Gender disparities in medical care have recently received increased attention. Gender bias inherent to the instructional materials used by medical students is one of many factors implicated in contributing to inequities in medical treatment (1—5).

Gender bias, in the form of role or social stereotypes, has been widely noted in instructional materials used in primary grades through college (6—8). Exposure to biased instructional materials increases gender-biased attitudes, particularly among children (9—11). Title IX of the 1972 Education Amendments Act has been instrumental in reducing stereotypic representation of men and women in educational materials used in federally funded primary and secondary educational institutions (8,12).

Naturally, as regards training of medical students, there are concerns that gender-biased instructional materials can impart values that are later incorporated into clinical practice. After all, a great deal of a medical student's time is spent reviewing such texts. Gender bias can be incorporated into instructional materials in several ways. In addition to stereotypic presentations of men and women, bias can be conveyed in the frequencies that men and women are portrayed in case scenarios or as models, or in the amount of attention directed to disorders more common to men vs. women.

Previous investigations of gender bias in anatomy or physical diagnosis texts have noted that illustrations using male models far exceeded those using female models (1,3). Consequently, investigators have suggested that medical students may be less familiar with female anatomy, and are at risk for becoming biased in perceiving the male body as the "norm" (3,4). The present study seeks to use a similar paradigm to examine whether bias is present in psychiatric texts. While introductory psychiatric textbooks do not typically emphasize anatomic illustrations, authors of psychiatric texts use case vignettes to illustrate psychopathology or management of psychiatric disorders. The frequencies of case vignettes featuring male and female subjects were examined. Introductory level texts were selected, as these are more apt to be used by medical students in psychiatry courses and during clinical clerkships. To our knowledge, no such review of gender bias in psychiatric texts has been done before.

Five texts were selected for review of gender bias: Introductory Textbook of Psychiatry (13), Review of General Psychiatry (14), Clinical Manual of Psychiatric Diagnosis and Treatment—A Biopsychosocial Approach (15), DSM-IV Casebook—A Learning Companion to the Diagnostic and Statistical Manual of Mental Disorders (16), and Clinical Psychiatry for Medical Students (17). Text selection was based upon common usage for introductory level instruction as well as the presence of case vignettes illustrative of various DSM diagnoses. The most recent editions of the textbooks were used; publication dates were between 1994 and 1995. In addition, two earlier versions of the DSM Casebooks, corresponding to the DSM-III and DSM-III-R (18,19), were assessed for comparison to the most recent edition. We conducted our study in 1997.

The Casebooks consist of a series of case vignettes, each of which is followed by explanations of how psychiatric diagnoses were assigned. Case vignettes in the other textbooks were either offset from the surrounding text by a different print font or were designated by inscriptions such as "case vignette" or "illustrative case."

Each vignette was reviewed to determine the gender of the subject. If the gender of the subject was not discernible, the gender was said to be unspecified. Additionally, if the vignette featured a male and a female subject, for example, a couple presenting with relationship difficulties, then the gender was categorized as "both."

The primary diagnosis illustrated in each of the case vignettes was recorded. Because frequencies for specific psychiatric disorders were apt to be small, disorders illustrated in the vignettes were classified along broad diagnostic categories based on DSM-IV (T1). In addition to the broad categories, some specific disorders were also examined separately, including depression and other depressive-related disorders; schizophrenia; psychoses other than schizophrenia; antisocial and borderline personality disorders; and disorders unique to women, for example, postpartum disorders. The number of males and females in each of the diagnostic categories was tabulated. Not all of the texts used case vignettes illustrating each of these diagnoses. An individual vignette may illustrate more than one diagnosis. If more than one was provided, each of the diagnoses was recorded. "Rule out" diagnoses were not recorded, however. Vignettes demonstrating something other than a psychiatric disorder, for example, a defense mechanism or an aspect of psychotherapy, were recorded for purposes of tracking gender; no psychiatric diagnosis was assigned unless clearly delineated in the case illustration.

Comparisons of frequencies of males and females used in each text were subjected to chi-square analyses. Cases with no clear male or female protagonist, that is, classified as "both" or "unspecified," were omitted from statistical analyses because of the small sample size.

A total of 962 case vignettes were examined in the seven texts. Of the vignettes used in the DSM-III and the DSM-III-R Casebooks, 42% and 82%, respectively, were subsequently re-used in the DSM-IV Casebook. Consequently, vignettes in the earlier editions of the Casebook were omitted from composite analyses. Of the remaining 5 texts, 292 vignettes featured male protagonists, 192 featured females, 8 featured both males and females, and 1 featured a patient whose gender was unspecified. The numbers of vignettes with male, female, or both protagonists for each text are outlined in T2, along with χ² andprobability values for individual comparisons within each text. The difference observed between the numbers of vignettes featuring males and females combined across the five texts was significantly different (χ²=20.66, df=1, P<0.0001). In all but two texts, the frequencies of male and female protagonists were significantly different (13,17). In one text, the number of vignettes featuring females exceeded those featuring males (17).

In addition, there were 214 vignettes in the DSM-III Casebook and 255 in the DSM-III-R Casebook. Male subjects were featured in 126 (58.9%) and 142 (55.7%) cases, respectively; female subjects were featured in 84 (39.3%) and 109 (42.7%); and both males and females in 4 (1.8% and 1.6%, respectively). Significantly greater numbers of vignettes featured males than females in both texts: χ²=8.40, df=1, P=0.004 and χ²=4.34, df=1, P=0.037, respectively. Compared with the fourth edition of the Casebook, the percentages of female and male subjects remained essentially unchanged (cf. T2).

A total of 560 diagnostic categories were illustrated in the vignettes used in the five texts. The frequencies for these diagnostic categories subdivided by gender are summarized in T1. Cases featuring both male and female subjects and the one gender-neutral case accounted for nine of the diagnoses, but were not included in T1. Of all 493 vignettes examined, only two were illustrative of disorders unique to women's health, that is, postpartum disorders and premenstrual dysphoric disorder. As expected, female subjects tended to be represented in vignettes illustrating somatoform disorders (67.6%), dissociative disorders (62.5%), eating disorders (72.2%), and depressive disorders (50.9%). Females were also depicted in 71.4% of cases depicting factitious disorders and in 50% of cases depicting borderline personality. Otherwise, the remaining vignettes illustrating all other diagnoses featured men predominantly.

Of the clinical case vignettes examined in this study, a disproportionate number featured male protagonists. Significant differences in the use of male as compared with female protagonists were observed in all but two of the textbooks.

The preponderance of male subjects was not an artifact of the use of vignettes depicting psychiatric disorders with higher prevalence rates among males. Such disorders, for example, substance-related disorders and antisocial personality, merely accounted for 11% of the vignettes. Those depicting disorders with higher prevalence rates among women accounted for 17.8% of the vignettes reviewed. Interestingly, of all vignettes examined, only 2 (0.4%) dealt with disorders unique to women's health, specifically postpartum disorders and premenstrual dysphoric disorder.

Overall, females tended to be subjects of vignettes illustrating disorders with higher prevalence rates among women; and males were subjects in vignettes illustrating disorders with higher prevalence rates among men. In such cases, protagonist gender was consistent with known epidemiologic trends of psychiatric disorders. However, vignettes illustrating disorders in which the prevalence rates are unknown or are otherwise thought to be equal among men and women tended to feature male subjects. In fairness to the text authors, it is noteworthy that when one breaks down the vignettes in each individual text by diagnosis, the numbers become small, making it difficult to detect whether these trends exist within each individual text.

It is interesting that women are underrepresented in clinical vignettes in psychiatric texts despite their predominance in psychiatric practice (20). Women comprise more than half of the population (2), and more women than men seek medical or psychiatric attention (20).

Despite efforts to draw attention to the need for more gender-equitable educational materials and journal articles in the late 1980s and early 1990s (21,22), several of the texts reviewed herein, although published between 1994 and 1995, continue to display gender bias. While each of the three versions of the DSM Casebooks was updated to correspond to modifications made within the DSM, no significant changes were made with regard to gender bias.

In future editions of psychiatric texts, subject gender in case vignettes should parallel known epidemiologic trends or patterns of psychiatric service utilization. In this way, subject gender will be consistent with trends observed in the clinical setting. When these trends are unknown or unclear, a gender-equitable approach, as was used in two texts examined herein (13,17), can be a consideration.

The tendency to use the male as a chief character has been observed in a variety of texts, from grade school through college. Our study expands on work focusing on gender disparities in other disciplines taught to medical students, for example, anatomy and clinical medicine (1,3). Therefore, the trends observed in this study are not unique to psychiatry and may reflect broader social biases.

The differences in which physicians react to, evaluate, or treat complaints made by male and female patients may be, in part, attributable to prejudices and misrepresentations of disorders as manifested by women in the medical literature (5,23). Previous research has demonstrated that male patients receive more extensive evaluations and follow-up studies than female patients, despite presenting to physicians with identical complaints (24—29). Whether a patient's gender influences the assessment and treatment of their psychiatric condition warrants attention. Furthermore, how this might be communicated in psychiatric texts might be an avenue of further research. It is quite possible that readers may not attend to the gender of subjects in illustrative vignettes or to the frequencies with which male or female subjects are used. The impact of gender disparities, if any, on readers requires further investigation.

Nonetheless, students need to be instructed that gender can enrich our understanding of the individuals seeking treatment, providing a valuable source of epidemiologic and prognostic information. Training in psychiatry should emphasize how the presentation, course, and prognosis of psychiatric disorders may differ for men and women. It is hoped that medical trainees will consequently appreciate gender differences and will be attuned to potential gender biases in other aspects of medical training and, even, clinical research. Scientific study has often used male subjects and assumed generalizability to females (30,31). Ultimately, it is hoped that when men and women are treated differently by physicians, it is based upon an awareness of the aforementioned gender differences, and not faulty assumptions, attributions, or misperceptions about men and women conveyed, implicitly or explicitly, in medical school.

The authors thank Kim Hughes and Dianne Szczerba for their assistance with data collection and analysis and Linda F. Pessar, M.D., and Jennifer Batterman-Faunce, Ph.D., for editorial suggestions.