value	β (standard error)	P value	β (standard error)	P value	β (standard error)	P value	β (standard error)	P value	β (standard error)	P value
Breakfast
1–2 vs 0	−0.85 (1.3)	0.51	−4.37 (1.2)	< 0.001	−35 (58)	0.55	0.92 (0.5)	0.09	−0.01 (0.1)	0.80	1.85 (1.3)	0.16	1.39 (0.9)	0.14
3+ vs 0	−5.31 (1.6)	0.001	−6.93 (1.5)	< 0.001	−1 (72)	0.99	−0.44 (0.6)	0.50	−0.04 (0.1)	0.55	0.26 (1.6)	0.87	0.48 (1.1)	0.67
Lunch
1–2 vs 0	−1.10 (1.2)	0.37	−1.83 (1.2)	0.13	0 (55)	0.99	0.08 (0.5)	0.88	−0.04 (0.1)	0.38	1.34 (1.2)	0.28	0.20 (0.9)	0.82
3+ vs 0	−6.50 (2.4)	0.008	−3.65 (2.4)	0.12	24 (109)	0.83	−0.93 (1.0)	0.36	0.03 (0.1)	0.76	3.21 (2.5)	0.20	1.04 (1.8)	0.60
Dinner
1–2 vs 0	−3.0 (1.4)	0.03	−1.32 (1.4)	0.34	3 (64)	0.97	−0.41 (0.6)	0.49	0.08 (0.1)	0.19	2.12 (1.4)	0.14	0.62 (1.0)	0.54
3+ vs 0	0.10 (2.7)	0.97	2.48 (2.5)	0.33	−124 (117)	0.29	1.70 (1.1)	0.13	0.05 (0.1)	0.66	6.90 (2.7)	0.01	2.48 (1.9)	0.19

^aExample of interpretation of the regression coefficient for breakfast and Healthy Purchasing Score: Compared with employees who skip 0 meals per week, employees who skip 3+ meals per week have a 5.31 point lower average Healthy Purchasing Score.

^bAnalyses adjusted for age, sex, race, ethnicity, education, shift work status, marital status, household size, job type, 1-year weight goal, and number of items purchased (excluding 1% lowest purchasers [< 34 items]).

^cHealthy Eating Index-2015 is a dietary quality score for the full day, calculated from Automated Self-Administered 24-hour Dietary Recalls.

^dAverage daily calories on days measured by 24-hour dietary recall (two Automated Self-Administered 24-hour Dietary Recalls).

The regression-adjusted mean prevalence of obesity, hypertension, and prediabetes/diabetes by meal skipping frequency is displayed in the Figure. There were no significant differences in obesity, prediabetes/diabetes, or hypertension prevalence by meal skipping frequency. Sensitivity analyses that examined continuous cardiometabolic outcomes had similar results. There were no significant effects of meal skipping frequency for BMI, HbA1c, or diastolic blood pressure for any meal. Skipping dinner ≥ 3 days per week was significantly associated with systolic blood pressure (β = 6.90 [2.7]; P = 0.01). Results are shown in Table 6. All models adjusted for age, sex, race, ethnicity, education, shift work status, marital status, household size, job type, 1-year weight goal, and number of cafeteria items purchased.

Figure.

Adjusted prevalence^a of obesity, prediabetes/diabetes, and hypertension by baseline weekly meal skipping frequency of 602 hospital employees participating in the ChooseWell 365 study in Massachusetts during September 2016 through February 2018. ^aAdjusted prevalence estimates are marginal means rounded to the nearest whole number from logistic regression models that used ‘Never Skip’ as the referent group. Models adjusted for age, sex, race, ethnicity, education, shift work status, marital status, household size, job type, 1-year weight goal, and number of items purchased (excluding 1% lowest purchasers [< 34 items]). No results were statistically significant.

DISCUSSION

In this sample of hospital employees working in a variety of occupations, meal skipping was prevalent and was associated with purchasing less healthy food and beverages at work. Employees who skipped breakfast ≥ 3 times per week had an HEI-2015 score more than five points lower than those who never skipped breakfast, a potentially clinically meaningful difference in dietary quality.^38,39 The observed differences in meal skipping frequency by race, education level, and job type indicate that there may be structural and socioeconomic factors in the lives of employees that increase susceptibility to frequent skipping and unhealthy eating patterns at work. Skipping meals was more common in employees who worked nonstandard shifts (eg, evening and night shifts) compared with those who only worked day shifts.

The ability to compare meal-skipping behavior to the healthfulness of objectively measured food and beverage purchases using the HPS is a unique strength of this study compared with prior literature on meal skipping. Although breakfast skipping has been shown to be associated with lower dietary quality,⁴⁰ including in samples of health professionals,⁴¹ the role of the workplace food environment on dietary choices is unknown. Several studies have shown that workplace accessibility of healthy foods and convenient spaces for food consumption influence employees’ eating behaviors, such as the number and type of calories consumed,⁴² snacking behaviors,^43,44 and fruit and vegetable consumption.^45–47 In a nationally representative sample of working adults in Korea, a country with breakfast skipping prevalence similar to that of the United States, employees who did not or could not utilize workplace foodservice locations reported more meal skipping than those who frequented these locations.⁴⁸ A study from Finland showed that employees who ate more often at workplace cafeterias that served healthy meals were more likely to follow nutrition guidelines.⁴⁹ Although the exact reasons employees in the sample skipped meals is unknown, these results and prior findings suggest that employers could potentially increase their employees’ healthy food intake and help mitigate the negative health impacts of meal skipping^8,10,11 by offering healthy food and beverages in easily accessible locations and providing common spaces to store, prepare, and consume foods at work (eg, refrigerators, microwave ovens, and tables).

The finding that meal skipping was more prevalent in evening or night shift workers is consistent with previous research that has documented the association of shift work and meal skipping,⁵⁰ particularly in health care workers such as nurses.^4,5 A recent systematic review of the influence of shift work on eating behavior found that evening or night shift workers across sectors were more likely to skip meals, eat at unconventional times, and consume more unhealthy foods (eg, saturated fats and sweetened carbonated beverages).⁵⁰ Given that has been documented with nurses,^4,5 this pattern of meal skipping in our sample of hospital workers may be more associated with heavy workload and difficulty taking work breaks than a weight management or dieting strategy. In fact, regardless of shift, more than a third of nurses in the United States report rarely or never taking breaks during their shifts.⁵¹ As evening and night shift employees work during hours when food retail locations may be closed, the availability and proximity of healthy food and spaces for food preparation and consumption are major influencers of eating vs skipping and of the quality of food consumed.^50,52 In hospital or other busy patient care settings, department managers could implement protocols that better facilitate regular breaks for employees to consume meals or snacks, thus providing the opportunity for healthier eating behaviors.

In this sample, meal skipping was also more prevalent among workers with less education and those working in low-wage job categories, such as administrative and service workers. Prior research has established that meal skipping is common in low income and unemployed individuals who are food insecure^53,54; however, few studies have examined meal-skipping patterns in employed adults who earn low wages. Low-wage workers comprise one-third of all workers in the United States and are essential to hospitals and many other industries.⁵⁵ Compared with higher-wage employees, low-wage employees are more likely to have cardiometabolic risk factors such as unhealthy diets and physical inactivity, and they are half as likely as higher wage workers to utilize preventive care.^56–58 This group of workers may be particularly vulnerable to meal skipping and other suboptimal dietary behaviors and may benefit from increased availability of low-cost healthy foods at the workplace and other easily accessible, low time-burden health promotion efforts.

Contrary to some prior research, no significant association was found between breakfast skipping and cardiovascular risk factors (eg, obesity, hypertension, and prediabetes/diabetes). These findings may be due, in part, to the fact that this was a relatively young and working population. One-third of the sample (36%) worked evening/night shifts, and although this group reported higher rates of skipping, they may be skipping or eating at delayed times for different reasons than the general population (ie, difficulty taking work breaks).

This study has limitations that are important to consider. The data are cross-sectional, which limits conclusions about causality. Meal skipping and 24-hour dietary recalls are both self-reported and not observed, and some participants (6.3%) completed only one dietary recall. Whereas 24-hour dietary recalls are considered among the most accurate measures of dietary intake,²⁸ a small number of recalls do not represent overall average dietary intake. The self-reported meal skipping measure reflects participants’ typical pattern of skipping meals throughout the week, whereas the cafeteria purchasing data was collected only during the days and hours the participants were at work, and thus may not capture all typical food and beverages consumed throughout the week. Recently, alternative intake patterns have contributed to some individuals choosing to reduce regular breakfast eating. Specifically, skipping breakfast has become a common intermittent fasting strategy⁵⁹ and breakfast may be the easiest meal to skip for those interested in fasting. Although intermittent fasting gained popularity in the United States after the data from this study was collected (September 2016 through February 2018), it is possible that some participants may have engaged in intermittent fasting with the goal of losing weight or improving health. Similarly, some participants may have been skipping breakfast purpose-fully in an effort to reduce daily calories for weight maintenance or loss. However, recent research has demonstrated that certain individuals may be genetically predisposed to not eating in the morning,⁶⁰ and skipping breakfast for these individuals may not be related to weight loss goals or dietary quality. Lastly, individuals in this sample were relatively young, had a high education level on average, and worked at a large urban hospital. Results may not be generalizable to nonemployed people or working populations with lower education levels, a higher mean age, or in rural settings.

CONCLUSIONS

The findings that meal skipping was associated with unhealthy food and beverage purchases at work and poorer overall dietary intake suggest that meal skipping patterns of employees should be considered in workplace health promotion efforts. Additional research is needed to understand employees’ reasons for skipping meals and to explore whether structural improvements to the workplace food environment could influence workers’ abilities to improve the frequency and quality of their food and beverage consumption. Future studies could explore whether attempts to facilitate healthier intake patterns at work have the potential to improve employees’ dietary quality and overall health.

RESEARCH SNAPSHOT.

Research Question:

Are meal-skipping patterns associated with the healthfulness of workplace food purchases, dietary quality, and cardiometabolic risk factors in employees?

Key Findings:

In this sample of 602 hospital employees, frequent skipping of any meal (breakfast, lunch, or dinner) was associated with less healthy food purchases at work. Skipping breakfast was associated with lower overall dietary quality. Meal skipping was not associated with prevalence of obesity, prediabetes/diabetes, or hypertension.