Interventions for Unhealthy Eating Patterns of Resident Otolaryngologists on Call

ABSTRACT

Objectives

Resident physicians commonly experience elevated stress levels and disrupted sleep while on call. Many also report intense cravings for comfort food, typically high in fat, sugar, and salt. The objective of this pilot study was to explore the physiological and psychological factors contributing to the desire for unhealthy food among US otolaryngology residents on call, and to present strategies for reducing these behaviors. The hypothesis was that the introduction of a simple behavioral intervention would be associated with reduced unhealthy food consumption.

Methods

Twelve otolaryngology residents from a single academic institution were prospectively enrolled and followed over a 61‐day study period consisting of a pre‐ and post‐intervention phase. During the initial 31‐day pre‐intervention period, participants completed a brief survey following each call shift and reported total sleep duration and whether unhealthy food was consumed. During the subsequent 30‐day intervention period, residents were instructed to complete box breathing and/or consume green tea, and reported intervention adherence, sleep duration, and whether unhealthy food was consumed.

Results

Survey responses reporting unhealthy food consumption were associated with shorter sleep duration compared to responses reporting no unhealthy food consumption (mean difference = 1.73 h, CI 1.02–2.45, p < 0.0001). Participation in the intervention phase was associated with a reduction in unhealthy eating habits; however, the change was not significant (p = 0.066). Similarly, residents who performed the intervention of box breathing and/or drinking green tea demonstrated fewer episodes of unhealthy food consumption; however, the magnitude of change was not significant (p = 0.857).

Conclusion

This pilot study evaluated practical behavioral strategies as potential approaches to reducing unhealthy food consumption during call. Although reductions in unhealthy eating episodes were observed, findings should be interpreted cautiously given the small sample size. Further research is needed to design, implement, and assess strategies that support healthier dietary practices among US resident physicians on call.

Level of Evidence

4.

1. Introduction

Globally, resident physicians across specialties report consistently elevated stress levels and frequent disturbances in sleep quality [1, 2, 3]. In the United States in 1984, the death of a patient, attributed in part due to sleep deprivation among training residents, prompted significant changes in the Accreditation Council for Graduate Medical Education (ACGME) guideline policies [4]. Subsequent studies demonstrated that these changes positively influenced resident well‐being [4, 5].

Seventy percent of otolaryngology residents across six institutions surveyed in one study reported that an 80 h workweek, including being on call every third night with no more than 24 h of continuous work without sleep, was a demanding yet reasonable work schedule [6]. However, excessive work hours have been associated with not only reduced sleep duration but also elevated body mass index (BMI), highlighting the broader implications of extended duty on resident physician health [7]. Considering the rising prevalence of lifestyle‐related conditions such as obesity or type II diabetes in recent decades, it is important to explore preventative strategies that also support healthy dietary habits [7].

Effective strategies to promote healthy dietary behaviors require a foundational understanding of the mechanisms driving the increased desire for unhealthy food consumption. Key physiological systems include the sympathoadrenal medullary (SAM) axis and the hypothalamo–pituitary–adrenal (HPA) axis [8, 9, 10, 11]. These hormonal pathways have also been linked to sleep deprivation and its downstream effects on appetite regulation. Consequently, recognizing the impact of sleep deprivation on appetite regulation helps to clarify the extent to which appetite can be modulated through alternative behavioral or physiological interventions. From a psychological perspective, concepts such as optimal foraging theory and sensory‐specific satiety offer further insight into behavioral patterns underlying overconsumption and present potential targets for intervention [10, 12].

Breathing exercises have been shown to reduce stress and anxiety [13]. In a study involving college students with no prior experience in meditation techniques, participants who engaged in structured breathing exercises exhibited a measurable reduction in cortisol levels, indicating an impact on the physiological stress response [14, 15]. Box breathing is a very simple technique which involves inhaling for 5 s, then holding for 5 s, then breathing out for 5 s, then holding again for 5 s [16].

Despite ongoing efforts to improve resident well‐being, sleep deprivation and nutrition management are inherently challenging aspects of resident lifestyles, particularly surgical residents on call. Furthermore, the pathogenesis and interrelationships between stress, sleep, and nutrition are complex. An understanding of these pathways, however, may help identify realistic feasible interventions to improve dietary choices. Thus, the objectives of this study are to describe the physiological and psychological mechanisms that lead to unhealthy food choices, then (1) evaluate if sleep duration and unhealthy food choices are related; (2) evaluate if there is concordance between reduced unhealthy eating and box breathing or drinking green tea.

2. Materials and Methods

2.1. Participants

In the United States, otolaryngology residents at a single institution scheduled for call from March 1, 2025 to April 30, 2025 were prospectively enrolled. Call at this institution is defined as split between day call and a dedicated night float system; 2 residents are assigned each day. All otolaryngology residents scheduled for call irrespective of PGY level were invited to complete the questionnaire. This included residents taking both home and in‐house call. Exclusion criteria for residents included reported food allergies or significant dietary restrictions.

2.2. Procedure

Local institutional review board approval for exemption was obtained.

All participants received standardized electronic communication. The informed consent agreement and questionnaire were emailed to each resident at 7:00 AM on the morning following their scheduled shift. Participants were informed that last names were collected solely for the purpose of accurate matching of pre and post intervention responses, and each participant was assigned a unique randomly generated identifier which was used for accurate matching and in all subsequent analysis performed. Details were also included in the informed consent. For the first 31 days, the pre‐intervention survey collected data on how many hours were slept, if unhealthy food was consumed, and what was consumed (Appendix A). On Day 31, residents were then educated via a one‐page handout on how to practice box breathing (Figure 1). Green tea, as well as protein shakes, protein bars, nuts, and low salt and sugar snacks were made free and available within the resident room, a less than 3‐min walk from the operating rooms. Following this, the post intervention survey collected data on how many hours were slept, if unhealthy food was consumed, what was consumed, and if box breathing was performed or green tea was consumed at any point during their shift (Appendix A).

FIGURE 1.

One‐page educational handout on box breathing and green tea consumption.

2.3. Measures

Unhealthy food was defined as any food with greater than 10 g of sugar per serving or greater than 600 mg of sodium per serving [17, 18]. A “Not sure” response was included to account for the subjectivity of self‐reported responses on unhealthy eating.

2.4. Data Analysis

All analyses were conducted using Microsoft Excel for Microsoft 365 (version 2404; Microsoft Corp., Redmond, WA, USA).

Descriptive statistics were used to report demographics of participating residents.

2.4.1. Objective 1

To assess if sleep duration differs according to unhealthy eating behavior. A two‐sample unequal variance t‐test was used to compare the mean sleep duration of pre‐intervention survey responses between participants who reported unhealthy food consumption and those who did not.

2.4.2. Objective 2

To determine if the intervention of box breathing and/or green tea consumption is associated with a reduction in unhealthy eating episodes. Because pre‐ and post‐intervention responses were paired by participant and sample size was small, changes in unhealthy eating episodes were first assessed using a Wilcoxon signed‐rank test, followed by an exploratory analysis using a Mann Whitney U test to compare participant‐level mean change scores between those that completed the intervention and those that did not.

3. Results

3.1. Participants

During the pre‐intervention phase, 62 of the 62 distributed surveys were collected over the first 31 days. These responses were contributed by 12 residents, with individual residents providing multiple responses during the study period based on the assigned resident call schedule. During the post‐intervention phase, 36 of the 60 distributed surveys were completed over the subsequent 30‐day period (Table 1). These responses represented 8 of the original 12 residents who participated in the pre‐intervention phase, meaning 4 residents did not complete any post‐intervention surveys. Of the 8 residents who completed the survey during the post‐intervention phase, 5 of the residents participated in performing the intervention of either or both box breathing and drinking green tea.

TABLE 1.

Number of pre‐ and post‐intervention survey responses collected from participants.

Resident	Count of post‐survey responses	Count of pre‐survey responses
0XP3	3	5
20GV	3	4
4VUP	0	8
CGDL	2	3
HTJM	4	5
JCY4	0	1
JWAN	0	3
KRN8	2	8
OYDE	4	3
VBKT	9	10
VVMC	9	10
XQ07	0	2
Total	36	62

Among the pre‐intervention survey responses, the majority of responses were from PGY2 (35%) with 21% from PGY1, 18% from PGY3, 10% from PGY4 and 16% from PGY5. Among the post‐intervention survey responses, the majority of responses were from PGY2 (39%) with 11% from PGY1, 25% from PGY3, 6% from PGY4 and 19% from PGY5.

Examples of food reported as healthy included fruits (oranges, strawberries, banana), vegetables (cucumbers, broccoli, potatoes), and protein (chicken, protein shake, pork loin, salmon, eggs). Examples of food self‐reported as unhealthy included sweets (Oreos, chocolate, cookies, candy), fast food (Whataburger, Wendys, Chick‐fil‐a, McDonalds), and soda. Thus, because unhealthy eating episodes were self‐reported by participants using a binary response (Yes/No), all reported unhealthy food consumption meeting the given criteria were classified equally regardless of food type, quantity consumed, or total caloric content.

Pre‐and post‐intervention survey responses were analyzed to assess whether participants may have already established healthy eating habits prior to the study, potentially influencing their ability to maintain such behaviors independent of the intervention. For instance, if participants reported few instances of unhealthy food consumption in both pre‐ and post‐surveys, they should be presumed to have pre‐existing healthy dietary habits. However, data from this sample, in part due to its limited power, yielded inconclusive insights regarding baseline eating behaviors (Figure 2).

FIGURE 2.

Display of the number of residents who reported not eating unhealthy food before (blue) and after (orange) the intervention. Variation across residents indicates differing individual responses, with some showing increased abstention, no change, or a decrease post‐survey.

3.2. Objective 1: To Determine if Sleep Duration and Unhealthy Food Choices Were Related

Survey responses reporting unhealthy food consumption were associated with shorter sleep duration. During the pre‐intervention period, survey responses reporting no unhealthy food consumption had a greater mean sleep duration than responses reporting unhealthy food consumption (Table 2). The mean difference in sleep duration was 1.730 h (95% CI, 1.02–2.45). The two sample t‐test was statistically significant with a difference in mean sleep duration between groups (p < 0.0001).

TABLE 2.

Comparison of sleep duration between pre‐intervention survey responses reporting unhealthy food consumption and no unhealthy food consumption.

Variable	Observations	Minimum sleep (h)	Maximum sleep (h)	Mean sleep (h)	Std. deviation
No unhealthy food consumption	30	3.000	8.000	6.600	1.192
Yes unhealthy food consumption	30	2.000	9.000	4.867	1.548
95% confidence interval	1.02–2.45
t (observed value)	4.860
DF	54.448
p‐value (two‐tailed)	< 0.0001

3.3. Objective 2: To Determine if There is an Association Between Reduced Episodes of Unhealthy Eating and Box Breathing or Drinking Green Tea

Among the eight residents who completed both pre‐ and post‐intervention surveys, five reported a decrease in self‐reported unhealthy eating episodes, three reported no change, and none reported an increase. The mean number of unhealthy eating episodes decreased from 2.38 ± 3.02 before the intervention to 1.00 ± 1.41 after the intervention (Figure 3). A Wilcoxon signed‐rank test demonstrated a trend toward reduced unhealthy eating episodes following the intervention; however, this change did not reach statistical significance (V = 10, p = 0.066). In exploratory analyses, participant‐level change scores were compared according to participation in the intervention using a Mann–Whitney U test (Figure 4). Although residents who participated in the intervention demonstrated a decrease in unhealthy eating episodes, the magnitude of reduction did not differ significantly between residents who did and did not participate in the intervention (U = 7, p = 0.857).

FIGURE 3.

Box plot illustrating the distribution of episodes of unhealthy eating during the pre‐ and post‐intervention phase. Participants during the post‐intervention phase demonstrated a decrease in self‐reported episodes of unhealthy eating.

FIGURE 4.

Box plot illustrating the distribution of the mean difference in episodes of unhealthy eating between the pre‐ and post‐intervention phase for those that performed the intervention and those that did not. No association between performing the intervention and a greater reduction in episodes of unhealthy eating was observed.

4. Discussion

This pilot study was designed to evaluate the feasibility of simple, low‐cost behavioral interventions aimed at reducing unhealthy food consumption among surgical residents during call and immediate post‐call periods. The findings should be interpreted within the context of several methodological limitations and interpreted with caution. The small sample size of responses substantially limits the statistical power resulting in major susceptibility to type I and type II errors. Although reductions in self‐reported unhealthy eating episodes were observed following the intervention, the study was not adequately powered to determine whether these changes represented a true intervention effect. Rather, these findings provide preliminary data to inform the design of larger, adequately powered investigations.

It is possible that the Hawthorne effect may be playing a role here. It is impossible to say which intervention—box breathing or drinking green tea—had the biggest effect given that the intervention took place concurrently. Additionally, an interesting consideration is the relationship between green tea consumption and sleep duration, given that green tea contains caffeine. This association was not examined in the present study; thus it cannot be evaluated whether green tea consumption influenced sleep patterns or whether sleep confounded the relationship between green tea consumption and healthier food choices. Further studies could evaluate these interrelationships. This pilot study went about utilizing the complex interplay of psychological, physiological, and environmental factors that contribute to feeding behaviors to identify interventions. Despite the lack of statistical significance of the results, the observed reduction in unhealthy eating episodes among participating residents who completed the intervention highlights the potential feasibility of implementing simple behavioral strategies within the demanding clinical environment of an otolaryngology residency. Even modest reductions in unhealthy food consumption may have meaningful implications given the cumulative impact of repeated call shifts. However, a larger resident cohort would be needed to provide more data to account for the baseline eating habits of the residents.

Results were also limited by measurement assumptions. Dietary behaviors that are self‐reported may be influenced by recall or social desirability bias, particularly given the small cohort size. Newer applications using the phone may allow for better capture of exactly what food is being eaten and provide a better appreciation of the macronutrients involved. Improved measurements would also address the limitation of implementing a binary classification of unhealthy food consumption which does not capture the differences between types of unhealthy foods. Ultimately, this pilot study underscores the need for larger investigations to improve the overall rigor and clinical utility of the interventions proposed.

Prior research has demonstrated that resident physicians experience stress and significant sleep loss [4, 5, 19, 20]. In the current study, survey responses revealed comparable results. Stress is defined as the perception of a threat resulting in a state of mental tension caused by a difficult situation [21, 22]. Sleep deprivation is acceptably characterized as a sleep episode of four hours or less per 24 h period [21, 22]. Many studies have also reported high levels of perceived stress among US residents [22]. In addition to stress, healthcare workers frequently report insufficient time to eat, poor access to food stations, and the predominance of unhealthy food options at available stations [23]. Consequently, there is a need to better understand the ways in which the unhealthy eating habits of surgical residents on call can be reduced, despite limited sleep.

4.1. Physiological Pathway

Stress triggers two major physiological pathways. The first is the SAM axis, where sympathetic nerve terminals stimulate the adrenal medulla to release adrenaline and noradrenaline [11]. A previous study examining the link between stress, craving popular snacks, and performance motivated tasks found a positive correlation between perceived challenge, craving for chocolate, and increased SAM axis activation [24]. These findings suggested that the SAM response to psychological stress due to a motivational state to perform under pressure may contribute to cravings despite noradrenaline release [24]. These findings may help motivate the need for further research into the underlying mechanism of food cravings, with particular attention to the types of foods most desired during periods of stress.

The second pathway is the HPA axis in which ultimately, the paraventricular nucleus of the hypothalamus is signaled to release corticotrophin releasing hormone (CRH) [11]. Within the hypothalamus, CRH modulates the expression of orexigenic signals NPY/AgRP which in turn relates to the sensitivity of leptin and ghrelin [8, 9]. Ghrelin is an appetite‐stimulating hormone and associated with increased food intake. Leptin is an appetite‐inhibiting hormone and is associated with decreased food intake. Early experimental studies investigated the relationship between sleep deprivation, leptin levels, and hunger and found that partial sleep deprivation was associated with decreased leptin, increased ghrelin, and increased hunger [25].

As mentioned, cortisol release in response to stress is also associated with reduced dopaminergic activity. To cope with a stress response, the body simultaneously releases endogenous opioids. One theory suggested that the consumption of palatable foods high in fat and sugar stimulates the release of endogenous opioids [10]. As a result, under chronic stress, the intake of such foods enhances dopamine levels and attenuates the HPA axis activation through opioid release [10, 26]. Moreover, high sucrose intake has been shown to alleviate anxiety‐like behaviors through inhibiting paraventricular CRH neurons, while low‐calorie foods do not appear to activate dopaminergic pathways [26, 27].

It is important to recognize that some residents respond to stress with reduced food intake. One hypothesis for this bidirectional response involves the distinction between acute and chronic stress. Acute stress activates the SAM axis, leading to the release of noradrenaline, an anorexigenic. In contrast, chronic stress is associated with hyperactivation of the HPA axis, resulting in the release of CRH and cortisol [28]. Under acute stress, CRH downregulates AgRP thus suppresses appetite. Under prolonged or less immediate stress, glucocorticoids can directly stimulate appetite [28].

4.2. Psychological Pathway

The optimal foraging theory proposes that organisms tend to overeat when food is abundant to compensate for times when food is scarce to ensure survival [10]. This theory is used to explain findings from studies in which individuals demonstrated stronger motivational responses to visual stimuli depicting high‐calorie food compared to low‐calorie foods [10]. Over time, these impressions may lead to conditioned associations between high‐calorie foods and stress relief, reinforcing behaviors of not only overeating but also overeating unhealthy foods. Another psychological concept is sensory‐specific satiety which describes the phenomenon in which a decrease in pleasantness of a particular food develops when eaten to satiety [12]. Attenuated sensory‐specific satiety is believed to promote overeating by facilitating larger meal consumption and be associated with chronic stress [12]. While numerous studies have linked stress to increased consumption of foods high in fat, sugar and salt, the type of food preference is influenced by multiple factors [10, 27, 29, 30]. It is reductive to evaluate the mechanism of stress eating solely from these perspectives; however, such an approach remains valuable for identifying areas for intervention (Figure 5).

FIGURE 5.

Physiologic and psychologic pathways underlying increased eating due to stressors.

5. Limitations

Several limitations should be considered when interpreting the findings of this study. First, the small sample size reduced the statistical power needed to detect significant associations, thus limiting the external validity of the results and increasing the risk of type II error. The potential for survey fatigue among residents reduced participation rates, thereby limiting the ability to collect sufficient data for matched analysis. Participants were surveyed from a single site with predominantly more junior residents on call, thus limiting the generalizability of this study. The study took place over 61 days, making it difficult to infer the longitudinal effects of the interventions on changing the participants' eating habits over the course of their residency.

The main goal of this study was to demonstrate whether residents could be encouraged to eat less unhealthy food. The approach attempted to intervene at different pathways (Figure 6). Box breathing and/or drinking green tea were chosen interventions. However, by introducing all interventions simultaneously, the effects of each are confounded. Moreover, the study relied on self‐reported unhealthy eating behavior which is subject to recall and reporting bias. The binary outcome of whether unhealthy food was consumed classifies all qualifying unhealthy food consumption equally. Consequently, not only is self‐reported data against nutritional thresholds unreliable, assessment of associated dietary behaviors in this study lacks granularity. A pre/post study design also evaluates associations with the intervention period; causation cannot be established.

FIGURE 6.

Proposed interventions to protect against eating unhealthy foods.

For future research, larger sample sizes and a longer follow‐up duration are recommended to yield more robust and generalizable results. Enhanced subject recruitment and participation would support matched analyses, allowing for more rigorous adjustment of confounding variables. Adjunct studies may focus on developing methodologies that collect resident data effectively while minimizing survey fatigue. If possible, measurement of cortisol testing of participants would provide a more objective assessment of baseline stress and the effect of calculated interventions. These findings support the value of further studies to better understand and more clearly demonstrate the association of simple interventions to reduce unhealthy eating patterns.

6. Conclusion

Taking call is a universal part of our American healthcare system. Residents across the country do the brunt of this work leading to poor sleep, high stress, and the hard‐to‐resist desire to eat unhealthy foods. Over the course of a residency, these dietary choices, along with perpetual sleep deprivation and chronic stress, are likely setting up our physicians of tomorrow with chronic health problems. This paper was a pilot study which sought to further explore the complex and interrelated physiological and psychological issues that lead to eating unhealthy food while on call. While the small sample size limited statistical power and reduced the ability to detect meaningful changes or associations between box breathing/drinking green tea and unhealthy eating episodes leading to insignificant results, the study calls for further research in this area. We highlight an opportunity to further develop the well‐being of residents. This area of research is important as there are chronic consequences of developing unhealthy eating patterns. This paper offers some practical, low‐cost solutions that are likely to work symbiotically to improve the dietary choices of residents on call and provides a framework for additional studies.

Funding

The authors have nothing to report.

Disclosure

The authors did not use artificial intelligence in this manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Appendix S1: Unhealthy eating patterns: pre data collection.

Appendix S2: Unhealthy eating patterns: post data collection.

Luk H. G. and Pine H., “Interventions for Unhealthy Eating Patterns of Resident Otolaryngologists on Call,” Laryngoscope Investigative Otolaryngology 11, no. 4 (2026): e70538, 10.1002/lio2.70538.

Data Availability Statement

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