The following case examples illustrate the heterogeneity in negative response bias displayed by noncredible mTBI test takers. The first case example involves a noncredible test taker claiming cognitive dysfunction secondary to complicated mTBI who demonstrated noncredible performance only on memory PVTs. However, as is shown in subsequent examples described below, some noncredible mTBI patients adopt feigning strategies other than poor memory performance.

Case #1: Feigned Memory/Attention Impairment: The patient was a 65-year-old male who had worked in food service and who had completed an Associate’s degree; he was tested 7 years post complicated mTBI. Glasgow Coma Scale (GCS) was initially 13/14, but improved to 15 while still in the emergency department; a severe brachial plexus injury left the patient with an inability to use his left arm. Initial brain imaging was normal, but the following day, a small subarachnoid hemorrhage was noted as well as small petechial hemorrhages. On examination the patient failed all PVTs involving verbal (Warrington Words: total = 27, 214″ [60]; WMS-III Logical Memory Effort Equation = 37 [90]; RAVLT Effort Equation = 4 [63]; RO/RAVLT discriminant function = −2.538 [64]; Rey Word Recognition = 5 [79, 80]) and visual (Digit Symbol recognition = 47 [62]; RO Effort Equation = 44 [61, 82]) memory, but scored within normal limits on PVTs involving attention (Digit Span Age-Corrected Scaled Score (ACSS) = 10, RDS = 9; 3-digit time = 1″ [69]), visual perception (Picture Completion Most Discrepant Index = 6 [84]), and processing speed (Dot Counting Test E-score = 7 [65]; b Test E-score = 36 [78, 88]). Standard cognitive scores were normal with the exception of variable scores in verbal and visual memory and some timed tasks.

Case #2: Feigned language impairment: The patient was a 36-year-old female with approximately 12 years of education who was suing for damages sustained in a motor vehicle accident 3 years previous to exam; her presentation is described in more detail in Cottingham and Boone [91]. She claimed mTBI and presented with prominent language symptoms that continued to evolve and worsen over time, including decreased word-retrieval, dysarthria, telegraphic speech, loss of prosody, difficulty deciphering written words, emergence of an Eastern European-type “foreign accent,” and English as a Second Language (ESL) grammatical errors (“How you say?”). However, interestingly, on examination, articulation errors were inconsistent (e.g., “turkey” was sometimes pronounced correctly and at other times was pronounced as “tur’keen” and “den tur’kee’un”) and overlearned number labels were not used (e.g., “eleven” was pronounced “one-one”). Neuropsychological test performance was normal with the exception of lowered scores on some measures involving verbal abilities and processing speed. The patient passed PVTs involving verbal (Warrington Words = 47 [60]; RAVLT Effort Equation = 17 [63]; Rey Word Recognition = 9 [79, 80]; RO/AVLT discriminant function = 0.91 [64]) and visual (RO equation = 60 [61, 82]) memory, attention/numbers (Dot Counting E-score = 11 [65]; Digit Span ACSS = 10, RDS = 10 [69]), and motor speed (Finger Tapping = 44 [85]), but failed indicators involving verbal repetition (time to repeat 3 digits = 3″, time to repeat 4 digits = 4.5″ [69]), letter identification (b Test E-score = 481.8 [78, 88]) and finger identification (Finger Agnosia errors = 5 [92]). Had only memory PVTs been employed, there would have been no psychometric evidence of noncredible symptoms.

Case #3: Feigned Processing Speed and Sensory impairment: The patient was a 42-year-old appliance repairman with 12 years of education who was suing in the context of claimed mTBI sustained in a workplace accident. On medical evaluation, GCS was 15 and the patient was alert and oriented, neurological examination was intact, brain imaging was normal, and the patient’s claims of loss of consciousness were not independently verified. On neurocognitive exam 3 years after injury, the patient failed PVTs involving processing speed (Warrington Words time = 369″ [60]; time per digit on forward Digit Span = 1.3″ [69]), sensory function (Finger Agnosia errors = 5 [92]) and overlearned math skills (Dot Counting errors = 4 [65]), but scored within normal limits on measures involving visual memory (Rey 15-item Memorization Test plus recognition = 26 [81]; RO effort equation = 62 [61, 82]), verbal memory (Rey Word Recognition Test = 12 [79, 80]; Warrington Words = 46 [60]; RAVLT/RO discriminant function = 0.096 [64]; RAVLT effort equation = 16 [63]), attention (Digit Span ACSS = 9, RDS = 8 [69]), rapid letter identification (b Test E-score = 74.4 [78, 88]), visual perception (“most discrepant” index on Picture Completion = 5 [84]), and motor speed (finger tapping = 53.7 [85, 93]), and standard cognitive scores were normal with the exception of scores on measures of processing speed and sensory/motor tasks.

Use of Multiple PVTs in Combination. It is imperative that multiple PVTs be administered during a neuropsychological evaluation because response bias may not be constant and, as illustrated in the cases above, noncredible test takers adopt differing strategies in their approach to feigning [66]. Further, multiple PVTs are needed because none have 100 % sensitivity and specificity. Therefore, if only one PVT is administered and a passing score is obtained, it cannot be definitively concluded that the patient is performing to true capability because the individual may fall within the subset of noncredible individuals who in fact pass the test. Likewise, failure on a single PVT administered cannot be used as definitive evidence of feigning in that cut-offs are set to allow a small subset of credible patients to fail (the exception would be cut-offs set to 100 % specificity).

Rather than confusing the situation, administration of increasing numbers of PVTs actually provides the clinician with more confidence in conclusions regarding credibility of performance [94]. That is, research has shown that failure on two PVTs meets or exceeds 95 % specificity, and three failures results in near perfect specificity (≥98.5 %) [5, 95–97]. We frequently encounter mTBI plaintiffs who fail 8, 9, 10, 11, and even 12 PVTs, which is incontrovertible evidence of negative response bias.

Because embedded PVTs do not involve additional test administration time, their use best realizes the goal of substantially increasing available PVT data. The following three mTBI cases illustrate how examination of embedded PVT data moved the likelihood of symptom feigning from probable to definitive.

Case #4: The patient was a 39-year-old male attorney involved in a motor vehicle accident 5 years previous; he did not sustain loss of consciousness, head was atraumatic on evaluation, he was alert/fully oriented, and he did not complain of head symptoms in the emergency department; brain CT and MRI were normal. He subsequently returned to work, often billing 15 h in a single day. On neurocognitive examination, he failed two of five dedicated performance validity tests; specifically, he failed the Warrington Recognition Memory Test (Words: 29/50; 11′44″) [60] and the Rey Word Recognition Test (4) [79, 80], but passed Dot Counting (E-score = 15) [65], the b Test (E-score = 59) [79, 88], and Rey 15-item plus recognition (24) [81].

However, he was also not credible on six of six standard cognitive tests sensitive to feigned performance: Finger Tapping dominant hand (38; cut-off for TBI males) [85], Digit Span (ACSS = 5, RDS = 7, mean time to recite 3 digits = 3″) [69, 98], RO Equation (30) [61, 82], RAVLT indices (Effort Equation = 4 [63]; RO/RAVLT discriminant function = −1.775 [64]), Picture Completion Most Discrepant Index (1) [84], and Finger Agnosia errors (4) [92].

Results of standard neurocognitive testing showed impaired scores in visual perception/constructional skill, verbal and visual memory, and finger dexterity, while borderline scores were documented in attention/processing speed; no scores were within normal limits.

What did the embedded PVTs contribute? Failure on two of five dedicated PVTs would be associated with 95 % specificity, but failure on eight total indicators increases specificity to 100 %. Further, the patient failed free-standing PVTs only involving verbal memory, but the failed embedded indicators revealed he was also underperforming in attention, processing speed, visual perception/memory, and motor sensory function, and showed that his symptom feigning strategy involved underperformance in most neurocognitive domains.

Case #5: This patient was a 52-year-old male general contractor with 12 years of education who was involved in a motor vehicle accident 3 years previous to exam. He reported loss of consciousness of unknown duration in the collision, but medical records indicated that he did not in fact lose consciousness and was ambulating at the scene, and he did not have any head complaints in the emergency department. The patient returned to work briefly, and then claimed cognitive symptoms interfered with ability to run his business. On neurocognitive examination, he failed one of four dedicated PVTs; specifically, he failed the Warrington Recognition Memory Test (Words = 39; 190″) [60], but passed the Rey Word Recognition Test (7) [79, 80], the Dot Counting Test (E-score = 10) [65], and the Rey 15-item plus recognition (30) [81].

However, he failed five of seven standard cognitive tests sensitive to feigned performance, namely, Finger Tapping dominant hand (37, cut-off for head injured males) [85], Digit Symbol recognition (57) [62], RO Equation (49) [61, 82], RAVLT indices (Effort Equation = 8 [63]; RO/RAVLT discriminant function = −1.538 [64]), and Picture Completion Most Discrepant Index (2) [84], but passed Finger Agnosia errors (3) [92] and Digit Span variables (ACSS = 10, RDS = 10, mean time to recite 3 digits = 1″) [69].

On standard neurocognitive testing, average scores or higher were obtained in attention, math skills, problem-solving/reasoning, visual spatial/constructional skill, language (word-retrieval), and visual memory, while low average scores were found in processing speed, borderline scores were present in verbal recall, and impaired range performance was documented in motor dexterity.

What did the embedded PVTs contribute? Failure on one dedicated PVT is associated with only 59 % specificity, but failure on six total indicators increases specificity to 100 %. Additionally, the failed dedicated PVT only involved verbal memory, but the embedded indicators revealed the patient was also underperforming in processing speed, visual perception/memory, and motor function.

Case #6: This 28-year-old male employed in computer sales and attempting to complete a college degree was involved in a motor vehicle accident 3.5 years previous to testing. He exhibited amnesia for the event, but GCS was 15 at the scene, and brain imaging normal. He returned to school and work full-time, but claimed symptoms interfered with ability to function, although there was no change in grades from pre- to post-injury and no poor work evaluations post-injury.

On neurocognitive examination, the patient failed one of five dedicated PVTs; specifically, he failed only the b Test (E-score = 1,157) [78, 88], while passing the Warrington Recognition Memory Test (Words = 46) [60], Rey Word Recognition Test (9) [79, 80], Dot Counting Test (E-score = 14) [65], and Rey 15-item plus recognition (29) [81].

However, the patient also exhibited noncredible performance on three of seven standard cognitive tests sensitive to feigned performance, including Finger Agnosia errors (8) [92], the RO/RAVLT discriminant function (−0.798) [64], and Picture Completion Most Discrepant Index (2) [84], but passed the RAVLT Effort Equation (13) [63], Finger Tapping dominant hand (52) [85], Digit Symbol recognition (138) [62], RO effort equation (59) [61, 82], and Digit Span variables (ACSS = 13, RDS = 13, mean time to recite 3 digits = 1″) [69].

On standard neurocognitive tests, average scores or better were documented in basic attention, constructional skill, visual memory, executive skills, language, and right hand motor dexterity, while low average/average scores were noted in processing speed, low average performance was found in left hand dexterity, and impaired scores were observed in verbal memory.

What did the embedded PVTs contribute? Like case #5, failure on a single free-standing PVT would be associated with a false-positive identification rate of 41 %, but failure on four total indicators increases specificity to 100 %. The single PVT failure occurred on a measure of processing speed/overlearned information, but the additional failed embedded PVTs also showed that the patient was underperforming in visual perception, verbal memory, and sensory function.

In cases 4 through 6, extra information was obtained, and specificity was increased, at no “extra cost” in terms of test administration time. The use of multiple PVTs provides more confidence in conclusions, and rather than complicating test interpretation, data from many PVTs bring results into “sharp focus.” Data from multiple PVTs best protect credible patients from being labeled as noncredible, particularly when ≥3 PVT failures are required.

Ideally, it would be preferable if each cognitive measure included an embedded performance validity indicator so that data regarding credibility could be gathered in “real time.” The explosion of literature on embedded PVTs [99–101] shows that we are on well on the way to achieving this goal.

Personality Inventories (MMPI-2-RF, PAI, MCM-III). In addition to PVTs requiring cognitive performance, personality inventories can provide information regarding cognitive symptom over-report in individuals presenting with mTBI. In particular, literature exists on the Minnesota Multiphasic Personality Inventory-2-Restructured Form (MMPI-2-RF) [102], the Personality Assessment Inventory (PAI) [103], and the Millon Clinical Multiaxial Inventory, Third Edition (MCMI-III) [104].

The MMPI-2-RF [102], the recently published and psychometrically advanced version of the MMPI-2 [105], appears to show particular promise in identifying noncredible symptom report in mTBI. A recent study conducted by Youngjohn and colleagues [106] examined litigants with claimed mTBI (n = 55), complicated mTBI (n = 13), and moderate/severe TBI (n = 14). Thirty-four of the entire sample failed at least one cognitive PVT (noncredible group), and 48 failed none (credible group). The authors found that FBS-r (Symptom Validity scale) successfully discriminated credible from noncredible litigants of all TBI severity levels, but also that the Neurologic Complaints (NUC), Head Pain Complaints (HPC), Gastrointestinal Complaints (GIC), and Malaise (MLS) scales predicted PVT failure, with the most variance in neurocognitive PVT performance accounted for by MLS. Interestingly, mTBI litigants were more likely to produce higher scores on HPC than were their complicated mTBI and moderate/severe TBI counterparts.

Jones and Ingram [107] examined several of the validity scales on the MMPI-2 [105] and MMPI-2-RF [102] in a mixed sample of 288 active duty military personnel, approximately 90 % of whom had sustained mTBI. A subset of 117 participants was judged to be engaging in response bias. The authors reported that that FBS-r, HHI (Henry-Heilbronner Index) [108], and RBS (Response Bias Index) [109] were better at discriminating between credible and noncredible participants than were the Fs (Infrequent Somatic Responses) scale and the F-family scales from the MMPI-2.

Several of the above-described cases illustrate the usefulness of the MMPI-2-RF [102] in mTBI evaluation. In case #3, scores on MMPI-2-RF validity scales indicated that the patient was over-reporting psychiatric, physical, and cognitive symptoms in a noncredible manner (F-r = 115T; Fs = 115T; FBS-r = 90T; RBS = 114T; RC1 = 99T; MLS = 87T; GIC = 80T; HPC = 85T; NUC = 91T; COG = 96T) not explained by random responding or inability to comprehend test questions (VRIN-r = 58T; TRIN-r = 50 T). Similarly, in case #4, MMPI-2-RF validity and clinical scales revealed physical, cognitive, and psychiatric symptom over-report (FBS-r = 108T; RBS = 101T; F-r = 106T; RC1 = 95T; MLS = 87T; GIC = 96T; HPC = 78T; NUC = 91T; COG = 80T), and of interest, the low score on VRIN-r (42T), indicated that he was more careful/consistent in completing the protocol than the typical test taker; his extreme carefulness in completing the inventory would not likely be possible if his low cognitive scores were accurate. In case #5, MMPI-2-RF validity scales did not indicate over-report, but the fact that the patient was able to complete the MMPI-2-RF in 1 h with a low VRIN-r score (39T) would argue that cognitive function was likely intact. In case #6, physical/cognitive, and psychiatric symptom over-report was indicated (FBS-r = 89T; RBS = 90T; F-r = 83T; RC1 = 74T; MLS = 87T; GIC = 72T; HPC = 65T; NUC = 86T; COG = 86T). Thus, in most of the cases, MMPI-2-RF data provided critical information regarding veracity of patient symptom report.

A search of the literature revealed two studies investigating usefulness of the MCMI-III [104] in measuring credibility of symptom report in mTBI. Aguerrevere and colleagues [110] examined the response patterns of mild (n = 76; 71 % of sample) and moderate–severe TBI patients (n = 31) on the MCMI-III; the groups were collapsed after analyses showed that injury severity was not associated with differences in MCMI–III scale elevations. Most of the sample had secondary gain, and approximately half met Slick et al. [111] criteria for malingered neurocognitive dysfunction (presence of external incentive, PVT failure, and performances upon evaluation are markedly discrepant from functional abilities). The authors found that all three validity scales (Disclosure, Desirability, and Debasement) accurately discriminated between malingerers and credible patients. Using the Disclosure, Desirability, and Debasement scales in combination (cut-offs of ≥67BR, ≤54BR, and ≥71BR, respectively), specificity was 96 % with sensitivity reaching 64 %. Less promising results were observed in an earlier study of compensation-seeking individuals, 94 % of whom were claiming TBI of unreported severity [112]. Thirty to 34 percent produced PVT failure (TOMM or RDS), but MCMI–III sensitivity was only 1.9 % for the Desirability Scale (cut-off of <24), 2.9 % for the Disclosure Scale (cut-off of ≥85), and 9.5 % for the Debasement Scale (cut-off of ≥85).

Whiteside et al. investigated the PAI [103] validity scales and cognitive PVTs in a mixed clinical sample (n = 222) [113], but mTBI patients were not examined separately and it unclear how many mTBI patients were included. Additionally, it is unclear how many of the mTBI patients were evaluated in the context of secondary gain, although the authors report that approximately 9 % of their entire sample had external incentive and approximately 11 % failed the TOMM [67, 114]. The authors examined the relationships between validity scales and TOMM performance and found that the Negative Impression Management (NIM) scale and the Infrequency (INF) scale were significantly related to poor performance on the TOMM. In a subsequent study by this group employing the same sample [114], clinical scale elevations and TOMM Trial 2 performance were examined. Only the Somatic Complaints (SOM) scale was significantly related to TOMM Trial 2 performance. Among subscales on the SOM scale, the Conversion subscale (SOM-C) was most associated with PVT performance. A T score of >87 on the SOM scale detected 76 % of those patients exhibiting response bias on the TOMM, at 93 % specificity. In fact, cases #3 and #6 above obtained scores of 92T and 93T, respectively, on the Somatic Complaints scale.

In conclusion, emerging data indicate an important role of personality inventories in documentation of symptom invalidity in the evaluation of mTBI.