Utility of the 20‐Item Noise Pareidolia Task (NPT‐20) for Assessing Visuoperceptual Disturbances Associated with Complex Visual Hallucinations in Parkinson's Disease

ABSTRACT

Background

Complex visual hallucinations (VH) are a common complication of Parkinson's disease (PD). Recent studies have demonstrated relevance of face pareidolia to VH in PD and Lewy body dementia (LBD).

Objective

This study examined utility of the 20‐item Noise Pareidolia Task (NPT‐20) in assessing visuoperceptual disturbances associated with VH in PD.

Methods

Retrospective chart review included 46 consecutive PD patients who completed NPT‐20 during clinical neuropsychological evaluation.

Results

About half the sample (43%) reported VH. PD with VH made significantly more false‐positive pareidolia errors on the NPT‐20 (p < 0.0001). A cut‐off of 2 errors yielded 40% sensitivity, 100% specificity to VH; cut‐off of 1 yielded 75% sensitivity, 81% specificity. NPT‐20 was not associated with any other clinical or demographic factor. Across groups, NPT‐20 evinced moderate correlations with visuospatial functioning and visual memory.

Conclusions

Current findings support utility of the NPT‐20 for evaluating visuoperceptual disturbances associated with VH in PD.

Complex visual hallucinations (VH) are common in Parkinson's disease (PD). VH can significantly affect the quality of life and caregiver burden, particularly when insight is poor due to cognitive impairment or unreliable due to cognitive fluctuations, and when other psychiatric disturbances such as delusions are present. ¹ , ² VH are also associated with increased risk for movement disturbances, including freezing of gait and fall risk. ³ While their prevalence increases at later stages of PD and with the presence of cognitive impairment, it is well documented that VH can occur early in the disease and in the absence of cognitive impairment. ⁴ , ⁵ , ⁶ , ⁷ Current methods to evaluate for the presence of VH rely on patient or caregiver‐based questionnaires and scales. One of the limitations of this approach is that it relies on the patient's recall and insight into the VH. In addition, the stigma of a potential diagnosis of mental illness may lead to under‐reporting. ¹ Having a method to increase the suspicion of VH and further investigate the possibility of VH will have a significant impact on care and prognosis of PD.

Pareidolia is defined as perception of a specific and often meaningful image in a random or ambiguous pattern. Cloud formations shaped like animals or patterns of tree bark resembling human faces are common examples. The phenomenon is not pathologic, per se, but increased pareidolia misperceptions have been associated with VH in parkinsonian disorders. In 2012, Uchiyama et al. found increased pareidolia rates on an experimental test battery discriminated DLB from Alzheimer's dementia (AD) with 88% specificity and 100% sensitivity. ⁸ While pareidolias did not discriminate DLB with VH from those without VH, lower rates of pareidolia were seen in those treated with acetylcholinesterase inhibitor. In another study, the same group found increased pareidolia rates discriminated VH in a non‐demented PD sample. ⁹ The authors subsequently developed a more readily deployable measure using 40 black and white images, 8 of which included outlines of human faces embedded within inkblots, and 32 images containing only inkblots. ¹⁰ Pareidolia errors (ie, false positive identifications of faces in noise) demonstrated sensitivity of 81% and specificity of 92% discriminating DLB from AD, and good 30‐day test–retest reliability in DLB, ICC = 0.82. From this task, a 20‐item version (NPT‐20) containing 7 images with faces and 13 with noise was standardized and included within the National Alzheimer's Coordinating Center (NACC) Lewy body dementia module. The task instructions and stimuli can be obtained free of charge at https://naccdata.org/data-collection/forms-documentation/lbd-3. In the validation study, performance on the NPT‐20 discriminated patients with DLB from AD, as well as suspected prodromal DLB from amnestic MCI. ¹¹ NPT‐20 errors were also higher in DLB patients reporting VH. ¹² In another study, the NPT predicted conversion to DLB in patients followed for REM sleep behavior. ¹³ Application of the NPT‐20 in clinical evaluations of suspected synucleinopathies has therefore been suggested;¹¹ however sensitivity to visuoperceptual disturbances associated with VH in PD has not yet been reported.

The aim of the present investigation was to assess utility of the NPT‐20 within the context of clinical neuropsychological evaluations in PD. Specifically, we examined the sensitivity and specificity of the NPT‐20 to VH in PD. We also explored associations between NPT‐20 performance with clinical characteristics and demographic factors, as well as other neuropsychological measures. Our hypotheses were that NPT error rates would be higher in PD patients with VH (PD VH+) than those without (PD VH−), and that performance would be associated with measures of visuospatial functioning.

Methods

This retrospective chart review was approved by the Human Subjects Protection Program at the Medical University of South Carolina, Pro00062817. Participants were 46 consecutive patients followed in a Movement Disorders clinic for PD at an academic medical center who were referred for neuropsychological evaluation. PD diagnosis was made according to UK Brain Bank criteria by a fellowship‐trained movement disorder neurologist. Participants were referred for neuropsychological evaluation of cognitive concerns as noted by the patient, informant, or referring neurologist.

In keeping with prior clinical practice, history of complex VH was assessed during clinical interview and rated as present or absent. Complex VH were defined by self‐reported perception of figures in the visual field (central or peripheral) that were stable enough to be identified or described in terms of shape and size, and subsequently determined to be hallucinatory by the patient and/or care partner. Presence/extracampine experiences, passage hallucinations (without shape or form), and illusions were considered to be simple hallucinations and not recorded. Hypnogogic/hypnopompic hallucinations and history of VH circumscribed to delirium, other illness, or acute medication side effects (eg, narcotics) were also not recorded as complex VH. History of dream enactment behavior suggestive of REM sleep behavior disorder (RBD) was similarly queried and documented.

Comprehensive neuropsychological assessment included a semi‐structured test battery administered by the first author. The NPT‐20 was completed at the end of the evaluation. Diagnosis of Mild Cognitive Impairment in PD (PD‐MCI) and dementia (PDD) followed Movement Disorders Society Task Force criteria ¹⁴ , ¹⁵ and did not consider NPT‐20 performance. Group differences based on VH were evaluated with independent sample t‐test or non‐parametric statistical tests as indicated by distributions, and associations of NPT‐20 errors with demographic, clinical, and cognitive measures were assessed with Spearman‐Brown rank‐ordered correlations.

Results

All patient participants (n = 46) accurately responded to the practice item and completed the NPT‐20. Total administration time ranged from 2–4 min. About half the sample (43%) reported complex VH. All VH reported in this sample involved living things, with people, pets, and “creatures” or “bugs” being most common. Only one individual reported both VH and auditory hallucinations. The majority (70%) of those with VH reported preserved insight and reality‐testing; however, veracity of this report was suspect in enough cases that further classifications were not made on this basis.

As shown in the Table 1, presence of complex VH was associated with greater disease severity, HY Stage = 2.60 versus, 2.26, t(44) = 2.183, p = 0.037, but no other clinical or demographic factors. NPT‐20 performance was not associated with any clinical or demographic factor, and was not associated with global cognitive function (DRS‐2). However, PD patients reporting VH made significantly more false‐positive pareidolia errors on the NPT‐20 (p < 0.0001). A cut‐off of 2 errors yielded 40% sensitivity and 100% specificity to VH; cut‐off of 1 yielded 75% sensitivity and 81% specificity. The distribution of pareidolia errors by group is shown in Figure 1. Notably, all 8 PD patients with 2 or more pareidolia errors also reported a history suggestive of RBD. As shown in the table, group differences were not statistically significant for any other cognitive test administered, though increased pareidolia errors were associated with worse performances on tests of visuospatial perception and visual memory.

TABLE 1.

Demographic, clinical, and cognitive measures for Parkinson's disease without visual hallucinations (PD VH−) and with visual hallucinations (PD VH+)

	Measure	PD VH− (n = 26)	PD VH+ (n = 20)	Group Comparison		Correlation with Pareidolia Errors
				t‐statistic	p‐value	rho	p‐value
Demographics	Age	67. 38 (8.02)	68.25 (5.71)	−0.409	0.685	0.142	0.346
	Gender (males)	21 (81%)	17 (85%)	X 2 = 0.141	0.707	0.048	0.75
	Education	15.54 (3.20)	15.68 (3.30)	−0.045	0.964	0.196	0.202
Clinical	Duration of Illness	8.43 (4.63)	10.53 (5.98)	−1.27	0.209	0.157	0.322
	H&Y Stage	2.6 (.53)	2.26 (.50)	2.183	0.037	0.238	0.11
	REM Sleep Behavior	14 (54%)	14 (70%)	X 2 = 0.926	0.372
	LEDD	824 (533)	829 (605)	−0.03	0.976	−0.019	0.902
Cognitive Status	PD Normal (n)	8 (31%)	4 (19%)	X 2 = 0.126	0.528	0.183	0.228
	PD‐MCI (n)	17 (65%)	14 (66%)
	PD Dementia (n)	1 (4%)	2 (10%)
NPT‐20	Face Errors (max = 7)	.24 (.51)	.55 (.88)	U = 214.5	0.191
	Pareidolia Errors	1.19 (.40)	1.85 (1.90)	U = 95	<0.001
Global Cognition	DRS‐2 (raw, max = 144)	138.94 (3.50)	138.08 (4.06)	0.629	0.534	−0.066	0.724
Visuospatial	JOLO (raw, max = 30)	23.92 (5.73)	21.41 (4.34)	1.539	0.131	−0.345	0.024
	NAB Figure	52.25 (15.33)	51.67 (11.20)	0.096	0.924	−0.336	0.137
Attention	Digit Span Forward	51.61 (8.15)	52.95 (7.02)	−0.584	0.562	0.204	0.174
	NAB Num/Let A Errors	49.14 (9.87)	43.63 (12.33)	1.53	0.135	−0.281	0.087
Processing Speed	NAB Num/Let A Time	45.27 (13.19)	40.81 (14.22)	0.996	0.326	0.029	0.862
	Trails A	49.28 (11.04)	44.6 (13.84)	1.263	0.214	−0.168	0.271
Language	NAB Naming	56.5 (2.26)	53.84 (6.71)	1.817	0.077	−0.001	0.995
	Animal Fluency	45.38 (11.63)	45.3 (12.00)	0.024	0.981	−0.121	0.424
Executive Functions	Digit Span Backward	44.38 (7.76)	46.7 (6.21)	−1.091	0.283	0.153	0.309
	Phonemic Fluency	45.24 (9.20)	45.15 1(13.13)	0.027	0.979	−0.181	0.234
	Trails B	43.92 (10.85)	41.84 (13.00)	0.57	0.572	−0.155	0.32
	D‐KEFS 20Q	42.70 (8.6)	46.67 (13.27)	1.076	0.29	0.051	0.772
Verbal Memory	HVLT Total Learning	42.04 (8.23)	41.11 (11.83)	0.304	0.763	0.131	0.401
	HVLT Delay Recall	41.24 (9.97)	44.64 (11.07	0.257	0.799	−0.05	0.751
Visual Memory	BVMT Total Learning	40.08 (11.98)	34.53 (11.29)	1.561	0.126	−0.343	0.023
	BVMT Delay Recall	42.28 (11.48)	35.94 (13.04)	1.709	0.095	−0.404	0.006

Note: Values reflect mean (SD) unless otherwise noted. Cognitive measures are provided as demographically‐corrected T‐scores (mean = 50, SD = 10). Group comparisons reflect independent sample t‐tests unless indicated. Spearman rank‐order correlation coefficients with pareidolia errors are provided.

Abbreviations: BVMT, Brief Visuospatial Memory Test; D‐KEFS, Delis‐Kaplan Executive Function System; DRS‐2, Mattis Dementia Rating Scale; HVLT, Hopkins Verbal Learning Test; LEDD, Levodopa Equivalent Daily Dose; NAB, Neuropsychological Assessment Battery.

FIG 1.

Distribution of Pareidolia errors on the NPT‐20 for Parkinson's disease patients without visual hallucination (PD VH−) and with visual hallucinations (PD VH+).

Discussion

Results from this study support clinical utility of the NPT‐20 as measure of visuospatial functioning associated with complex VH in PD. Performance on the NPT‐20 correlated with other standardized visuospatial measures and discriminated PD patients reporting complex VH with good sensitivity and specificity. Importantly, NPT‐20 performance was not associated with overall cognitive status, other clinical factors, or demographics characteristics.

The assessment of illusions to determine the presence of VH may appear orthogonal as they are by definition different. Hallucinations are defined by visual percept not associated with a real object and illusions as a real object perceived incorrectly. ¹⁶ However, the clarity of these definitions may be difficult to translate to clinical practice where whether the perception was triggered a stimulus is difficult to determine, particularly in subjects with cognitive impairment. More importantly, illusions and VH often coexist in PD and DLB, suggesting the possibility of a common pathophysiological mechanism. ⁸ Therefore, harnessing the ability to evaluate for illusions in the clinic and research settings can lead to a better understanding of the nature of psychosis in PD.

In our study, the correlation between impairment in visuospatial abilities and the presence of VH and errors in the 20‐NPT suggests impaired visuospatial abilities play a role in psychosis. Whether illusions, such as pareidolia, precede or are concomitant with the emergence of VH remains to be determined. In any case, the detection of illusions may serve as an objective marker of disease progression or treatment response. In addition to the NPT‐20, other methods to elicit visual misperception such as the Scene Pareidolia Test and the Bistable Percept Paradigm, have been suggested as surrogate markers for VH in DLB ¹⁷ , ¹⁸ and could be applied in PD. Finally, the finding that all subjects who had more than 2 errors reported both VH and RBD warrants further investigation, as pareidolias in the context of RBD is associated with worse cognitive performance and a higher likelihood of developing symptoms of parkinsonism and dementia. ¹³

There are several limitations to this retrospective chart review. First, history of complex VH was obtained through clinical interview rather than a standardized inventory. As such, it is possible that another clinician might obtain or interpret responses differently. Use of a formal scale that included frequency and severity ratings, such as the Neuropsychiatric Inventory (NPI), ¹⁹ would also be useful in determining clinical meaningfulness of highly elevated pareidolia error rates. Similarly, determination of whether patients had RBD was done through interview and not polysomnography, which is the gold‐standard for diagnosis. Finally, given that these evaluations were done in the context of clinical assessment, the generalizability of the results to subjects without cognitive concerns may be limited. Despite these limitations, current results are highly consistent with previous studies and encourage application in both clinical and research settings. For instance, the NPT‐20 could be used to identify PD patients more likely to experience psychosis with dopamine agonist therapy or benefit from antipsychotic intervention. In clinical trials, the NPT‐20 could be applied as a criterion for inclusion or exclusion, depending on the indication. Prior findings of reduced pareidolia in LBD treated with donepezil warrant consideration as a secondary efficacy outcome in PD psychosis trials. Finally, given the brevity of administration and scoring time, the inclusion of the NPT‐20 in comprehensive longitudinal studies of PD neuropsychiatric phenomenology, such as the PPMI, is suggested. ²⁰

Author Roles

(1) Research project: A. Conception, B. Organization, C. Execution; (2) Statistical Analysis: A. Design, B. Execution, C. Review and Critique; (3) Manuscript Preparation: A. Writing of the first draft, B. Review and Critique.

T.H.T.: 1A, 1B, 1C, 2A, 2B, 3B.

F.R.P.: 1B, 2C, 3A.

Disclosures

Ethical Compliance Statement: This study was approved by the Institutional Review Board for Human Research at the Medical University of South Carolina, Pro00062817. This study was a retrospective chart review at an academic medical center; informed consent was not required. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this work is consistent with those guidelines.

Funding Sources and Conflict of Interest: No specific funding exists for this work. The authors declare that there are no conflicts of interest that relate to the research covered in the article, regardless of date.

Financial Disclosures for the previous 12 months: Dr. Turner has received financial compensation for consultation related to clinical trial design from WCG‐VeraSci (now WCG Clinical Endpoint Solutions) and Scion Neurostim. Dr. Rodriguez‐Porcel declares that there are no additional disclosures to report.