The prospective revision of the Diagnostic and Statistical Manual of Mental Disorders, 4th Ed., Text Revision (American Psychiatric Association 2000) is an opportune time to consider appropriate changes in the nosology. This manuscript addresses the status of malingering, a V code, or “other condition that may be a focus of clinical attention” (American Psychiatric Association 2000, p. 739). At the time of this writing in August, 2010, there do not appear to be any moves to revise the DSM text on malingering, based on review of the APA website. In light of the serious concerns reviewed below about the conceptual and practical utility of the directions given in the DSM text for identifying malingering, it is recommended that this condition be dropped from DSM-5 and replaced with a more empirically based alternative.

The term “malinger” appears to emerge from a late eighteenth century French idiom (malingrer) meaning either “to suffer” or “pretend to be ill” (Online Etymology Dictionary 2010). Modern connotations typically reflect only the latter, more incriminating, and pejorative nuance. In the current DSM text on malingering, it is defined as “…the intentional production of false or grossly exaggerated physical or psychological symptoms, motivated by external incentives…” (American Psychiatric Association 2000, p. 739). The entry goes on to state that malingering should be “strongly suspected if any combination of the following is noted”: medicolegal context of presentation, marked discrepancy between self-reported stress or disability and objective findings, lack of cooperation during diagnostic evaluation or with prescribed treatment, or presence of antisocial personality disorder. No further detailed criteria for establishing the presence of this condition are given in the text.

Conceptual Concerns

There are a number of conceptual issues that bedevil the text on malingering in the DSM-IV-TR. Like many of the diagnoses of mental disorders in the DSM, malingering is presented as a categorical condition. That is, one either is or is not a malingerer. Recent evidence calls this assumption into serious question. Walters et al. (2009) reported that taxometric analyses of results from psychiatric outpatients completing a health inventory with embedded validity scales indicated that feigning of physical symptoms was dimensional, not categorical. Similarly, Walters et al. (2008) analyzed Minnesota Multiphasic Personality Inventory 2nd Edition (MMPI-2) and Structured Inventory of Reported Symptoms (SIRS) data from forensic neuropsychiatric evaluees and found evidence for feigned psychiatric symptoms as a continuous, not a categorical phenomenon. Finally, Walters et al. (2009) analyzed results on objective neurocognitive feigning instruments in forensic neuropsychiatric patients and reported evidence for a dimensional structure of the construct. Thus, at a fundamental level, the categorical DSM criteria for malingering do not map on to available objective data on the nature of the phenomenon.

The DSM conceptual model for malingering has been criticized from other perspectives as well. Surveying the criteria noted above whose presence was supposed to increase suspicion of the presence of malingering, Rogers (1990) argued that the only integrating theme was “badness” or a criminological model. As evidence of this, Rogers suggested that the DSM criteria essentially characterize the malingerer as a “bad person” (antisocial personality disorder), in “bad circumstances” (legal difficulties), doing “bad things” (uncooperative with evaluation and treatment). Unlike other diagnoses, this conception of malingering seems less related to a description of distinct symptoms and behaviors and more related to a proscription of one’s character.

In contrast, Rogers (1990) offered an “adaptational” model, which views malingering as a reasoned response by an individual facing adversarial circumstances with few alternatives perceived to be available. This perspective removes some of the moralistic baggage of the DSM criteria and focuses on identification of the feigning response to situations perceived to be challenging by the patient. In this sense, recognition of feigned symptoms may assist the clinician to understand basic anxieties (e.g., fear of military service) or insecurities (e.g., limited financial resources) that may in some circumstances benefit future care. As Stone and Boone (2007, p. 11) state, “recognition of feigning behaviors may prove to be the first therapeutic step to understanding the patient’s actual needs.”

Another conceptual concern with the present DSM text on malingering is the implied assumption that malingering is monolithic and that malingerers will fake bad on most, if not all of the procedures they undergo. In contrast to this perspective, recent research has identified at least three distinct targets for false symptoms, including psychiatric, physical/somatic, and cognitive/neuropsychological feigning (Rogers 2008a, b). Evidence for at least partial independence in two of these domains was reported by Alwes et al. (2008), who administered objective tests for feigned psychiatric symptoms and false neuropsychological deficits to over 300 individuals undergoing forensic neuropsychiatric evaluation. They reported that while the base rate of probable psychiatric feigning was about 7.5%, the prevalence of probable feigned neuropsychological deficit was 24.4%, and only 13 patients met criteria for probable feigning of both types of symptoms. Consistent with this evidence for independence of feigned symptoms, Nelson et al. (2007) factor analyzed MMPI-2 validity scales and multiple neurocognitive feigning instruments from a compensation-seeking traumatic brain injury sample. The two domains of feigning detection indices loaded on separate, distinct factors. The authors concluded that “litigants and claimants can exhibit complex presentations and unique forms of response bias (cognitive, psychological, or both)” (p. 447).

Practical Issues

A significant practical concern is that the text describing malingering in DSM-III (American Psychiatric Association 2000) was barely changed 20 years later in DSM-IV-TR (American Psychiatric Association 1980) and, as noted above, apparently will not be updated in DSM-5. From a scientific perspective, this is very troubling in light of the fact that the literature on malingering has expanded tremendously in the 30 years since the appearance of DSM-III. Figure 1 shows the number of publications with malingering as a keyword from 1980 to 2009. It can be seen that at the time DSM-III appeared, only two or three papers per year were published on malingering, whereas in 2009, the last year for which complete data were available, there were more than 90 publications on this topic. In the 20 years between 1989 and 2009, over 1,200 papers on malingering appeared (ISI Web of Science). In the 3 years prior to preparation of this manuscript, four edited volumes have been published on malingering (Boone 2007a, b; Larrabee et al. 2007; Morgan and Sweet 2009; Rogers 2008a, b).

Fig. 1
figure 1

Annual number of publications with keyword malingering from 1980 to 2009. Note: Image captured from ISI Web of Knowledge

Full size image

The exponential growth in malingering literature has also supported development of several professional position papers (e.g., AACN Board of Directors 2007; Bush et al. 2005; Heilbronner et al. 2009). These works offer empirically based, clinically relevant, and consensus-defined recommendations related to the assessment and diagnosis of malingering. It is regrettable that many professionals are more inclined to consult these references than what is meant to be the definitive reference tool in the practice of mental health professions (i.e., DSM). In short, the failure to update the criteria for malingering in DSM-5 ignores more than 30 years of empirical and theoretical work on the topic. The evolution of symptom validity and malingering literature in recent decades has culminated in a sophisticated conception of malingering that essentially renders DSM-IV-TR criteria obsolete.

In addition to the failure to incorporate more recent data on the condition, the text guiding identification of malingering is problematic in clinical application as well. As noted earlier, the key features of both the DSM-III and DSM-IV-TR descriptions of malingering include (1) the intentional (voluntary) production of false or grossly exaggerated physical or psychological symptoms and (2) motivated by external incentives (American Psychiatric Association 1980, pp. 331–332; American Psychiatric Association 2000, pp. 739–740). Considering the criterion of intentionality, in the absence of a confession, or perhaps a video recording performance of an act a patient claims to be unable to carry out, it is difficult to discern an objective procedure for meeting this prong. In fact, Erickson (Pankratz and Erickson 1990) emphasizes that attributions of intention, as required by the DSM criteria, are likely to be highly speculative. Thus, the current framework for identifying malingering virtually requires subjective determination of a key part of the diagnosis of this condition. Turning next to external motivation as a requirement, Boone 2007a, b points out that it is exceptionally difficult to arrive at an objectively based discrimination between externally motivated malingering and internally motivated somatoform disorder. In fact, it seems likely that most behaviors are driven by both intrinsic and extrinsic factors, and once again, in the absence of honest self-report, objective and validated methods for drawing this distinction are lacking, leaving the other core aspect of DSM-defined malingering also to be established by subjective means. Given the severe consequences of a determination of malingering in many circumstances such as civil and criminal legal actions, it seems undesirable to base the relevant criteria solely on clinical judgment, which has been found unreliable in most studies of its accuracy rates (Garb 1998; Zimmerman 2003).

The DSM-IV-TR text on malingering has also proven problematic in application in another important way. As noted above, clinicians are directed to strongly suspect the presence of malingering when two or more of four conditions are met: medicolegal context of presentation, marked discrepancy between claimed stress or disability and objective findings, lack of cooperation in diagnostic process and in compliance with treatment regimen, and presence of antisocial personality disorder. Unfortunately, when Rogers (1990) evaluated the accuracy of these criteria, correct classification rates were only 20%. Thus, the screening criteria for malingering provided in DSM are far too inaccurate to be used in clinical practice.

To summarize, the current DSM text on malingering is terribly flawed on both conceptual and practical grounds. This suggests that this V code should be replaced with a more empirically supported alternative. With the goal of moving toward such a replacement, current perspectives on alternative models will be reviewed next.

Alternative Models for Identifying Malingering

In response to growing concerns about the difficulties in implementing the recommendations of DSM for identifying malingering, a number of alternative frameworks have been proposed. These vary considerably in their complexity and completeness. In recognition of the domain-specific nature of malingering reviewed earlier, these will be presented in separate sections addressing feigning of psychiatric, cognitive or neuropsychological, and physical or somatic symptoms.

Detecting Feigned Psychiatric Symptoms

In an extensive research program, Richard Rogers has explored conceptual models, detection strategies, and classification accuracies for identifying feigned psychiatric symptoms. One of his important points of clarification is that tests and other objective procedures do not identify malingering, as defined by DSM, but rather detect fabricated or significantly exaggerated symptoms “without any assumptions about… goals” (Rogers et al. 2010, p. 5). Note that this focuses attention on objective documentation of false symptom reports, without attempting to impute volition or external goals. This sidesteps the seemingly intractable problem of inferring motivation and volition without validated, objective methods. At the time of this writing, the final form of DSM-5 has not yet emerged. However, it is possible that changes to the diagnostic criteria for conversion disorder and factitious disorder may mitigate the need to distinguish consciously determined from unconsciously determined symptoms as well as to discriminate internal from external goals.

Rogers (1990) provided an extensive review of strategies for detecting feigned psychiatric symptom reports that had appeared in the published literature. This early work was updated in Rogers (2008a, b). Table 1 summarizes the detection strategies he concluded were validated, as well as objective scales from various instruments implementing these approaches to detecting malingering. Rogers (2008a, b) divides these into two broad categories: unlikely presentations (first four strategies) and improbable symptoms (last five strategies). It is interesting to note that many of these approaches to detecting feigned psychopathology have been endorsed in a more qualitative form by Resnick and Knoll (2008), who review literature on detection of malingered psychosis from the standpoint of forensic psychiatry. For example, they endorse reviewing reports of hallucinations and delusions from the perspective of atypical content, unbearable severity, and inconsistency of reported versus observed behavior.

Table 1 Selected strategies for identifying feigned mental disorder
Full size table

As reviewed in Table 1, there are several instruments and scales that allow objective detection of feigned symptoms, including validity scales from the MMPI-2 (Butcher et al. 2001), MMPI-2-RF (Ben-Porath and Tellegen 2008), Personality Assessment Inventory (Morey 2007), and the Psychological Screening Inventory. Additionally dedicated instruments validated for this purpose include the SIRS and the Miller Forensic Assessment of Symptoms Test. The varying operating characteristics of these instruments and scales are reviewed by Berry et al. (2008). Thus, using appropriate methods, it is possible to achieve reliable and valid identification of false psychiatric symptom reports.

Consideration of the available literature on detection of malingering suggests the following:

  1. 1.

    Focusing on objective identification of feigned psychiatric symptom reports, without attempting to infer volition or motivation, may enhance the focus and probably the accuracy of criteria.

  2. 2.

    Systematic review of the published literature on techniques for detecting feigned psychiatric symptom reports may yield useful recommendations for clinical identification of this condition.

  3. 3.

    Employment of multiple strategies for identifying false psychiatric symptom reports will likely increase accuracy of classification.

  4. 4.

    A focus on minimizing false-positive rates for any single detection strategy may allow adequate sensitivity to feigning with minimal cost in specificity when multiple strategies are used.

Detecting Feigned Cognitive/Neuropsychological Symptoms

Neuropsychologists and neuropsychiatrists often evaluate patients with purported brain injuries in secondary gain contexts, settings in which substantial benefits may be attained by appearing more impaired than is in fact the case (e.g., civil lawsuits, disability evaluations, competency to stand trial, etc.). Clinicians working with this population have long been dissatisfied with the DSM framework for determining malingering. An immediate concern was that the DSM criteria were apparently written primarily with traditional psychiatric disorders in mind. Whereas feigning a psychiatric disorder might involve complaints and behaviors concerning a discrete set of signs and symptoms during evaluation and observation, neuropsychologists and clinical psychologists measure level of ability using standardized tests of intelligence, memory, problem solving, and other areas of cognitive function.

Several early tests for malingered neurocognitive deficit (MND) such as the Rey 15-Item Test were described in Lezak’s influential text on neuropsychological assessment (Lezak 1976). However, a major breakthrough occurred with the adoption of forced-choice testing formats as described by Binder and Pankratz (1987). Typically, this involves posing a question or problem to a test-taker with two answers to choose from. This approach means that the probability of a correct response in the absence of any preserved ability is 50%. Over a large number of trials, a feigning patient may be clearly identified when he or she scores statistically significantly below chance. Such a result strongly implies that the test-taker knew the correct answer and instead chose the incorrect alternative. Although early work in this area used the conservative statistically significantly below chance criterion, studies using simulated malingerers determined that only a fraction of instructed feigners performed this poorly with any regularity (Guilmette et al. 1993). Recently, Greve et al. (2009) confirmed this low prevalence by reporting that statistically significantly below chance performance occurred in only about 12% of over 1,000 forensic neuropsychological evaluees given multiple symptom validity tests.

The low prevalence of statistically significantly below chance performance by known feigners led to a move to identify “norm-based” cutting scores, in which results from large groups of patients with the disorder in question who were not seeking compensation were used to set a “floor” on expected performance from cooperative patients (Guilmette et al. 1994). Compensation-seeking test-takers who scored below this level but not statistically significantly below chance were said to have invalid test data. Vickery et al. (2001) meta-analytically reviewed several of these tests and concluded that, on average, they were moderately sensitive and highly specific for the detection of feigned neuropsychological impairments.

There are now a large number of well-validated tests for detection of feigned cognitive deficits. Grote and Hook (2007) reviewed several of the most widely studied procedures. They drew generally positive conclusions about many of these instruments, which included the Computerized Assessment of Response Bias, the Portland Digit Recognition Test, Test of Memory Malingering, Victoria Symptom Validity Test, Medical Symptom Validity Test, Letter Memory Test, Word Memory Test, and Validity Indicator Profile.

After reasonably accurate tests for feigned neuropsychological deficits appeared, attention turned to providing a framework for diagnosing MND. Early researchers (e.g., Greiffenstein et al. 1994; Nies and Sweet 1994) provided preliminary suggestions for detection of cognitive malingering. However, the system that has proven the most influential in this area to date has been that proposed by Slick et al. (1999). Table 2 provides a summary of the framework.

Table 2 Summary of diagnostic criteria for malingered neurocognitive dysfunction
Full size table

Using these criteria, Slick et al. (1999) went on to propose diagnostic categories for MND. These were definite, probable, and possible MND, as follows:

  • Definite MND

    1. 1.

      Presence of a substantial external incentive (A)

    2. 2.

      Definite negative response bias (statistically significantly below chance performance) (B)

    3. 3.

      Behaviors from group B are not fully accounted for by other factors (D)

  • Probable MND

    1. 1.

      Presence of a substantial external incentive (A)

    2. 2.

      Two or more types of evidence from neuropsychological testing excluding definite response bias (B) (or) one type of evidence from neuropsychological testing and one type of evidence from self-report (B and C)

    3. 3.

      Behaviors from group B and C are not fully accounted for by other factors (D)

  • Possible MND

    1. 1.

      Presence of substantial external incentive (A)

    2. 2.

      One or more types of evidence from self-report (C)

    3. 3.

      Behaviors from group B and C are not fully accounted for by other factors (D)

Although suggestions for improving this system have appeared (Boone 2007a, b; Larrabee et al. 2007), it has been widely used in published research on the detection of feigning during neuropsychological examinations. This system also may provide useful insights into revision of the DSM framework for identifying malingering, as follows:

  1. 1.

    When forced-choice neuropsychological tests are used, it is possible to identify an objective threshold that strongly implies deliberate choice of incorrect answers (e.g., statistically significantly below chance performance) that is perhaps the most compelling evidence for feigning.

  2. 2.

    Criteria for identifying feigned symptoms may need to be modified depending on the specific type of symptom or impairment that is involved (e.g., false psychiatric versus cognitive/neuropsychological symptoms).

  3. 3.

    Consideration of the quality and magnitude of evidence for false symptom reports may guide assignment of certainty levels for the presence of feigning, such as possible, probable, or definite.

  4. 4.

    It is possible to provide a framework for identifying the presence of feigned neuropsychological deficits that relies primarily on well-validated, objective tests.

  5. 5.

    It is possible to chain results from multiple forced-choice tests to provide a probability of feigning using several different approaches, as described by Larrabee (2007a, b, c, pp. 14–26).

Detecting Feigned Physical or Somatic Symptoms

It is perhaps unsurprising that it is in the area of objective detection of feigned physical or somatic symptoms that mental health professionals have made the least progress, though Heilbronner et al. (2009) offer suggestions for symptom validity assessment in this area and directions for future study. It should be noted that various medical specialties have procedures for detection of feigned physical symptoms; Granacher has described these as “detection methods employed by physicians (that) are part of the art of medicine; they are based on the identification of nonanatomical or non-physiological presentations of apparent neurological and physical disorders” (Granacher and Berry 2008, p. 149). Lanyon (1978) included a health symptom overreporting scale on the Psychological Screening Inventory. Although some attention has been devoted to detecting feigned somatic symptoms on the MMPI-2 (Larrabee 2003, 2007a, b, c; Bianchini et al. 2008) and MMPI-2-RF (Wygant et al. 2009), much of the work in this broad area by mental health professionals has involved identification of feigned pain symptoms. Drawing on Slick et al. (1999), Bianchini et al. (2005) proposed a diagnostic framework for identifying malingered pain-related disability, as outlined in Table 3. These authors also grade the malingered pain-related disability into possible, probable, or definite with increasingly stringent requirements. Finally, they also recommend employing multiple strategies for identifying feigned pain symptoms.

Table 3 Selected proposed criteria for the diagnosis of malingered pain-related disability
Full size table

Integration of Models for Detecting Feigned Symptom Reports

Consideration of these three alternative models leads to several suggestions that might usefully be incorporated in a revised text for DSM-5. First, when substantial incentives are present for successful faking, it would be helpful to focus on identification of feigned symptoms without attempting to infer conscious intent or intended goal of the patient. As reviewed earlier, there are very few procedures or situations in which objective determination of volition or intent may be determined (even in the case of statistically significantly below chance performance on forced-choice tests, only a small minority of known feigners breaches this threshold). Therefore, the DSM-5 text should focus on identification of feigned symptoms, rather than malingering as previously defined.

Another conclusion from considering the alternative models is to acknowledge that feigned symptoms may take many forms. Thus, specifiers, such as psychiatric, cognitive/neuropsychological, or somatic/physical, should be included in a determination of feigned symptoms. It seems likely that additional potential foci for false reports may be identified, and these could be added as specifiers as the research literature grows.

A third area involves the need to employ multiple strategies for documenting the presence of feigned symptoms. This can be seen in the strategies for detecting feigned symptoms outlined by Rogers (2008a, b) and reviewed earlier, such as improbable or absurd symptoms, symptom combinations, reported versus observed symptoms, etc. In the Slick et al. (1999) and Bianchini et al. (2005) frameworks, this is illustrated by requiring documentation of feigning from both neuropsychological testing and self-report in order to identify probable feigning. Wherever possible, these criteria should be met using standardized, objective measures, although behavioral observations by two or more individuals or video recordings could also establish evidence of feigning. Cutting scores for standardized scales for determining feigned symptoms, particularly if multiple scales are involved, should be selected to minimize false-positive results. In this regard, symptom validity tests typically set cutting scores that hold false-positive rates at 5–10%.

In recognition of the varying levels of certainty for identifying feigned symptoms, a hierarchy of possible, probable, and definite modifiers seems appropriate. Of course, depending on the nature of the false symptoms, it is likely that operationalization of these modifiers will need to be tailored to each category. For example, using forced-choice cognitive tests, statistically significantly below chance performance is a useful and objective means of documenting definite feigning of cognitive/neuropsychological symptoms. At present, similar objective procedures may not be available for false psychiatric symptoms. However, video recording of a defendant who claims psychosis interacting normally with peers might achieve the level of definite feigned psychiatric symptoms.

Finally, false symptom reports could be graded by severity into mild, moderate, or severe ranges. This could be accomplished by considering the difference between claimed disability and actual functioning, or perhaps by the magnitude of the feigned deficits. This would address the apparent dimensional nature of feigning as reviewed earlier.

In summary, the present DSM criteria for identifying malingering are badly flawed at both conceptual and practical levels. Given the availability of successful models for detecting feigning of psychiatric, cognitive/neuropsychological, and somatic/physical symptoms, the DSM text on malingering should be replaced by empirically based criteria for detecting false symptom reports. Deletion of the requirement for usually subjective judgments of uncertain validity about volition and intent is recommended. Replacement text regarding identification of feigned symptoms using well-validated, objective techniques can be based on published research and divided into feigned psychiatric, cognitive/neuropsychological, or somatic/physical symptoms. Certainty about the presence of feigned symptoms could be graded into possible, probable, or definite categories. Lastly, in recognition of the apparent dimensional nature of feigning, ratings of mild, moderate, or severe could be operationalized. This proposed revision, while requiring further fleshing out for clinical application, could do much to address the weaknesses of the current DSM text on malingering and be of great potential benefit to clinicians working with patients who have substantial incentives for exaggerating or fabricating symptoms.