Abstract
The relationships between gender and malingering have received little attention in the literature. Our study examined data from 1,748 patients committed as incompetent to stand trial between 2008 and 2017, of whom 397 were women. Scores on a structured assessment of feigned psychiatric symptoms were only slightly higher for men than for women. Yet evaluators believed that over 23 percent of men but less than 15 percent of women were malingering. Our data suggest that these gender differences in rates of malingering may be attributable to symptom constellations and extent of criminal arrest history.
Competence to stand trial evaluations are the most common assessments conducted by forensic evaluators,1 and guidelines indicate that the assessment of feigning is a critical component of these evaluations.2,3 Research has shown that a substantial portion of criminal defendants feign a mental disorder or defect to avoid prosecution,4,–,6 with some estimates as high as 21 percent.7 Offenders frequently use multiple strategies when feigning, such as feigning cognitive or memory deficits in addition to psychiatric symptoms.7,–,11 As is the case with most literature examining questions pertaining to the criminal justice system, the majority of these studies contained few women. Furthermore, no studies have documented gender differences in feigning for the purpose of being adjudicated incompetent to stand trial (IST). There are reasons to suspect, however, that there may be gender-based differences in rates of malingering.
First, there are gender differences in patterns of criminal offending, including types of charges and criminal histories. For example, men commit 85 percent of all violent crimes12,13 and are arrested more frequently than women for all offenses except for embezzlement and prostitution.14,15 Men are also more likely to have prior offenses, with only 39 percent of incarcerated men having one or no prior offenses, compared with 51 percent of incarcerated women.14 Defendants charged with more serious offenses may be more motivated to malinger,16 and repeated incarcerations provide opportunities for inmates to learn how to feign mental illness from their peers.17 If women and men face different types of charges and have different criminal arrest histories, they may malinger at different rates.
Second, differences in mental health characteristics between men and women who are involved in the criminal justice system may lead to different rates of suspected malingering. Incarcerated women are more likely to have mental health problems than their male counterparts.12,18 For example, female inmates are two to three times more likely than male inmates to have borderline personality disorder and posttraumatic stress disorder (PTSD).19,–,21 Symptoms of borderline personality disorder and PTSD can lead to false positives on structured assessments of feigned psychopathology,22,–,27 suggesting that, on structured assessments, incarcerated women may be more likely to be incorrectly identified as malingering than incarcerated men. Because scholars often recommend using structured instruments to assess for feigning, it is important for evaluators to understand their accuracy for both men and women.
With the growing number of women incarcerated and the expanding number of IST defendants burdening state hospital systems, it is critical that we understand gender differences in malingering to improve the accuracy of competency to stand trial evaluations. The objectives of this study were to describe gender differences in rates of malingering and to explore those factors related to clinicians' opinions of malingering. We also sought to describe gender differences in demographic, clinical, and legal characteristics of individuals committed as IST to a forensic psychiatric hospital.
Methods
This research was conducted as part of a collaborative effort between the California Department of State Hospitals-Napa (DSH-Napa) and University of California-Davis School of Medicine, Department of Psychiatry and Behavioral Sciences, funded by the California Department of State Hospitals. This research was approved by the Human Subjects Committee at DSH-Napa, the State of California Committee for the Protection of Human Subjects, and the University of California-Davis School of Medicine institutional review board. Because these were data collected for clinical purposes, the review board granted a waiver of informed consent.
Participants
This study was conducted at DSH-Napa, a large inpatient state psychiatric hospital located in northern California. Approximately 380 beds at this facility are allocated for the competence restoration of patients committed as IST. With rare exceptions, only defendants charged with felony offenses are committed to DSH-Napa. At the time of this study, the maximum length of commitment for restoration of offenders with felony charges was three years. The records of 3,134 unique patients found IST and admitted to DSH-Napa for restoration of competence between March 25, 2008 and October 31, 2016 were eligible for inclusion in the study.
Procedure
All patients whose data were included in this study were found IST by the courts and admitted directly from the referring county jails for restoration. An effort was made to conduct an interview with all admissions, which included a brief assessment of symptoms, competence to stand trial, and a structured screening tool to assess for feigned psychiatric symptoms. The purpose of this brief evaluation was to provide information to the treatment providers; results were not shared with the criminal justice system. The majority of these interviews were conducted by a psychologist or a psychiatrist. The second author initiated this screening procedure and conducted approximately 24 percent of these interviews. As one component of this interview, several structured assessments were administered. Some admissions could not be interviewed (e.g., too thought disordered or agitated to participate in the evaluation), and their information was excluded from the analyses.
Once the patient was interviewed and the assessment tools were completed, the interviewer made clinical judgments about each new admission in three areas: competence on each of the prongs of the Dusky standard, as well as overall competence28; whether the evaluator suspected that the patient was malingering psychiatric symptoms or cognitive or memory deficits; and the presence of possible cognitive deficits. In general, the evaluators based these judgements on their interviews in conjunction with the results of the administered assessments. Clinicians were not required to document their rationale for their clinical opinion.
All evaluators were trained in the assessment of malingering. Opinions were documented on a coding sheet that also included information gathered from the patient's record, including basic demographic information, clinical information (e.g., current medications, prior psychiatric treatment), criminal arrest information (e.g., most serious commitment offense, prior IST finding, number of prior arrests), as well as the scores obtained from the assessments. All interviewers were trained in these procedures, although inter-rater reliability for the structured assessments was not established.
Measures
The second author developed a semi-structured interview designed to obtain information about the patient's history. Questions included: “What is your understanding of why you are at a hospital?”; “What does not competent to stand trial mean?”; “Do you think you are competent?”; and “Why do you think you were found not competent?” The evaluators also asked the subjects questions about their past IST commitments, previous mental health treatment, current medications, and educational background, including any enrollment in special education classes. Interviewers were permitted to ask other follow-up questions as appropriate.
The Miller Forensic Assessment of Symptoms Test
The Miller Forensic Assessment of Symptoms Test (M-FAST) is a screening instrument designed to identify feigned psychopathology. It is a 25-item structured interview that can be administered in approximately five minutes. Although the M-FAST contains multiple subscales, the manual suggests that a total score of six or greater indicates that a more extensive assessment of feigning is needed.29
The Brief Psychiatric Rating Scale
The Brief Psychiatric Rating Scale (BPRS) is a widely used instrument that consists of 18 items designed to assess affective symptoms, hostility, and positive and negative symptoms of psychosis.30 In this study, the brief screen consisted of the four items that quantify positive psychotic symptoms (thought disorganization, suspiciousness, hallucinations, and unusual thought content). Items are rated on a seven-point scale, with 1 indicating that the symptom was not observed, and 7 indicating that the symptom was very severe.
Competency Screening
The evaluators screened individuals on admission for competence on both prongs of Dusky (i.e., whether the patient understood the nature of the criminal proceedings, and whether the patient was able to assist counsel in the conduct of a defense in a rational manner) by asking five questions.28 To assess for understanding of the proceedings, the evaluators asked about the roles of courtroom personnel (i.e., judge, jury, prosecutor, defense attorney). To assess their ability to assist, the evaluators asked the patients to name their defense attorney and to provide their opinion of their attorney's performance. Each of the five items was scored as 0, 1, or 2, with 2 indicating an adequate answer, 1 a partially correct answer, and 0 no answer or an incorrect or delusional response; thus, scores ranged from 0 to 10.
Data Analysis
We analyzed the data using SPSS 25 (IBM Corp., Armonk, NY). Statistical analyses included frequency distributions to provide information regarding basic demographics and base rates of possible feigning. We conducted chi-square analyses and analyses of variance to assess gender differences on many of the variables.
Hierarchical conditional logistic regression analyses were conducted for each gender separately to statistically identify those factors most strongly as-sociated with evaluators' judgments concerning which subjects were malingering. Using the evaluator judgments as the dependent variable, we entered the M-FAST scores on the first step because there was a clear relationship between M-FAST scores and evaluators' opinions. In the second step, we included the individual scales of the BPRS, the overall competence assessment score, past criminal arrest history (dichotomized as 0 to 2 or 3 or more prior arrests), past psychiatric history (yes/no), and seriousness of offense (dichotomized as murder or robbery versus all other offenses) as independent variables.
Results
There were a total of 3,134 unique IST admissions available for review during the specified time period, but only 1,748 of the 3,134 records contained complete information based on an interview. Of these 1,748 records, 397 (22.7%) were women. Gender and commitment offense did not differ between individuals who were and were not interviewed (p = .840 and .158, respectively). Individuals of Hispanic or Asian descent were less likely to have been interviewed because of language barriers. Only data from the summaries of these admission interviews were included in the analyses.
Scores on the M-FAST in the sample ranged from 0 to 25, with mean ± SD score of 4.45 ± 4.95. The modal score was 0; however, 31.1 percent scored at or above 6, the threshold score for suspected feigning.29 On the BPRS, scores in this sample ranged from 1 to 25, with higher scores indicating more severe psychotic symptoms. A score of less than 4 occurred when there was not enough information to rate all four items. The mean ± SD BPRS score for the sample was 8.64 ± 4.67. On the competence screen, the mean ± SD score was 6.03 ± 3.21.
We observed gender differences on several factors. As Table 1 shows, although there were no differences in educational attainment between men and women, men were more likely to have been placed in special education classes. There were also significant differences between men and women on the commitment offense, with women less likely to be found IST for homicide, weapons charges, and threats of violence. There were no differences between genders for assault or battery charges. Women were more likely than men to have been found IST for nonviolent offenses such as theft, drug charges, and miscellaneous offenses (typically vandalism). Not surprisingly, the overwhelming majority of IST patients charged with sex offenses were men.
Gender Differences on Demographic Variables
Table 2 provides the differences between genders on criminogenic risk factors. This table indicates that women evidenced less extensive criminal arrest histories. Consistent with previous literature,16 we recoded our representation of “commitment offense” to the dichotomy of murder or robbery versus all other offenses. When categorized in this fashion, women were as likely as men to commit a serious offense. We also found a relationship between offense type and suspected malingering. More specifically, for the total sample, we found that over 35 percent of patients charged with murder or robbery were suspected of malingering. For all other offenses combined, less than 19 percent were believed to be malingering (chi-square (1) = 40.98, p < .001). This relationship was true for both men (chi-square (1) = 35.49, p < .001) and women (chi-square (1) = 4.24, p = .039).
Gender Differences on Criminal Justice Variables
As seen in Table 3, women were more likely to have a history of inpatient or outpatient psychiatric treatment. There were no gender differences in overall BPRS scores or on any of the BPRS subscales, with one exception: women were less likely to endorse hallucinations. In examining the relationship between symptoms and M-FAST scores, although many correlations were statistically significant because of the large sample size (correlations ranged from –0.053 to 0.292, p ranged from .028 to < .001), the only meaningful relationship we found between BPRS subscale scores and total M-FAST scores was that the hallucination subscale was positively related to the M-FAST score (r = 0.292, p < .001). In other words, the higher the score on the hallucination subscale, the higher the score on the M-FAST; this was true regardless of gender.
Gender Differences on Psychiatric Variables
Table 4 demonstrates the difference between genders on malingering opinions and M-FAST scores. Although neither the M-FAST total score nor the percent above the cutoff score were statistically different between genders, evaluators nevertheless judged women as substantially less likely to be malingering on admission (23.8% of men compared with 14.6% of women, chi-square (1) = 15.35, p < .001). Moreover, when examining the relationship between gender and discharge diagnosis, we found that men were returned to court as competent with a diagnosis of malingering almost five times as frequently as women (8.6% for men compared with 1.8% for women, chi-square (10) = 29.32, p < .001).
Gender Differences on Malingering Variables
Table 5 provides the results of the hierarchical logistic regression analyses. The ability to predict evaluators' opinions was similar between genders, with an overall accuracy of 87.1 for men and 92.2 for women (100 represents perfect predictive power). In both genders, classification was most accurate in predicting who was believed not to be malingering. As can be seen in Table 5, although there were commonalities between genders with respect to which factors were most highly associated with evaluators' opinions of malingering (i.e., total M-FAST score, conceptual disorganization, and offense seriousness), the model for men contained several additional variables to improve accuracy (i.e., prior arrest history, past psychiatric history, and competence screening score). For women, only scores on the hallucination subscale of the BPRS improved the predictions of the malingering opinions. The odds ratio provides information on the direction of the relationship, with numbers above 1 indicating that the factor increased the evaluators' suspicions of malingering. Numbers below 1 indicate that the factor decreased the evaluators' suspicions of malingering. For both men and women, higher M-FAST scores and a commitment offense of murder or robbery increased frequencies of suspicions of malingering, whereas higher scores on the conceptual disorganization subscale of the BPRS decreased suspicions. For men, not having any prior psychiatric history increased the suspicion of malingering; fewer prior arrests and higher scores on the competence evaluation decreased suspicion. Higher scores on the hallucination subscale of the BPRS were associated with decreased frequencies of suspicions of malingering, but only for women.
Factors Related to Opinion Formation of Malingering by Gender
Discussion
Our data document several differences between men and women on criminal justice variables. Consistent with previous research,12,–,14 women were less likely to have an extensive criminal arrest history, and they also evidenced a different pattern of offending. Women were more likely to be arrested for minor or theft-related crimes, whereas men were more likely to be arrested for murder and sex offenses. Surprisingly, when murder and robbery offenses were combined, women and men committed offenses belonging to this single merged category at a similar rate. This is inconsistent with previous literature, which has shown that men tend to commit more serious offenses when offense type is dichotomized in this manner.16
In this study, although the combined offenses of murder and robbery did not differ between genders, when faced with these serious charges, both men and women evidenced increased rates of suspected feigning compared with men and women with other commitment offenses. In California, both murder and robbery carry longer sentences than most other offenses. The adaptational model of malingering suggests that individuals who malinger weigh the consequences of their actions and ultimately decide that malingering may be the most effective method for obtaining a desired outcome.31,32 Previous research has shown that motivation is relevant when evaluating base rates of malingering. For example, malingering rates for individuals involved in civil forensic cases are generally substantially higher than in criminal defendants, likely due to highly motivating financial incentives.33 Research has also shown that defendants with more serious charges are more than twice as likely to malinger than those with less serious offenses.16
In this study, evaluators opined that men were more than 1.5 times as likely as women to be malingering despite the fact that there were no statistically significant gender differences in M-FAST scores. Moreover, consistent with the evaluators' opinions on admission, men were much more likely to return to court as competent with a diagnosis of malingering. Our data indicate that the differences in malingering rates between men and women are related to patterns of symptoms coupled with the extent of their criminal arrest history. Other potential explanations for our results include gender bias against men, an under-identification of men as feigning on the M-FAST, or an over-identification of women as feigning on the M-FAST.
Certain types of symptoms decreased the evaluators' suspicions of malingering. For example, in both men and women, the presence of a thought disorder decreased the suspicion of feigning. This result is not surprising because numerous authors have noted that a formal thought disorder is difficult to feign.34 The hallucination subscale of the BPRS was modestly correlated with the total M-FAST score, suggesting a relationship between feigning psychiatric symptoms and the endorsement of hallucinations. This is consistent with literature that suggests that individuals who feign psychotic symptoms often endorse hallucinations but do not exhibit other signs of psychosis.4,34,35 Surprisingly, the presence of hallucinations decreased evaluators' suspicions of malingering in women. This unexpected finding may be attributable to the fact that women were less likely than men to endorse or evidence hallucinations. This suggests that when hallucinations were present in women, they represented a more severe and clearly genuine psychosis.
Although it appears that rates of malingering may in fact differ between genders and that these differences are related to both symptom constellations and criminal arrest history, there may be other explanations for the observed differences. For example, gender biases may have contributed to the observed gender differences in opinions of malingering. Despite the fact that men and women had similar M-FAST scores, evaluators were much more likely to opine that a male patient was malingering. Gender bias may lead evaluators to consider men to be more manipulative and antisocial, and thus more likely to be malingering.5,36,37 This explanation seems less viable in our study, however, because not only did evaluators opine that more men than women were malingering on admission, but men were also much more likely to return to court as competent with a diagnosis of malingering. Although this gender-variable difference in the ultimate malingering opinion also could reflect gender bias, extensive evaluation (including the administration of multiple structured assessments coupled with behavioral observations) is required before returning a patient to court as competent with a diagnosis of malingering. These comprehensive assessments are less likely to be subject to evaluator bias than the initial screening interviews.38 Therefore, it is more likely that the gender-variable differences in rates of malingering were attributable to symptom constellations and exposure to the criminal justice system than to gender bias.
Although multiple factors were related to clinicians' opinions of malingering, the M-FAST score was the strongest predictor in both genders. Therefore, another plausible explanation for our findings is that the M-FAST under-identifies men as feigning because men may use a wider range of strategies to malinger than women do. For example, it is possible that more men were considered to be malingering because the evaluator suspected that they were malingering symptoms not detected by the M-FAST (e.g., cognitive deficits). In a recent study of IST defendants who were determined to be malingering, most of whom were men, subjects were more likely to malinger both mental illness and cognitive deficits.9 Therefore, although men and women in our study had similar scores on the M-FAST, an instrument designed exclusively to detect feigned psychiatric symptoms, evaluators may have been more likely to opine that men were malingering because they believed men were using other strategies to malinger. In contrast, it is also possible that women are over-identified as feigning on the M-FAST. As previously noted, symptoms of borderline personality disorder and PTSD (diagnoses common among women involved in the criminal justice system) may result in false positives on structured assessments of feigned psychopathology.22,25,26 In a study of traumatized patients, the majority of whom were women, Rogers and colleagues39 reported that the Structured Interview of Reported Symptoms (SIRS) tended to over-identify individuals with significant trauma as malingering. This over-identification ultimately led to a revision of the scoring of the SIRS in an effort to decrease false positives.39 Therefore, our results may be related to evaluators' beliefs that many of the women with M-FAST scores at or above 6 were not actually feigning symptoms. It is unclear, however, if the M-FAST is as sensitive to trauma symptoms as other, lengthier assessments of feigning such as the SIRS. This is especially true given that the M-FAST primarily targets symptoms that are psychotic in nature. As Rogers noted, the false positives occurred on the subscales of the SIRS that do not evaluate feigned psychotic symptoms.39 Further research is needed to determine whether the M-FAST performs differently based on gender.
The primary limitation of this study is that multiple evaluators completed the admission evaluations, and the inter-rater reliability of the assessments was not measured. Although the M-FAST has demonstrated good inter-rater reliability, primarily because of the structured nature of the instrument, the BPRS scoring requires clinical judgment.
Conclusion
Our results indicate that rates of malingering differ between genders both on clinician judgment on admission and final diagnoses at discharge. These observed differences are at least in part associated with varying patterns of symptoms and exposure to the criminal justice system. As more women enter the criminal justice system, more structured research will be necessary to explore these differences.
Footnotes
Disclosures of financial or other potential conflicts of interest: None.
- © 2021 American Academy of Psychiatry and the Law