Incidence, Nature, and Consequences of Oncologists' Experiences With Sexual Harassment.

PURPOSE: The incidence and impact of workplace sexual harassment (SH) of oncologists requires rigorous characterization.
METHODS: Oncologists identified by ASCO's Research Survey Pool and social media outreach completed validated measures of SH (encompassing gender harassment, unwanted sexual attention, and sexual coercion) and four outcomes (mental health, job satisfaction, turnover intentions, and sense of workplace safety) over the previous year. Multivariable regression models assess the impact of SH on the four outcomes.
RESULTS: Of 271 cisgender respondents (153 women and 118 men), 189 (70%) experienced SH in the past year alone by peers and/or superiors (80% of women v 56% of men, P < .0001). Specifically, 186 (69%) experienced gender harassment (79% of women, 55% of men, P < .0001), 45 (17%) unwanted sexual attention (22% of women, 9% of men, P = .005), and 7 (3%) sexual coercion (3% of women, 2% of men, P = .42). SH by patients and/or families in the past year was experienced by 143 (53% overall: 67% of women, 35% of men, P < .0001). Specifically, 141 (52%) experienced gender harassment (66% of women, 34% of men, P < .0001), 15 (6%) unwanted sexual attention (5% of women, 6% of men, P = .80), and 3 (1%) sexual coercion (1% of women, 1% of men, P = .72). Multivariable analysis showed that past-year SH by peers and/or superiors was significantly associated with decreased mental health (β = -0.45, P = .004), sense of workplace safety (β = -0.98, P < .001), and job satisfaction (β = -0.69, P = .001), along with increased turnover intentions (β = 0.93, P < .0001). Past-year SH by patients and/or families was significantly associated with decreased mental health (β = -0.41, P = .002), sense of workplace safety (β = -0.42, P = .014), and increased turnover intentions (β = 0.58, P = .0004). There were no significant interactions between the respondents' gender and the SH scores in any of the four outcome models, signifying no difference in impact between men and women oncologists.
CONCLUSION: This study using validated measures of SH to systematically characterize oncologists' workplace experience demonstrates substantial incidence of SH in the previous one year alone and its impact on men and women oncologists, informing the need for and design of effective protective and preventive measures.

INTRODUCTION

Broad cultural movements focused on sexual harassment (SH) such as #metoo and #TIMESUP have reached medicine, where harassment and its impact on physician well-being and professional outcomes are increasingly recognized.[1-3] This motivates efforts to characterize the scope, nature, and impact of experiences with SH in oncology with the same rigor as in other fields.[4,5] Clinical oncology encompasses diverse clinicians from various practice settings, cultural backgrounds, and subspecialties.[6,7] Understanding exactly what happens, where, when, and to whom is essential to inform efforts to transform culture and eradicate problematic behaviors.

CONTEXT

Key Objective

Sexual harassment (SH) in the workplace of clinical oncologists remains to be fully characterized. Here, we conducted a prospective cross-sectional study of clinical oncologists in the United States using a survey with rigorous measures of SH (encompassing gender harassment, unwanted sexual attention, and sexual coercion) over the previous year alone and four outcomes, specifically mental health, job satisfaction, turnover intentions, and sense of workplace safety.

Knowledge Generated

The majority of the 271 respondents report one or more incidents of SH in the past year by institutional peers and/or superiors (70%, n = 189) and by patients and/or families (53% overall, n = 143), with more women oncologists being affected than men. Experiencing SH in the past year was associated with a negative impact on mental health, job satisfaction, and turnover intentions among both men and women respondents.

Relevance

This study systematically characterizes oncologists' experience of workplace SH and demonstrates substantial incidence over one year and impact, informing the need for and design of effective protective and preventive measures.

To standardize the study of SH, organizational psychologists developed the Sexual Experiences Questionnaire (SEQ), an extensively validated, behaviorally based survey instrument.[8,9] The SEQ captures all three dimensions of SH identified by social scientists: gender harassment, unwanted sexual attention, and sexual coercion (Fig 1).[10] Gender harassment includes verbal and nonverbal behaviors conveying hostility to, objectification of, exclusion of, or second-class status about one gender.[3] Unwanted sexual attention describes unwanted sexual advances, including unwanted touches or attempts to establish a sexual relationship despite discouragement. Finally, sexual coercion involves making job-related threats or promising job-related benefits to coerce compliance with sexual demands.[11] To evaluate behaviors perpetrated by members of health care organizations and by patients and families, investigators developed an SEQ version appropriate for physicians.[12]

FIG 1.

Defining the three subtypes of sexual harassment. This figure provides definitions of the three forms of sexual harassment that have been described by social scientists[3] and are measured by the Sexual Experiences Questionnaire instrument used in the present study.

Although previous studies have explored SH in cancer medicine, few to our awareness in oncology use comprehensive validated measures to investigate the experiences of oncologists.[13-15] To that end, we sought to rigorously evaluate oncologists' lived experiences and consequences of SH perpetrated by both institutional insiders (peers and/or superiors) and patients and/or families in a cross-sectional survey.

METHODS

Study Sample and Survey Administration

After approval by the University of Michigan institutional review board, we partnered with ASCO to send this survey study from September to November 2020 to 1,000 randomly selected members of ASCO's voluntary opt-in Research Survey Pool (RSP), who met the eligibility criteria: physicians (attending or in training), working full-time at their current institution for at least 1 year, practicing a clinical oncologic subspecialty (adult or pediatric hematology and/or medical oncology, surgical oncology, gynecologic oncology, or radiation oncology). Social media outreach through Twitter and Facebook's Hematology-Oncology Women Physicians Group also solicited oncologists who met the above eligibility criteria to participate. Participants clicked the link to the survey platform where upon consent, they authenticated their ASCO membership to confirm eligibility and avoid duplicate responses. Once authenticated, participants were immediately deidentified and taken to the survey. Weekly reminders to participate were sent through the ASCO RSP for 5 weeks. To mitigate response bias, recruitment outreach made no specific mention of SH; eligible participants received a nonspecific invitation to a study titled “Workplace Experience of Oncologists” to examine the work environment experienced by oncologists.

Study Outcomes

The primary objective was to characterize by gender the prevalence of recent (past year) SH of physicians practicing oncology, including that perpetrated by institutional insiders (peers and/or superiors) and by patients and/or families, by type (gender harassment, unwanted sexual attention, and sexual coercion). Secondary objectives included evaluating associations between experiences of harassment and consequences for the respondent (including measures of mental health, job satisfaction, sense of safety at work, and turnover intentions).

Survey Instruments

Consistent with best practices in survey design, participants completed the survey of demographics and questions on constructs of interest, using previously validated instruments where available, with verbiage to specify clinical oncology (Data Supplement, online only).[16] Participants were reminded throughout the survey to only respond about unwanted behaviors in the previous one year.

SH was measured using the 20-item SEQ, a self-reported behaviorally based inventory of three different SH types: gender harassment, unwanted sexual attention, and sexual coercion.[11,12,17] Respondents indicated on a four-point scale the frequency of experiences with unwanted behaviors in the previous year (0 = never, 1 = once or twice, 2 = sometimes, 3 = often, and 4 = many times), with higher scores signifying more episodes of SH. The SEQ items were presented twice, first to capture experiences perpetrated by institutional insiders (peers and/or superiors), followed by select items to capture experiences perpetrated by patients and/or families as previously described.[12] Binary indicators for the experience of any overall harassment and its subtypes were created to reflect if at least one component question was answered affirmatively.

Mental health was captured through the five-item Mental Health Index-5, a widely used screening instrument developed and validated from the Medical Outcomes Study 36-Item Short Form Health Survey.[18,19] Using a five-point scale (1 = never and 5 = always), respondents indicated the extent to which they agree with five statements about symptoms of anxiety (“been a very nervous person”) and depression (eg, “felt downhearted and blue”), with a higher average signifying better mental health.

Job satisfaction was measured using two items from the Michigan Organizational Assessment Questionnaire.[20,21] Using a five-point scale (1 = strongly disagree and 5 = strongly agree), participants indicate the extent to which they agree or disagree with two job-specific statements: “All in all, I am satisfied with my job” and “In general, I don't like my job.” A higher average sum of two items indicates greater job satisfaction.

Participants' sense of safety at work was assessed on a five-point scale (1 = strongly disagree and 5 = strongly agree) for one item, adapted from the study by Clancy et al on the extent to which they agreed with the statement “I feel safe at my workplace.”[22]

Turnover intentions measured the respondents' intentions to quit their job using three items designed for clinical medicine on desire to change the field, move to a different institution, and/or leave the medical field.[12] Turnover intention items were measured only for respondents not in training. Item responses were scored on a five-point scale (1 = never and 5 = always), standardized (z-scored), and averaged where higher values indicate greater turnover intentions.

Statistical Considerations

After removing substantially incomplete responses, we narrowed the analytic sample to cisgender participants, given the small number of respondents (n = 2) reporting a noncisgender identity. Descriptive statistics summarized the respondents' demographics, experience, and practice characteristics. We analyzed the incidence of any harassment for SEQ-Insider, SEQ-Patient, and each subdimension (1 = experiencing at least one unwanted behavior in the past year and 0 = experiencing no unwanted behaviors in the past year). We analyzed whether the incidence differed significantly by gender, subspecialty (medical hematology-oncology and others [gynecologic, surgical, and radiation oncology combined]), and career stage (early career [physician currently in or within 5 years of completing an oncologic residency or fellowship; < 5 years of experience], midcareer [5 to < 15 years], and senior [15+ years]). Then, separate multivariable linear regression models examined the adjusted association between the experience of SH score from either institutional insiders or patients and/or families and physician outcomes: mental health, job satisfaction, sense of safety at work, and turnover intentions. Adjustment covariates included ethnicity (White, under-represented minorities [URM; Black, Hispanic], and Asian or Pacific Islander), career stage (early career, midcareer, and senior), and specialty (Adult or Pediatric Heme and/or Med Onc v Surg, Gyn, Rad Onc, or Multiple). We evaluated pairwise interactions between gender and harassment scores. All statistical analyses were conducted using the SAS System version 9.4 (Cary, NC).

RESULTS

Participant Demographics

In total, 304 practicing oncologists including 238 of the 1,000 targeted through the ASCO RSP and 66 through social media outreach, all authenticated their identity with confirmation of their eligibility to participate and subsequently accessed the survey link. Of these, 273 provided responses (215 [79%] via RSP and 58 [21%] via social media). Appendix Table A1 (online only) compares the characteristics of the RSP vs social media respondents; of note, there were no differences in rates of reporting SH between RSP respondents and social media participants.

TABLE A1.

Sample Demographics by Survey Type Completed

Among the 1,000 RSP invitees, comparing demographics of the responders versus nonresponders shows that response rates were similar by gender, race and ethnicity, and geographic practice location. Response rates were significantly different for practice setting (with respondents more likely to be in academic settings and less likely in training programs) and age (with respondents less likely to be in the youngest age group and more likely to be in the middle age groups).

The respondents included 153 cisgender women and 118 cisgender men. One person was gender nonconforming, and one responded that none of the gender descriptors describe them. Overall, the 271 respondents who self-identified as cisgender formed the analytic data set (Table 1), of whom 153 (56%) were women and 118 (44%) were men. One hundred forty-four (53%) were White, and 95 (35%) were Asian or Pacific Islander, whereas 30 (11%) self-identified as African American or Hispanic. Two hundred fifty-six (94%) identified as heterosexual; 15 (6%) identified as a sexual or gender minority. Most respondents (n = 172, 63%) were born in the United States, and 237 (87%) held US citizenship. Of 271 respondents, 250 were physicians in practice and 21 were resident or fellow physicians; 168 (62%) practiced in academic settings, and 236 (87%) practiced medical oncology.

TABLE 1.

Characteristics of Oncology Physicians in the Analytic Data Set

Incidence of SH

Figure 2 and Appendix Table A2 (online only) detail the incidence of past-year SH and its three subtypes by both perpetrator types (institutional insiders and patients and/or families). Overall, 189 respondents (70%), including 80% of women and 56% of men (P < .0001), endorsed past-year SH by institutional insiders. Gender harassment was indicated by 79% of women and 55% of men (P < .0001), unwanted sexual attention by 22% of women and 9% of men (P = .005), and sexual coercion by 3% of women and 2% of men (P = .42). Similarly, past-year SH by patients and/or families was endorsed by 143 (53%) oncologists overall, including 67% of women and 35% of men oncologists (P < .0001). Specifically, patient- and/or family-perpetrated gender harassment was indicated by 66% of women and 34% of men (P < .0001), unwanted sexual attention by 5% of women and 6% of men (P = .84), and sexual coercion by 1% of women and 1% of men (P = .70).

FIG 2.

Incidence of SH in previous year alone by type and by respondent characteristics. (A) Proportion of respondents reporting at least one experience perpetrated by institutional insiders (peers and/or superiors) of any SH and then each SH subtype separately, by respondent gender, career stage, race and ethnicity, and specialty. (B) Proportion of respondents reporting at least one experience in the previous year alone perpetrated by patients and/or families of any SH and then each SH subtype separately, again by respondent gender, career stage, race and ethnicity, and specialty. NOTE. Six respondents did not answer the Patient-Family SEQ, one Insider SEQ, two did not answer their race and ethnicity. SH, sexual harassment.

TABLE A2.

Incidence of Any Sexual Harassment in the Previous Year, by Type and by Perpetrator

Downstream Impact of SH on Workplace Wellness

Table 2 details eight separate multivariable regression models examining the associations between past-year SH by institutional insiders and by patients and/or families, separately, and the four outcomes of interest (mental health, job satisfaction, turnover intentions, and sense of workplace safety), after controlling for respondent demographics. In the multivariable analyses, past-year SH by institutional insiders was significantly associated with decreased mental health (β = –0.45, P = .004), sense of workplace safety (β = –0.98, P < .001), and increased turnover intentions (β = 0.93, P < .0001). Similarly, past-year SH by patients and/or their families was significantly associated with decreased mental health (β = –0.41, P = .002), decreased sense of workplace safety (β = –0.42, P = .014), and increased turnover intentions (β = 0.58, P = .0004). Past-year SH by institutional insiders (β = –0.69, P = .001) but not patients (P = .21) was significantly associated with the respondents' job satisfaction. No significant interaction between gender and SEQ score for each outcome was found, suggesting that the SEQ-measured impact on the outcomes is similar by gender. The models that included SH by patients and/or families suggest lower job satisfaction (β = –0.24, P = .026) and sense of workplace safety (β = –0.24, P = .034) among women oncologists although gender was not significant in the models including SH by institutional insiders. None of the outcomes were otherwise significantly associated with physician gender, career stage, race and ethnicity, and/or oncologic subspecialty.

TABLE 2.

Regression Results Predicting Mental Health, Job Satisfaction, Sense of Workplace Safety, and Turnover Intentions Among Physician Respondents

DISCUSSION

To our knowledge, this is the first study in oncology to use validated measures of SH to systematically characterize the incidence and nature of past-year SH experienced by a diverse sample of oncologists, using best practices as recommended by the National Academies of Sciences, Engineering, and Medicine (NASEM).[3] The SEQ scoring–based finding of past-year SH by peers and/or superiors was significantly associated with decreased mental health, lower job satisfaction, less workplace safety, and higher turnover intention, with a similar significant impact on all outcomes but job satisfaction for SEQ scoring–based finding of SH by patients and/or families. Although the incidence of any SH between men and women physicians was significantly different, the downstream impact was not. No significant interactions existed between the respondents' gender and the impact of SH for any outcome. These findings demonstrate the impact of SH on men and women oncologists in multiple domains of workplace experience. This study provides critical data to inform the need for and design of effective protective and preventive workplace policies in oncology.

NASEM's landmark 2018 report on SH showed that female medical students were 220% more likely than students in non-science, technology, engineering, and mathematics disciplines to experience SH. By far, the most common form of harassment is the sexist remarks and crude behaviors that constitute gender harassment. Indeed, NASEM disseminated the metaphor of SH as an iceberg, whereby much attention focuses on the rare egregious cases of unwanted sexual attention and sexual coercion, but the bulk of behaviors lurk beneath the surface in the form of gender harassment. The few studies that have used sensitive validated measures to evaluate physicians' experiences of workplace SH have shown strikingly high rates of past-year harassment, with the overwhelmingly most common form being gender harassment.[4,12,14,23] Disappointingly, the present study provides compelling evidence that the rates of gender harassment in oncology are substantial, experienced by the majority of not only women but also men studied here. Thus, although attention is often focused on shocking individual cases of sexual coercion, the data suggest that the problem of SH is both more widespread and nuanced in its manifestations than commonly recognized.[24,25]

Experiences with gender harassment are not without impact, as further illustrated by our findings. Behavior need not be sexually predatory to derogate, demean, or humiliate individuals on the basis of sex in a way that has consequences for their well-being.[26] Indeed, consistent with decades of research in organizational psychology that has demonstrated clear associations between experiences of workplace SH and the physical, mental, and professional well-being of workers, we found significant associations between past-year experiences of SH and the four outcomes that we measured: mental health, job satisfaction, turnover intentions, and sense of workplace safety.[27,28]

To address workplace SH, studies to demonstrate what behaviors are currently occurring, to whom, by whom, and with what effect, within a specific field, can provide a powerful call to action. Our findings that women are more likely to experience harassment than men, but that men are also frequently experiencing unwanted behaviors, are important to both motivate and guide changes to policy and practice.

To date, although documents exist to guide entities in developing policies and procedures for misconduct in the workplace, no nationally standardized, widely adopted strategies to counter SH across the United States exist.[29] Instead, individual institutions, health systems, and other entities develop their own approaches and policies with limited reporting on impact and outcomes. Common unifying themes of such programs center on creating a safe workplace culture with accountability, robust institutional policies specifically covering SH, continuous education efforts of local stakeholders to ensure broad awareness of these policies, robust reporting processes, and supportive services for those affected by SH with an emphasis on ensuring continuous enforcement of policies and standard governing behaviors in the workplace as it pertains to employees as well as patients and families.[30]

Through our findings, we recognize not only the implications of these behaviors on the workplace experience of oncologists but also the broader impact of these behaviors on other members of the oncologic workforce, whose experience was outside of the scope of this study. We hope that our study findings will lead to evidence-based intervention, including innovative approaches to foster cultural transformation through the cultivation of civility and respect, allyship, and empowerment of bystanders.[2,31,32] Findings that highlight the role of patients and families as perpetrators can further guide efforts to include strategies such as development of patient rights and responsibilities statements to address this common challenge.[33-36]

Limitations in interpretation of our data include a modest response rate among those who were invited to participate. We took care not to advertise the specific subject of the investigation in our request for participation, to avoid demand effects and selection bias, and we are reassured that the rates of response to our survey were similar to other studies conducted using the ASCO RSP (and were among the highest response rates for surveys fielded by the RSP during the disruptions that occurred in 2020). Nevertheless, there is a risk that the incidence estimates might not be generalizable; this is less of a concern for the associations observed. It is challenging to quantify whether our respondents are representative of the national pool of clinical oncologists, given limited contemporaneous data characterizing the cross-disciplinary US oncologic workforce during the survey period and that our study is limited to physicians in an oncology subspecialty in full-time practice at their current institution for at least a year. Compared with published data characterizing the general oncology workforce at large, there may be higher representation of certain demographic groups among our study's RSP invitees, who were randomly selected members of the voluntary RSP pool meeting our study eligibility criteria.[37,38] This includes a higher proportion of academic practitioners and those identifying as Asian in the RSP sample who received our study invitations (50% in academia and 25% self-identified as Asian in their ASCO membership profile per the organization's definition) and, in turn, in our respondents (62% academia and 35% Asian or Pacific Islander per our methodology). Given that a portion of our outreach was through social media that included women physicians' sites, we characterized the demographics of our participants by the recruitment approach; we reassuringly found that, among those who were targeted by the ASCO RSP, the respondents were similar to nonrespondents by gender, and those who responded after social media outreach were not more likely to report harassment than those who responded via the RSP. Therefore, given the extremely high incidence rates, this study provides compelling evidence of a problem of sufficient magnitude to warrant action. Even if very few nonrespondents experienced harassment in the past year, the fact that such a large majority of respondents did would mean that the underlying population rate of harassment is unacceptably high. Furthermore, the limited participation from each oncologic subspecialty other than medical oncology, oncologic physicians in training, and noncisgender respondents precluded any separate analyses into these groups' unique experiences. Our findings call for further study of the experiences of oncologists facing the challenges of intersectional membership in multiple historically marginalized groups. Future research would also be valuable to integrate the understanding of how SH experiences may relate to the strikingly high levels of burnout documented in oncology.

Given the limited data characterizing the nature and scope of SH in oncology, this study presents critical data to inform effective policies to protect the oncology workforce that provides care and produces research that serves patients and society.