Loneliness has been associated with several adverse health outcomes, with the largest effects on mental health outcomes and overall well-being [1]. This study explored how often medical students, residents, fellows, and faculty physicians who completed a suicide screening questionnaire at a large US academic medical center endorsed intense loneliness and the relationships of loneliness with burnout, depression, and suicidal thoughts and behaviors.

Methods

Data were obtained from a voluntary, anonymous online screening and referral tool used to connect respondents in distress with counselors of the Healer Education Assessment and Referral program and additional mental health treatment. Medical students, trainees, and faculty physicians at a large academic health care institution are invited by email to complete the survey approximately twice per year and can also complete the survey anytime through the widely disseminated online site. Responses between May 2009 through September 2021 were used.

The screening and referral tool includes the Stress and Depression Questionnaire from the American Foundation for Suicide Prevention’s Interactive Screening Program (ISP) [2]. The ISP questionnaire is a screening instrument, not a validated research tool. The questionnaire includes measures of intense emotional distress that have been linked to depression with suicidal ideation [3], a modified version of the Patient Health Questionnaire (PHQ-9) [4], and current suicidal thoughts, behaviors, plans and past suicide attempts. In 2017, three items related to burnout were added. Only basic sociodemographic data, such as gender, age, and academic position (e.g., student, resident, or faculty) are requested to maximize anonymity and participation. Respondents were not asked their medical specialty. For all items, excluding questions about depression and lifetime history of suicide, a 4-point scale was used to rank the frequency. Item response choices were 0 = not at all, 1 = some of the time, 2 = a lot of the time, and 3 = most or all the time. Participants were asked, “During the last 2 weeks, how often have you been bothered by feeling intensely lonely?” Participants were categorized as feeling “intensely lonely” if they responded with choice 2 or 3.

Derived from the Affective State Questionnaire of the ISP [2, 3], participants rated the frequency of the following states in the preceding 4 weeks: feeling nervous or worrying a lot, becoming easily annoyed or irritable, feeling your life is too stressful, having arguments or fights, feeling intensely anxious or having anxiety attacks, feeling intensely lonely, feeling intensely angry, feeling hopeless, feeling desperate, and feeling out of control. A response of 2 or 3 was considered positive.

Three questions related to burnout were added to the survey in 2017. One was a global measure similar to other single-item measures often used to screen for burnout [5], “feeling burned out from your work,” and the other two covered the burnout dimensions of exhaustion and depersonalization addressed in the Maslach Burnout Inventory: “feeling emotionally drained from work” and “having become more callous toward people since you took this job” [6]. A score of 2 or 3 was considered positive.

Depression severity in the prior 2 weeks was evaluated using a modified version of the PHQ-9 [4]. Participants were classified as having no or minimal depression (PHQ-9 score of 0–4), mild depression (score of 5–9), or moderate to severe depression (score of 10–27). A score in the moderate to severe range was defined as screening positive for depression.

Respondents rated the frequency of the following in the preceding 2 weeks: “Have you…had thoughts about taking your own life?” “…planned ways of taking your own life?” “…done things to hurt yourself?” A positive response was an answer  ≥ 1. Participants were also asked to reply yes or no about a lifetime history of any past suicide attempt.

Contingency tables were generated with Chi-square coefficients to explore the association of loneliness with sociodemographic characteristics and the binary outcomes of interest (burnout, depression, intense negative emotions, suicidality). The data are presented as odds ratios (OR) with 95% confidence intervals (CI). Statistical significance was defined as two-sided p-value  < 0.05.

Analysis of variance analysis (ANOVA) and t-test of independent samples was used to compare means. The correlation analysis was performed to explore the relationship between loneliness and an individual’s age, score for burnout, depression, and negative emotions. The Spearman correlation coefficients (r) were reported due to non-normal distribution of variables, as determined by the Shapiro–Wilk test. IBM SPSS Statistics 28.0.1.0 software (Armonk, NY) was used for all analyses.

Results

Of the approximately 10,000 students, trainees, and faculty physicians enrolled at the institution during the study period, 2,016 (~ 20%) completed the screening tool. The mean age of respondents was 34.30 ± 11.33 years. The majority identified as women (n = 1176, 58.3%) and White (n = 1076, 53.4%).

Among respondents, 18.0% met the criteria for intense loneliness (Table 1), and 21.0% of medical students, 24.7% of residents, 22.2% of fellows, and 9.0% of faculty physicians reported feeling lonely a lot of the time or most of the time. Overall, there was a statistically significant difference in mean loneliness scores between these cohorts (ANOVA F = 40.378, p < 0.001), with faculty physicians reporting significantly less intense loneliness than medical students (t = 9.294, p < 0.001), residents (t = 10.256, p < 0.001), and fellows (t = 6.098, p < 0.001). There were no significant differences in mean loneliness scores of medical students compared to residents, medical students compared to fellows, or residents compared to fellows.

Table 1 Loneliness severity among survey respondents by medical background
Full size table

Table 2 shows that intense loneliness was more likely to be reported by individuals under age 40 (OR = 2.899, p < 0.001, χ2 = 32.837), women (OR = 1.298, p = 0.030, χ2 = 4.778), and non-Whites (OR = 1.700, p < 0.001, χ2 = 10.819). Individuals who reported feeling intensely lonely were more likely to have a position other than faculty physician (OR = 0.341, p < 0.001, χ2 = 54.84).

Table 2 The association of loneliness with individuals’ sociodemographic data, self-reported burnout, depression, and suicidality
Full size table

Lonely individuals were more likely to feel “burned out from work” (OR = 3.500, p < 0.001, χ2 = 31.709), “more callous toward people” (OR = 3.565, p < 0.001, χ2 = 33.994), and “emotionally drained from work” (OR = 3.290, p < 0.001, χ2 = 29.598) than their non-lonely counterparts. They were more likely to have moderate to severe depression (OR = 12.336, p < 0.001, χ2 = 457.929), as well as report feeling nervous or worrying a lot (OR = 4.740, p < 0.001, χ2 = 165.976), easily annoyed or irritable (OR = 4.342, p < 0.001, χ2 = 161.719), intensely stressed (OR = 5.292, p < 0.001, χ2 = 182.065), argumentative or combative (OR = 4.850, p < 0.001, χ2 = 115.387), anxious (OR = 5.493, p < 0.001, χ2 = 176.200), intensely angry (OR = 8.338, p < 0.001, χ2 = 181.252), and out of control (OR = 6.260, p < 0.001, χ2 = 190.506).

Lonely individuals were more likely to report recent suicidal thoughts (OR = 7.104, p < 0.001, χ2 = 173.038), plans (OR = 12.958, p < 0.001, χ2 = 95.575), and behaviors (OR = 6.403, p < 0.001, χ2 = 27.913), as well as lifetime histories of suicide attempts (OR = 3.467, p < 0.001, χ2 = 21.101), compared to non-lonely respondents.

Discussion

The prevalence of loneliness was particularly high among medical students, residents, and fellows compared with faculty physicians. To our knowledge, our study is the first to report the prevalence of loneliness among U.S. medical students and residents. Previous studies that used different scales and cutoff points to measure loneliness have reported higher rates of loneliness among U.S. physicians (43.0% to 44.9%) [7, 8]. If we included respondents who experienced loneliness “some of the time” in the category of “intensely lonely” individuals, we would have found the prevalence of loneliness to be 31% among faculty physicians, closer to prior findings.

The observed lower prevalence of loneliness among clinical faculty physicians may be due to several factors. First, senior physicians may be more acclimatized to the health care environment and skilled in achieving work-life integration. Second, practicing physicians may have more time available to spend with their family and friends than resident physicians, who work long hours during their years in training. Furthermore, a greater number of faculty physicians may have spouses and children than younger trainees, mitigating loneliness. Although clinical faculty physicians had relatively lower rates of loneliness within our cohort, overall, our findings demonstrate that a sense of social disconnection can be an issue throughout all phases of a medical career.

Several sociodemographic factors were associated with loneliness: self-identifying as a woman, non-White, and being of younger age. Ofei-Dodoo et al. [7] similarly reported more severe loneliness among female versus male physicians. The gender imbalance in loneliness rates may be related to female physicians facing unique challenges related to balancing a demanding career and time with family. To our knowledge, racial differences in loneliness among health care professionals have not been studied; however, our findings are consistent with data showing higher loneliness rates in racial/ethnic minorities in the general population [9]. Physicians, trainees, and other health care workers of color continue to experience the effects of racial injustice due to longstanding structural inequities and may also face interpersonal racism in the workplace or learning environment, including implicit bias and microaggressions [10].

Few studies have looked at the relationship of loneliness with burnout among physicians. In a study of 124 internal medicine residents, Shapiro et al. [11] found that individuals with significant burnout were more likely to report feeling lonely than those without significant burnout. Karaoglu et al. [12] found a positive correlation between burnout and loneliness in pediatric residents.

Severity of depression was robustly associated with loneliness. Since this was a cross-sectional study, we could not draw causal conclusions about whether loneliness is a risk factor for depression, as several previous studies have suggested, or a consequence of depression [13, 14]. We can say with some certainty, however, that the presence of loneliness is a red flag for underlying depression and that it may be a modifiable risk for burnout, depression, and suicide risk. We recommend targeting loneliness and social isolation through wellness initiatives designed with these risk factors in mind.

Suicide is the leading cause of death among male residents and the second leading cause among female residents [15]. Female physicians have higher rates of suicide than females in the general population [16]. We are unaware of any prior studies linking loneliness to suicide risk in medical students, residents, or faculty physicians. Thus, our novel and alarming finding that lonely medical students, residents, fellows, and clinical faculty physicians were 13 times more likely to plan ways to harm themselves and 6.4 times more likely to harm than their less-lonely counterparts bears close attention and further study. In the meantime, we urge all clinicians to include loneliness and social isolation in suicide screening efforts and to consider loneliness an important target for suicide prevention strategies.

Our findings are relevant not only to individual physicians but also to medical education programs and health care institutions. Other authors have discussed the structural, professional, and social types of isolation that can contribute to physician distress and loneliness [17]. The physical fragmentation of medical offices and increased use of telehealth services, especially during the COVID-19 pandemic, are examples of how changes in the workplace can contribute to feelings of isolation among healthcare professionals. Our findings, taken together with prior work, suggest that interventions to maximize meaningful social connections within the hospital culture may be an actionable way that an institution can minimize loneliness among its employees and reduce their risk of other serious outcomes such as burnout, depression, and suicidality. Scheduling recurring group meetings for participants to discuss work-related challenges and work-life integration, which have been done through Balint and Schwartz Rounds, can strengthen meaningful connectedness among health care professionals [18, 19]. Other potential interventions are brief resilience exercises with mental health care experts; supervision among classmates, colleagues, and work unit and multidisciplinary teams; peer support groups for junior medical professionals and trainees or women in health care; physician lounges; and routine social events.

Several methodological limitations are important to consider. First, the data for this study all came from the ISP questionnaire, which is not a validated research instrument but, rather, a screening and referral tool. For this reason, we relied on a single question to assess loneliness and other screening measures to evaluate burnout and other mental health states, rather than empirically validated research scales. Further, loneliness was not explicitly defined in the questionnaire, and as a result, the term may have been interpreted differently by respondents. We did not control for factors such as depression when evaluating the association between loneliness and other outcomes. It would be interesting for future studies to explore whether loneliness or depression drives the association with suicidal ideation, among other associations. Due to the limitations of our dataset, we could not control for other variables that may have influenced our results, such as psychiatric history, personality characteristics, substance use, year of training, or specialty. Further, our analysis was based entirely on self-reported instruments, which introduces recall and social desirability biases.

Our sample was not fully representative because respondents came from a single academic medical institution and only a small proportion of eligible students, faculty, and trainees completed the screening questionnaire. It is likely the sample was biased toward more distressed individuals. Faculty physicians were relatively underrepresented than other study cohorts, which may be explained by faculty struggling less with the studied outcomes, feeling too overwhelmed to complete the questionnaire, and/or feeling less comfortable navigating to the online survey. The study included both pre- and post-COVID cohorts without considering differences over time. Both burnout and loneliness have increased since 2017 [20].

Interesting future research directions would be to explore the nuanced differences in how loneliness relates to burnout, depression, suicidal thoughts, and other indictors of distress since the onset of COVID, to design and implement hypothesis-driven, longitudinal studies that might allow for investigation of causal relationships, and to assess the effectiveness and impact of interventions designed to mitigate loneliness.

In summary, a large percentage of medical students, physician trainees, and faculty physicians reported feeling intensely lonely, and loneliness was significantly associated with burnout, depression, and suicidal thoughts and behaviors. Institutions seeking to improve the well-being of their medical students, trainees and attending physicians must acknowledge the importance of and implement interventions to optimize social connection.