Introduction

Psychiatric disorders are the second leading cause of disability in the world [1], and despite expansion of evidence-based psychotherapeutic and pharmacological options, public health measures of mental health indicate that the suicide rate has risen by 33% in the past two decades and the rates of deaths of despair have risen dramatically [2, 3]. These trends have worsened in the past year, given the COVID-19 pandemic and the resulting public health and socioeconomic consequences: In June 2020, 40% of the US adult population reported at least one symptom of adverse mental or behavioral health condition [4]. Improved interventions are desperately needed for our patients and our society.

One way to improve psychiatric interventions is through standardized assessments and measurement-based care [5,6,7], as clinics that implement these systems consistently show that their patients experience better outcomes compared to usual care [27]. The PCARES registry itself was considered a clinical quality improvement project by the relevant Institutional Review Board (IRB). It did not meet the criteria for human subject research, therefore IRB approval was not required and the need for consent to participate was waived. However, any research project using the de-identified PCARES data, including this project, requires approval by the relevant IRB. The current study was conducted according to the ethical principles of the Declaration of Helsinki and approved by the Pennsylvania State University IRB (IRB #183 and #7926). All patients were evaluated by board-certified psychiatrists or licensed clinical psychologists. All patients were included in the registry regardless of psychiatric diagnosis or purpose of visit. The PCARES Registry includes self-report measures and select diagnostic and demographic information extracted from patients’ electronic medical records (EMR). Although previous studies using these data have been published (28,29,30,31), the current study’s aims are unique and have never been reported.

Demographics

Demographic data were extracted from the EMR. Demographic data included self-identified gender (male, female, patient declined, unavailable), race (American Indian/Alaskan Native, Asian, Black or African American, Native Hawaiian or Pacific Islander, White, two or more races, other race, patient declined, unavailable), ethnicity (Hispanic, Latino, or Spanish origin; not Hispanic, Latino, or Spanish origin; multiple; patient declined; unavailable), and marital status (single, married, and formerly married including divorced, separated, and widowed, patient declined, unavailable). Two steps were used to create participants’ socioeconomic profiles. First, participants’ home address/zip code were extracted from the EMR. Second, education and income at the zip code-level were obtained from the 2016 American Community Survey (ACS) five-year estimates database (32) so that we could identify the education and median income for each patient’s zip code. We also obtained the county level rural/urban profile from the Center for Rural PA, and classified patients into binary categories by county (rural/urban) and municipality (rural/urban). Insurance status was classified as follows: commercial insurance included preferred provider organizations (PPO), Blue Cross/Blue Shield related organizations, health maintenance organizations (HMO), and other commercial insurance payers (state-funded Medicaid payers, Medicare, or self-pay).

Clinical diagnoses

Clinical diagnoses were extracted from the first visit EMR documentation. Body mass index (BMI) was calculated from measured height and weight recorded in the EMR during the first visit.

Self-report measures

Participants completed a battery of self-report measures at the first visit. However, the battery was changed based on clinician feedback midway through data collection. Herein, we report first-visit data from the first and second versions of the battery, termed “battery 1” and “battery 2” (see Supplemental Table 1 for details of battery schedule). Self-report measures were available in the English language.

Trauma exposure

The Brief Trauma Questionnaire (BTQ) was used to report lifetime trauma exposure [33].

Transdiagnostic symptom measures

The DSM-5 Level 1 Cross-Cutting Symptom Measure [21, 22] is a self-report screening tool assessing depression, mania, anxiety, somatic symptoms, suicidal ideation, psychosis, sleep, memory, obsessions and compulsions, dissociation, personality styles, and substance use. If participants reported that anger, anxiety, depression, mania, obsessions and compulsions, sleep, or somatic symptoms bothered them more than slightly or rarely, the screener was determined to be positive. In battery 1, if the Level 1 screener was positive, the patient completed the Level 2 in-depth questionnaire for that symptom area. The DSM-5 Level 2 adult assessments included the PROMIS Emotional Distress, Depression Short Form; PROMIS Emotional Distress, Anger Short Form; PROMIS Emotional Distress, Anxiety Short Form; and PROMIS, Sleep Disturbance Short Form. Each of these Level 2 assessments was scored using T scores. The Personality Inventory for DSM-5-Brief Form (PID-5-BF) was used to assess personality traits and domains. The Columbia Suicidality Severity Rating Scale (CSSRS) was used to assess suicide risk [34].

In battery 2, regardless of whether patients screened positive for any Level 1 items, all patients completed the following measures. The Patient Health Questionnaire (PHQ-9) assessed depressive symptoms [35]. The Generalized Anxiety Disorder scale (GAD-7) assessed anxiety [36]. The Altman Self-Rating Mania Scale (ASRM) assessed manic symptoms [37]. The Mood Disorder Questionnaire (MDQ) screened for symptoms of bipolar disorder [38]. The Adult ADHD self-report scale (ASRS) [39] assessed for any ADHD, inattentive presentation, hyperactive presentation, or combined presentation.

Risky behavior

The Alcohol Use Disorders Identification Test (AUDIT) [40] screened for risky alcohol or substance use.

Functioning

The WHO Disability Assessment Schedule 2.0 categorized disabilities in cognition, mobility, self-care, getting along with others, ability to participate in life activities such as household and school or work activities, and social participation. Overall summary score and domain scores were calculated by using the item-response-theory based scoring method [41].

Statistical analysis

To compare the demographic and clinical characteristics between men and women, t-tests and chi-square (or Cochran-Mantel–Haenszel) tests were used for continuous and categorical variables, as appropriate. To enhance the interpretability of the comparisons, we further calculated the effect sizes for the gender differences for (1) continuous variables as Cohen’s d [i.e., (Meanwomen – Meanmen/SDoverall)]; and (2) categorical variables as difference in proportions (i.e., %women - %men). SAS 9.4 was used to perform all statistical analyses. Statistical significance was set at p ≤ 0.05.

Results

Demographic characteristics

This PCARES Registry cohort included 3,556 individuals (37% men, 63% women) with a mean age of 42.2 ± 17.0 years (see Table 1). Most participants (84.5%) identified as White; 5.9% identified as Black or African American, 9.7% identified as another race, and 4.4% identified as Hispanic. Participants’ marital status included 47% who reported being single, 37% married, and 16% formerly married. Almost half the sample included participants who reported having public insurance or self-paying for healthcare. Demographic factors of age, education (mean high school graduation rate 90.1%, SD 4.6%), and race and ethnicity did not differ significantly between men and women. Household median income (mean $59,815, SD $10,973) did not differ between men and women, and was below the median household income for Pennsylvania 2015–2019 [42]. Significant differences by gender were reported in marital status (more women [19%] than men [11%] were formerly married) and insurance type (more men [59%] than women [55%] with commercial insurance). Rurality distribution included 5.2% in rural county/rural municipality (N = 180); 2.4% in rural county/urban municipality (N = 85); 8.8% in urban county/rural municipality; and 83.6% urban county/urban municipality (N = 2923). BMI did not differ by gender.

Table 1 Patient demographics and trauma exposure stratified by gender
Full size table

Trauma

Men reported higher frequency of exposure to war-related trauma, serious accidents, major disasters, and seeing others injured seriously categories than women. Women reported significantly higher frequency of unwanted sex than men (Table 1).

Clinical Diagnosis

The most common diagnoses were major depressive disorder (41.2%), bipolar disorder (10.4%), and generalized anxiety disorder (21.5%; see Table 2). In order of the magnitude of gender difference by effect size (ES), more women were diagnosed with major depressive disorder (MDD) (44.1%, ES = 8.3% ), generalized anxiety disorder (22.9%, ES = 4.0%), post-traumatic stress disorder (PTSD) (7.7%, ES = 3.8%), panic disorder (6%, ES = 2.1%), bipolar II disorder (3.7%, ES = 1.7%), and eating disorders (1.4%, ES = 1.2%), while more men were diagnosed with autism spectrum disorder (10.9%, ES=-8.4%), ADHD (9.8%, ES=-4.5%), and psychotic disorder (4.5%, ES=-2.0%). Women had more diagnoses than men, with a mean of 1.6 compared to 1.4 for men, with a small effect size (Cohen’s d = 0.18, p < 0.01).

Table 2 Clinical Diagnoses stratified by gender
Full size table

Transdiagnostic measures

The percentage of positive DSM-5 Level 1 Cross-Cutting Symptom Measure screener areas (Table 3) differed between men and women for most items. In order of the magnitude of difference by ES, women had higher rates of anxiety (ES = 8.4%), somatic symptoms (ES = 7.2%), anger (ES = 7.1%), dissociation (ES = 7.0%), memory (ES = 5.6%), depression (ES = 4.5%), personality symptom domains (ES = 4.3%), and sleep disturbance (ES = 4.2%), whereas men had higher rates of tobacco use (ES=-4.7%), psychosis (ES=-3.3%), and substance use (alcohol and other substances (ES=-2.4%). Women reported significantly higher DSM Level 1 Cross-Cutting Measure overall symptom score and number of symptom domains than men, with small effect sizes. In the DSM-5 Level II symptom scores, women had significantly higher mean T-scores in sleep, depression, anxiety, and anger than men. Similarly, on the battery 2 measures (Table 4), women had higher scores on depression (PHQ-9, ES = 0.30) and anxiety (GAD-7, ES = 0.31), and higher rates of ADHD combined presentation, as compared to men (ASRS, ES = 8.1%). On the MDQ, men scored higher than women on the risky behavior item (ES=-7%). On the PID-5-BF, women reported significantly higher negative affect and detachment, while men reported higher antagonism and disinhibition (Table 3). Women scored higher than men on suicidality (ES = 0.14, Table 3).

Table 3 Transdiagnostic measure, personality, suicide and alcohol/substance use screeners stratified by gender
Full size table
Table 4 Mania, Depression, Anxiety, ADHD, Bipolar Disorder measures stratified by gender
Full size table

Substance use

Men scored significantly higher than women on the AUDIT alcohol (ES=-0.07) and substance use subscales (ES=-0.19, Table 3).

Functioning

Disability scores (Table 5) indicated that the sample overall had worse functioning than 88% of the general population [43]. Women reported significantly greater disabilities in all domains compared to men, with disparities most apparent in life activities (ES = 0.24), mobility (ES = 0.19) and social participation (ES = 0.19). Both women and men rated social participation as the most impaired domain and self-care as the least impaired domain.

Table 5 Functioning as assessed by the World Health Organization Disability Assessment Scale (WHODAS) scores stratified by gender
Full size table

Discussion

We examined socioeconomic factors, trauma exposure, transdiagnostic measures, and functioning by gender at first visit in an ambulatory mental health clinic, including the first report to our knowledge of gender differences in the transdiagnostic instrument DSM-5 Level 1 Cross-Cutting Symptom Measure. Our sample was representative of the local population, and the gender balance of the sample is representative of broader treatment patterns seen for mental illness, with women seeking mental health care more often than men [44].

Women reported significantly higher severity of transdiagnostic psychopathology and number of comorbid mental illnesses, as well as a higher impact of mental health problems on functioning compared to men. This is contrary to other findings which indicate that men present with more severe mental illness at first visit, possibly as a result of increased reluctance to seek health care [45,46,47]. However, symptom prevalence on the DSM-5 Level I Cross-cutting Symptom Measure and the rates of diagnosis are consistent with previous findings that women have a higher rate of depressive disorders [48], bipolar II disorder [49], anxiety disorders [50], eating disorders [51], and PTSD [52]. We also found gender differences in trauma exposure, where women reported higher rates of unwanted sexual intercourse while men reported more experiences of war zone or casualty trauma. This generally maps onto known gender differences in lifetime trauma exposure [50].

While the diagnosis of ADHD was almost twice as high in men than in women, the combined presentation of ADHD self-report score was higher in women than men [53]. This is consistent with some research reporting that undiagnosed adults who screen positive for ADHD are more likely to be women versus men [54]. In a large meta-analysis, women with ADHD were more frequently diagnosed with the inattentive subtype, and men were diagnosed more frequently with combined subtype [55]. However, a study of the ASRS used in a sample of patients in mental health clinics found that gender differences were not noted in subtype scales [56]. One explanation for the difference is that self-report scales may pick up more subtle presentations that are not uncovered by clinicians. An alternate explanation is that clinicians are accounting for ADHD symptoms as a part of another disorder. Further work needs to be done to determine the benefit of screening instruments for ADHD in a clinical diagnostic approach.

Women had higher sleep disturbance than men as measured by DSM5 Level I and II screeners. While population-based studies demonstrate that women have higher percentage of sleep time and slow wave sleep, with less sleep disturbance in response to an external stressor than men [57], the predominance of mood and anxiety disorders in women could contribute to more women than men reporting sleep difficulties.

A greater proportion of women than men reported anger as a prominent symptom on the DSM5 Level 1 and Level II severity screeners. While subjective anger is not part of the diagnostic criteria for depression in adults and is not routinely measured in studies of mental illness that are not directly studying aggression or suicide, anger has been posited to be an “alternative” expression of low mood in depression. In fact, an interesting analysis of the National Comorbidity Study Replication (NCS-R) showed when anger and aggression were measured (along with risk taking and substance use) and scored as primary symptoms of depression, gender differences in diagnosis of depression disappeared [58, 59]. Because the NCS-R was a population-based sample and our sample is a naturalistic, clinic-based sample, the findings are not directly comparable. The finding of prominent anger in women highlights the importance of measuring symptoms in transdiagnostic domains in the clinic setting.

Somatic symptoms were more prevalent among women than men, consistent with available studies [60] [61]. Women also reported higher rates of suicidality than men, consistent with findings that women have higher rates of non-fatal suicide attempts and men have higher rates of suicide deaths [62]. Women reported higher negative affect and detachment, but lower antagonism and disinhibition than men in our sample, though effect sizes were small. While some studies show gender discrepancies in personality traits and disorders, others do not; the differences found here may be due to measuring personality traits through self-report compared to structured diagnosis [63]. While there may not be gender differences in population rates of personality traits or disorders, an interesting next question is how strongly expressed personality traits impact women and men in day-to-day functioning.

In our sample, women reported experiencing more impairment participating socially, accomplishing household tasks, and completing daily work or school activities compared to men. The greatest discrepancy between men and women was impairment in completing life activities. One potential explanation is that women are more commonly responsible for household tasks than men resulting in more opportunities for an impact in functioning (a floor effect in male functioning). Only a few studies have assessed gender differences in functional impairment associated with psychopathology [64]. Results similar to our study were reported from a large prospective multi-center study which found that mental health problems were more likely to affect women’s marital functioning but men’s work functioning [65]. Overall, the finding that psychiatric symptoms differentially impact functioning highlight the need to monitor in our patients both psychiatric symptoms themselves and how symptoms affect functioning.

Strengths and Limitations

Strengths in our approach include using validated, transdiagnostic self-report measures in a framework recommended by the DSM-5 in a naturalistic, real-world cohort. These results have generalizability to patients seeking care in psychiatric clinics, however our sample includes a majority White, non-Hispanic, English speaking population and may not generalize to racial, ethnic and gender identity populations underrepresented in medical research. A limitation of using this approach is that data are gathered for clinical purposes and extracted from the EMR and the diagnostic description are not as comprehensive as structured interview [28]. The self-report measures do not adequately capture cognitive functioning and neurodevelopmental domains, and those with neurodevelopmental disorders experience more comorbid psychopathology. The sample is biased by the fact that they are presenting for clinical care and we cannot compare our sample directly to a group that did not obtain clinical care. We acknowledge the importance of gender expansive measurements, such as transgender or nonbinary identities, which have significant implications in mental health, however, the variable to represent gender is self-reported by patients during enrollment in the EMR and did not at the time allow for non-binary description of gender identity. Future studies should consider reporting gender expansive categories on these measures [66].

Clinical implications

Questions remain in how to integrate such layered levels of data to personalize care. DSM-5 added the transdiagnostic assessment paradigm to promote the assessment of several dimensional areas across categorical diagnoses. Symptom scales that take a transdiagnostic domain approach and include social factors, trauma exposure and domains of functioning allow clinicians to more efficiently and accurately identify areas of concern and distress for patients. The impact of mental illness and psychopathology are affected by biological, psychological, social and cultural factors that influence both the presence of symptoms and resultant level of functioning.

We have found that the higher level of psychopathology and functional impairment, exposure to sexual trauma, and anger in women when compared to men at first treatment suggests that women are waiting longer in the course of illness to seek treatment than men. This may affect rates of recovery and highlights a need to promote earlier treatment intervention in women’s health. Because mental illnesses can have a severe impact on daily functioning, understanding the mediating and moderating factors between mental illness and impairment in men and in women may generate targets for further study, including data-driven approaches to treatment matching.