Perceived cognitive impairment is related to internalizing psychopathology but unrelated to objective cognitive performance among nongeriatric adults presenting for outpatient neuropsychological evaluation

Abstract Objective: This study investigated the relationship between perceived cognitive impairment, objective cognitive performance, and intrapersonal variables thought to influence ratings of perceived cognitive impairment. Method: Study sample comprised 194 nongeriatric adults who were seen in a general outpatient neuropsychology clinic for a variety of referral questions. The cognition subscale score from the WHO Disability Assessment Schedule served as the measure of perceived cognitive impairment. Objective cognitive performance was indexed via a composite score derived from a comprehensive neuropsychological battery. Internalizing psychopathology was indexed via a composite score derived from anxiety and depression measures. Medical and neuropsychiatric comorbidities were indexed by the number of different ICD diagnostic categories documented in medical records. Demographics included age, sex, race, and years of education. Results: Objective cognitive performance was unrelated to subjective concerns, explaining <1% of the variance in perceived cognitive impairment ratings. Conversely, internalizing psychopathology was significantly predictive, explaining nearly one-third of the variance in perceived cognitive impairment ratings, even after accounting for test performance, demographics, and number of comorbidities. Internalizing psychopathology was also highly associated with a greater discrepancy between scores on perceived and objective cognitive measures among participants with greater cognitive concerns. Clinically significant somatic symptoms uniquely contributed to the explained variance in perceived cognitive impairment (by ∼13%) when analyzed in a model with internalizing symptoms. Conclusions: These findings suggest that perceived cognitive impairment may be more indicative of the extent of internalizing psychopathology and somatic concerns than cognitive ability.


Introduction
Approximately one in nine adults in the united States report some degree of cognitive impairment or decline in cognitive functioning (CDC, 2021;Stewart, 2012).Perceived cognitive impairment can have adverse effects on self-beliefs and attitudes and precipitate health-related anxiety that results in individuals seeking extensive medical care (Kessler et al., 2012;Mol et al., 2008;Roh et al., 2021).Such perceived difficulties are routinely elicited as part of a clinical interview in neuropsychology clinics, presumably because symptoms are informative of underlying neurocognitive dysfunction.yet, there is limited evidence that perceived cognitive symptoms are reliable indicators of cognitive impairment as measured objectively by neurocognitive tests (Buckley et al., 2013;Caramelli & Beato, 2008;Mendes et al., 2008;Minett et al., 2008).Most research studies report small correlations between subjective and objective indices of cognition, generally finding that patients overestimate the severity of impairment (for review, see Burmester et al., 2016;Crumley et al., 2014;Mendonça et al., 2016;Mitchell et al., 2014).It has been suggested that, in geriatric samples, perceived cognitive impairment may be an early sign of neurodegeneration that manifests before cognitive decline becomes apparent on objective testing (Burmester et al., 2016;Jessen et al., 2014).Conversely, among general clinical populations of younger adults without known history of neurodegenerative disease, endorsement of subjective concerns may be underlain by different etiologies (CDC, 2021, Stewart, 2012).The factors that influence perceived cognitive impairment in younger general clinical populations are not well understood.
A better understanding of intrapersonal factors that influence the relationship between perceived and objectively measured cognitive impairment can help streamline the evaluation process, improve diagnostic clarity, and provide more appropriate treatment recommendations and referrals to specialty clinics (e.g.see Daglas-Georgiou et al., 2022).Such improvements may secondarily reduce financial or emotional burden on patients and families who seek extensive medical workup (Horsfall et al., 2010;Ton et al., 2017).Furthermore, understanding the relationship between subjective and objective cognitive impairment may impact epidemiological research where subjective cognitive impairment is commonly used as a proxy measure of cognitive function (Stewart, 2012).Consequently, the main aim of the current study was to examine the relationship between perceived and objectively measured cognitive impairment in relation to relevant intrapersonal factors in a general sample of nongeriatric patients presenting for comprehensive outpatient neuropsychological evaluation.

Internalizing psychopathology and perceived cognitive impairment
Several correlates of perceived cognitive impairment have been described in the literature, with internalizing psychopathology arguably demonstrating the strongest relationship with reported cognitive symptoms (e.g.Hill et al., 2016;Hughes et al., 2019;Roh et al., 2021;Taylor et al., 2018).Specifically, individuals with greater internalizing symptoms of depression and anxiety tend to overestimate their degree of cognitive impairment when referenced against objective cognitive test scores (Burmester et al., 2016).
Although severe clinical depression has well-described cognitive sequelae (Conradi et al., 2011), the self-perception of cognitive functioning may be distorted by negative appraisals and expectancies and maladaptive coping strategies (for review, see Hill et al., 2016;Zimprich et al., 2003).Depression symptoms are associated with a negativity bias in which individuals become overly vigilant of normative lapses in cognition, perceiving them as pathological (Beck, 1987).The emotional distress regarding cognitive impairment may, in turn, exacerbate the depressive symptoms and thus result in greater overestimation of impairment (Bhang et al., 2020;Hill et al., 2016).These negative appraisals are amplified when individuals believe their cognitive symptoms interfere with daily functioning (Hill et al., 2016).
Comorbid anxiety and depression have been reported to have a cumulative impact on perceived cognitive impairment (Hill et al., 2016).Anxiety also may independently contribute to the discrepancy between perceived and objective cognitive functioning (Hill et al., 2016).Patients with anxiety who demonstrate intact cognitive performance sometimes report subjective cognitive impairment if their anxiety symptoms interfere with daily activities of living, potentially implying that these patients may mistake functionally disruptive internalizing symptoms for cognitive symptoms (Hill et al., 2016).Similar to depression, patients with anxiety may demonstrate negative biases and appraisals of their cognitive processing that result in greater perceived impairment than is observed on objective measures (Dux et al., 2008;Pearman et al., 2014).

Medical comorbidities and perceived cognitive impairment
Studies have also documented an association between the number of medical comorbidities and perceived cognitive impairment (Margolis et al., 2021;Stein et al., 2021;Taylor et al., 2020).Taylor et al. (2020) reported that having a history of three or more chronic medical conditions is associated with increased subjective cognitive impairment.number of comorbidities appears to influence perceived cognitive impairment ratings independent of age, race, or ethnicity (Gupta, 2021;Taylor et al., 2020).It is possible that certain types or medical diagnoses may particularly influence the endorsement of perceived cognitive impairment.Caracciolo et al. (2013) found that musculoskeletal, respiratory, and urological diseases are associated with higher endorsement of cognitive symptoms.Perceived cognitive impairment has also been reported in patients diagnosed with heart disease, stroke, diabetes, chronic obstructive pulmonary disease, asthma, arthritis, and chronic kidney disease (Biessels et al., 2006;Bugnicourt et al., 2013;national Academies of Sciences, engineering, and Medicine, 2017;Roh et al., 2021;Taylor et al., 2020).Most of these studies have been carried out in diagnostically homogeneous samples and did not account for objectively measured cognitive impairment.More research is needed to understand how medical comorbidity influences the perceived cognitive symptoms in mixed clinical samples.

Somatic symptoms and perceived cognitive impairment
The relationship between cognitive and somatic symptoms, such as pain and fatigue, has received much attention in the literature.Pain and fatigue have not only been demonstrated to correlate with perceived cognitive impairment (Cockshell & Mathias, 2014;Cvejic et al., 2016), but to also influence the relationship between perceived impairment and internalizing psychopathology (esmael et al., 2021;Rasouli et al., 2019), as well as the discrepancy between subjective and objective indices of cognition (Cvejic et al., 2016;Hughes et al., 2019).Fatigue in particular is endemic to several psychiatric and neurologic diseases.In neurologic samples, fatigue may influence perceived cognitive impairment when comorbid depression is present, or independently (Cockshell & Mathias, 2014;Hughes et al., 2019;Kinsinger et al., 2010;Kobelt et al., 2019).Kinsinger et al. (2010) found that solely treating fatigue in patients with multiple sclerosis resulted in reduced subjective endorsement of cognitive impairment while cognitive performance remained stable.
Mechanisms underlying the link between somatic symptoms and perceived cognitive impairment remain ill-defined.It has been suggested that pain and fatigue result in excessive interoceptive monitoring, which reduces externally-directed attention.Reduced externally-directed attention can hinder routine cognitive processes.Memory perfectionism and increased self-monitoring that is associated with pain and fatigue may then cause one to overinterpret these labored experiences (e.g.Siciliano et al., 2021;Teodoro et al., 2018).Patients who report unusually high somatic symptoms may also develop negative appraisals about their own cognitive processing that are similar to those with depression and anxiety (e.g.Cvejic et al., 2016).

Demographics and perceived cognitive impairment
Demographic factors including age, sex, level of education, race and ethnicity are proxies for social determinants of health and have been reported to influence the self-perception of cognitive impairment.Higher subjective cognitive impairment ratings and greater discrepancy between subjective symptoms and objective cognitive impairment have been disproportionately described among Hispanic/latinx and non-Hispanic Black individuals residing in the united States of America (Brown & Patterson, 2020;Gupta, 2021;Jang et al., 2022), with some evidence suggesting that these associations may be moderated by the severity of depressive symptoms (Ayers et al., 2020;Rodriguez et al., 2021).Several studies have reported negative associations between years of formal education perceived cognitive impairment (Aghjayan et al., 2017;Brown & Patterson, 2020;Stewart et al., 2008) but findings are mixed (e.g.Benito-leon et al., 2010;Mendes et al., 2008).Findings regarding the relationship between perceived cognitive impairment in women versus men are also mixed (e.g.Tomita et al., 2014;Wang et al., 2004).One study reported higher subjective ratings of cognitive impairment among women undergoing the menopausal transition (Schliep et al., 2022).Transgender and non-binary individuals have been shown to report higher ratings of perceived cognitive impairment than both men and women (Brown & Patterson, 2020;Flatt et al., 2021;lambrou et al., 2022); however, this relationship may be attenuated when accounting for depression (Flatt et al., 2021).lastly, findings are mixed regarding the relationship between age and perceived cognitive impairment ratings among nongeriatric adult patients (Rowell et al., 2016;Zapater-Fajari et al., 2022).More research is needed to better understand the nature of the associations between demographic factors and perceived cognitive impairment in general clinical populations.

Summary and limitations of the literature
Overall, extant research suggests that self-ratings of cognitive symptoms among younger (i.e.nongeriatric) adult patient populations may not be strongly related to objectively measured cognitive impairment.At the same time, internalizing psychopathology, medical comorbidities, somatic symptoms, and some demographic factors may significantly influence the severity of perceived cognitive impairment.The generalizability of these research findings is limited as most studies were conducted either with diagnostically homogeneous samples such as patients with major depressive disorder, chronic fatigue syndrome, multiple sclerosis, cancer, traumatic brain injury, or functional neurologic disorder (e.g. esmael et al., 2021;Hill et al., 2016;Hughes et al., 2019;Hutchinson et al., 2012;Rasouli et al., 2019;Teodoro et al., 2018), or with circumscribed patient populations such as military veterans or first responders (e.g.Donnelly et al., 2018;Stein et al., 2021).likewise, the relationship between perceived cognitive impairment and medical comorbidities has mostly been studied in older adults (e.g.Aarts et al., 2011;Gupta, 2021;lee et al., 2020;Margolis et al., 2021).yet, comorbidities, internalizing psychopathology, and somatic symptoms are pervasive among younger adult patients, such as those who typically present for outpatient neuropsychological evaluation (Goldberg & McGee, 2011;Rosenthal et al., 2008;Stein et al., 2021;Violan et al., 2014).Findings informing the relationship between perceived cognitive impairment and objective cognitive test performance are limited by the use of brief cognitive assessment instruments, with few studies administering batteries that assess cognition across multiple cognitive domains (Jessen et al., 2014).This decreases the sensitivity of detecting a significant relationship and may increase the probability of type II error.Furthermore, the methodologies used in studies with nongeriatric clinical populations are limited in scope.Most studies have explored correlates of perceived cognitive impairment; but such findings are not necessarily informative regarding which factors influence the accuracy of perception of cognitive impairment.Controlling for objective cognitive performance in statistical analysis or quantifying the discrepancy between the perception of cognition and performance on objective cognitive measures as an outcome variable is one way to address this limitation.Another gap in the current understanding of subjective cognitive impairment is whether certain intrapersonal factors are characteristic of patients who endorse a high level of perceived cognitive impairment.This is a clinically relevant group as they may be experiencing substantial emotional distress that may impede treatment efforts (Kessler et al., 2012;Mol et al., 2008;Roh et al., 2021), and because these patients may exert an undue burden on healthcare resources (Županič et al., 2022).

Current study
The current study utilized a retrospective cross-sectional design to examine the relationships between perceived cognitive impairment, objective cognitive test performance, and relevant intrapersonal factors in a mixed clinical sample of nongeriatric adults presenting for an outpatient neuropsychological evaluation.The first aim was to ascertain the strength of the relationship between perceived cognitive impairment and objective performance on a comprehensive battery of cognitive tests.The second aim was to explore the associations between perceived cognitive impairment and relevant intrapersonal factors including internalizing psychopathology, demographics, and number of medical and neuropsychiatric comorbidities.A supplementary aim using a smaller subsample of patients examined the role that clinically significant somatic symptoms may play in the relationship between the same intrapersonal variables, perceived cognitive impairment, and cognitive test performance.A second supplementary aim using a subsample of patients who endorsed a moderate-to-severe level of perceived cognitive impairment was to ascertain which intrapersonal factors most strongly influence the discrepancy between perceived and objectively measured cognitive impairment.
Based on prior findings from various clinical samples, we hypothesized that perceived cognitive impairment would be positively and more strongly associated with internalizing psychopathology than the other intrapersonal factors, including objective cognitive performance, demographics, comorbidities, and somatic symptoms.

Participants and procedures
The sample comprised 194 adult outpatients, 100 of whom identified as female.The majority of the sample (79%) identified as non-Hispanic White, 11% as non-Hispanic Black, 5% as Hispanic/latinx, and 5% identified as a mixed race/ethnicity.The average age was 47.21 years (SD = 16.35;range 18-74) and average educational attainment was 15.41 years (SD = 2.39; range 9-20).Consistent with the sociodemographic makeup of our sample, mean premorbid Full-Scale IQ was estimated at the 63 rd percentile (Wechsler Adult Intelligence Scale, 4 th edition Test of Premorbid Functioning standard score = 105).As a whole, our sample performed 0.3 standard deviations below the normative mean (z-score range = −3.02-1.54) on the composite measure of objective cognition.Participants presented with diverse medical histories, with the average participant having four mental health comorbidities (range 0-6) and eight medical comorbidities (range 0-19).Details regarding demographics and clinical characteristics of the sample are provided in Table 1 and Supplementary Table S1.
All participants underwent outpatient clinical neuropsychological evaluations at a large Midwestern academic medical center for a wide range of referral questions.evaluations were conducted by licensed staff neuropsychologists between 2017-2020.evaluation procedures included a record review (e.g.medical, psychiatric, educational and vocational records when appropriate), clinical interview, and a comprehensive assessment of cognitive and psychological functioning.This project was approved by the institutional review board at northwestern university Feinberg School of Medicine.
Data were harvested from participants' electronic medical records with the help of an enterprise data warehouse algorithm supplemented with manual record review.The algorithm was set to only extract data from participants who completed the perceived cognitive impairment rating measure used for this study.Once the data were extracted from the participants' electronic medical records, we excluded cases (n = 208) who were not administered the neurocognitive tests used in the composite measure of objective cognitive performance (see below for details).An additional two cases were excluded due to inadequate symptom or performance validity.Specifically, in accordance with accepted practice guidelines (Jennette et al., 2022;Sweet et al., 2021), participants who scored below the cutoff on two or more freestanding or embedded performance validity tests (Medical Symptom Validity Test, Green, 2004;Test of Memory Malingering, Tombaugh, 1997;Word Memory Test, Green et al., 1999; California Verbal learning Test Forced Choice Recognition; Reliable Digit Span) and/ or two or more embedded symptom validity tests (indices from the Minnesota Multiphasic Personality Inventory, Ben-Porath & Tellegen, 2008;Personality Assessment Inventory, Morey, 2007a) were excluded from the analyses.For the reasons discussed above, we additionally excluded four cases who had a diagnosis of probable neurodegenerative disease documented in their medical record at the time of data collection.Supplementary analyses examining the role of somatic symptoms were conducted with a subset of the participants (n = 83) who completed the Personality Assessment Inventory (Morey, 2007a), which is a broadband self-report measure of personality and psychopathology that includes an index of somatic symptoms.The subsample did not significantly differ from the overall sample in terms of demographic or clinical characteristics (see Table 1).Additional supplementary analyses examined the correlates of the discrepancy between perceived and objective cognitive impairment in a subset of the participants (n = 71) with moderate-severe ratings of perceived cognitive impairment.This subsample also did not significantly differ from the overall sample in terms of demographic or clinical characteristics except for the perceived cognitive impairment ratings (see Table 1).

Intrapersonal variables
Types of medical and mental health comorbidities were coded based on diagnostic categories as defined by the International Classification of Diseases, 10 th Revision (ICD − 10; World Health Organization, 1992).We recorded the number of different diagnostic categories recorded in participants' electronic medical records to index the breadth of medical and mental health comorbidities at the time of neuropsychological evaluation (see Table 1 and Supplementary Table 1 for a complete listing).
The severity of internalizing psychopathology was indexed by computing a standardized composite of scores from the following anxiety and depression measures: Beck Depression Inventory-Second edition (Beck et al., 1996), Geriatric Depression Scale (yesavage, 1988), Beck Anxiety Inventory (Beck et al., 1988), Geriatric Anxiety Inventory (Pachana et al., 2007), Generalized Anxiety Disorder 7-item (Spitzer et al., 2006), and State-Trait Anxiety Inventory Trait scale (Spielberger et al., 1983).Specifically, raw scores from each of these measures were z-transformed and averaged together to compute a single index of internalizing psychopathology.For supplementary analyses, participants were classified as endorsing clinically significant somatic symptoms if they attained T-scores ≥70 on the Personality Assessment Inventory Somatic Complaints scale (Morey, 2007a;2020).
Objective cognitive performance was measured via a standardized composite of scores from 16 performance-based measures 1 that were selected to assess the full range of abilities across multiple neurocognitive domains.Because the current study examined a referred clinical sample, participants did not receive a fixed neurocognitive battery.Scores from different measures of the same cognitive ability were used interchangeably (i.e.D-KeFS number Sequencing and Trail Making Part A; D-KeFS number-letter Sequencing and Trail Making Part B; D-KeFS letter and Category Fluency and Controlled Auditory Word Association Test letter and Category Fluency, respectively; neuropsychological Assessment Battery naming and Boston naming Test).To control for method variance, each of the measures that were used interchangeably were regressed on perceived cognitive impairment ratings to ensure that the combined measures yielded similar β coefficients.none of these measures had more than a two decimal point difference in their β coefficients.Because the primary aim of the current study was to examine relationships between subjectively and objectively measured cognitive impairment that is reflective of actual clinical practice, demographically corrected standardized scores (as opposed to raw scores) from each test were included in the creation of the composite score.In accordance with practice guidelines (American Academy of Clinical neuropsychology, 2007), the types of demographic adjustments applied reflected recommendations from test developers and neuropsychologists' clinical and ethical judgment.Although some tests had standardized scores that were age-corrected and others had standardized scores adjusted for additional demographic factors (e.g.sex, race, and/or education), preliminary analyses revealed no significant (p > .05)associations between demographics (age, sex, race, years of educational) and the standardized scores from each cognitive test.There was also no significant association between demographics and the overall composite score.All the standardized scores from each test were then linearly transformed into z-scores to create a uniform metric.none of the test z-scores were differently weighted.Consequently, the individual test z-scores were averaged together to produce a single index of objective cognitive performance.

Outcome variables
Perceived cognitive impairment was indexed using the total score on the Cognition (understanding and Communicating) subscale of the World Health Organization Disability Assessment Schedule 2.0 (WHODAS 2.0; Üstün et al., 2010).Prior research has demonstrated that this measure can sufficiently assess perceived cognitive impairment among individuals with various neuropsychiatric and medical conditions and ethno-racial backgrounds (Üstün et al., 2010).The six-item subscale assesses the perceived level of difficulty with several aspects of cognitive processing including concentration, learning and memory, decision-making, and language.Responses are rated on a one to five likert scale with responses indicating no difficulty, mild, moderate, severe, and extreme/cannot do, respectively.Total scores ranged from 6 to 30 with higher scores indicating greater severity of perceived cognitive impairment.
Discrepancy between perceived and objective cognitive impairment was indexed by computing standardized residuals from a linear regression of objective cognitive impairment scores on the WHODAS 2.0 Cognition scores.As such, greater residual scores in either direction indicate a greater discrepancy between objective test scores and perceived cognitive impairment.This discrepancy method was used to exclusively capture cognitive concerns that were not observed on objective cognitive measures, improving our ability to examine perceived impairment that is not reflective of actual cognitive impairment.

Data analyses
All assumptions were met and post-hoc power analyses indicated that all findings had an observed power greater than 80%.each statistical model was created a priori to address study questions based on the existing literature.
We addressed our first aim by conducting a Pearson product-moment bivariate correlation to ascertain the strength of the relationship between subjectively and objectively measured cognitive impairment.We addressed our second aim by conducting a hierarchical multiple linear regression to explore the associations between perceived cognitive impairment and relevant intrapersonal factors, above and beyond the effects of objective cognitive performance.Treating perceived cognitive impairment scores as the dependent variable, we structured the regression hierarchy in three steps to parse out the unique contributions of the intrapersonal factors.In Step 1, we entered the objective cognitive test score composite.In Step 2, we examined the extent to which demographics (age, sex, race, and years of education) and number of comorbidities influenced perceived cognitive impairment ratings, above and beyond the effect of cognitive performance.In Step 3, we examined the extent to which internalizing psychopathology influenced the ratings of perceived cognitive impairment, above and beyond the effect of cognitive performance, demographics, and number of comorbidities.
For the supplementary analysis, we expanded upon the second study aim by additionally accounting for the presence of clinically elevated somatic symptoms.Treating perceived cognitive impairment ratings as the dependent variable, we entered objective cognitive impairment in Step 1, demographics (age, sex, and education) and number of comorbidities in Step 2, level of somatic symptoms in Step 3, and internalizing psychopathology in Step 4. We were unable to include every demographic variable in this model because using all the variables would have required a larger sample size than we had available in order to maintain sufficient statistical power.We used the best subset selection method, to determine which demographic variables could be removed.This method uses a variable selection approach to derive the most parsimonious combination of demographic variables to be included in the model (for review, see Kassambara, 2018).The analysis revealed that race was non-contributory, and was therefore not included in this model.We included somatic symptoms in Step 3 to examine the extent that such concerns were associated with perceived cognitive impairment above and beyond the effects of cognitive performance, demographics, and breadth of comorbidities.This also allowed us to examine whether participants' ratings of perceived cognitive impairment were still associated with the relevant intrapersonal factors when adjusting for somatic symptoms.
We addressed our second supplementary aim by conducting a hierarchical linear regression to ascertain which intrapersonal factors influence the discrepancy between subjective and objective cognitive impairment among individuals who endorsed moderate-to-severe levels of perceived cognitive impairment (i.e.those who provided WHODAS scores in the 18-30 range).Treating subjective-objective discrepancy scores as the dependent variable, we structured regression hierarchy in two steps.In Step 1, we entered demographics (age, sex, race, and years of education) and number of comorbidities.In Step 2, we examined the extent to which internalizing psychopathology influenced the subjective-objective discrepancy scores, above and beyond the effect of demographics and number of comorbidities.

Correlates of perceived cognitive impairment
no interaction effects were observed between the covariates within any of the models described below, indicating that the statistically significant intrapersonal variables were uniquely associated with perceived cognitive impairment and subjective-objective discrepancy scores.
The objective cognitive composite scores were not significantly correlated with perceived cognitive impairment scores (r = .08,p > .05),explaining less than 1% of the variance in perceived cognitive impairment ratings.Table 2 presents results of hierarchical multiple linear regression models examining the associations between perceived cognitive impairment and intrapersonal variables.The overall model accounting for cognitive performance, demographics, comorbidities, and internalizing psychopathology explained a substantial and significant proportion (~36%) of the variance in perceived cognitive impairment ratings (F[9,184] = 11.53,p < .001).neither demographics (age, sex, education, and race) nor the number of comorbidities significantly contributed to the model (F[8,185] = 1.09, p = .369),explaining less than 5% of the variance in perceived cognitive impairment ratings.However, the model significantly improved when further accounting for internalizing psychopathology, (F[9, 184] = 11.47,p < .001)suggesting that a greater severity of internalizing psychopathology is correlated with higher ratings of perceived cognitive impairment.About 31% of the variance in perceived cognitive impairment ratings was uniquely explained by internalizing psychopathology scores (β = .59,95% CI [-0.05, 1.23]).

The role of somatic symptoms
Supplementary analyses indicated that the addition of somatic symptoms to the model explained a unique proportion of variance in perceived cognitive impairment above and beyond the variance explained by objective cognitive test scores, demographics (age, sex, and education), and the number of comorbidities (R 2 = .19,F[9,72] = 3.15, p = .003).Participants who endorsed clinically significant somatic symptoms also endorsed higher perceived cognitive impairment ratings, with approximately 13% of the variance in perceived cognitive impairment ratings independently explained  by somatic symptom scores (β = .40,95% CI [-1.87, 2.67]).The model significantly improved again when further accounting for internalizing psychopathology (R 2 = .44,F[10,71] = 7.44, p < .001).Internalizing psychopathology uniquely explained 25% of the variance in perceived cognitive impairment in the expanded model (β = .57,95% CI [-0.57, 1.69]).Taken together, models accounting for participants' cognitive performance, demographics, comorbidities, internalizing psychopathology, as well as somatic symptoms explained ~44% of the variance in perceived cognitive impairment ratings (see Supplementary Table S2 for details).
To further parse out the role of somatization, we examined the relationship between intrapersonal factors and perceived cognitive impairment separately in subgroups of participants who endorsed clinically elevated versus subclinical (i.e.PAI Somatic Complaints T-score < 70) levels of somatic symptoms.In both subsamples, internalizing psychopathology was the only variable that was significantly associated with perceived cognitive impairment.However, in the subsample of participants who endorsed elevated somatic symptoms, internalizing psychopathology uniquely explained 38.40% of the variance in perceived cognitive impairment ratings, whereas in the non-elevated subsample internalizing psychopathology only explained 18.94% of the unique variance in perceived cognitive impairment ratings.Interestingly, no interaction effect was found between internalizing psychopathology and somatic symptoms [F(1, 80) = 1,66, p > .05],indicating that somatization explains a small but significant proportion of variance in perceived cognitive impairment ratings independent of depression and anxiety symptoms.

Correlates of the Perceived-Objective discrepancy in participants with high perceived cognitive impairment
Table 3 presents findings from a hierarchical sequence of multiple linear regression models examining the associations between subjective-objective discrepancy scores and intrapersonal variables in a subsample of individuals with high perceived cognitive impairment.neither demographics nor the number of comorbidities significantly contributed to the overall model (F[7, 63] = 1.46, p < .20),explaining about 4% of the variance in the subjective-objective discrepancy scores.However, the model significantly improved when further accounting for internalizing psychopathology (F[8, 62] = 4.12, p < .001),suggesting that more severe internalizing psychopathology was associated with greater discrepancy between perceived cognitive impairment and objective cognitive performance.About 22% of the variance in subjective-objective discrepancy scores was uniquely explained by internalizing psychopathology (β = .52,95% CI [0.27, 0.77]).
Overall, the combined models accounting for participants' demographics, number of comorbidities, and internalizing psychopathology explained a substantial and significant proportion (~26%) of the variance in the discrepancy between subjective and objective cognition for participants with high perceived cognitive impairment ratings.

Discussion
This study is among the first to examine the associations between perceived cognitive impairment and objectively measured cognitive performance in a mixed  transdiagnostic sample of nongeriatric adults seen for outpatient neuropsychological evaluation.Our results are consistent with published findings from various clinical samples demonstrating weak associations between scores on subjective and objective measures of cognition (Buckley et al., 2013;Caramelli & Beato, 2008;Mendes et al., 2008;Minett et al., 2008).Our hypothesis that perceived cognitive impairment is more strongly associated with internalizing psychopathology than any other intrapersonal factor was supported by the current findings, and in line with prior research (Daglas-Georgiou et al., 2022;Kinsinger et al., 2010;Zapater-Fajarí et al., 2022).In our study, internalizing psychopathology was also strongly associated with a greater discrepancy between scores on subjective and objective measures of cognition among participants who reported moderate to severe cognitive concerns.These findings provide indirect evidence for internalizing psychopathology as a potential psychological mechanism underlying cognitive symptom over-endorsement in younger adult patients.Our results extended prior findings (Cvejic et al., 2016;Hughes et al., 2019) by providing evidence that clinically significant somatic symptoms may also uniquely contribute to the endorsement of cognitive concerns, albeit not as strongly as internalizing psychopathology.The model that explained the greatest proportion of variance in perceived cognitive impairment included both internalizing psychopathology and somatic symptoms, suggesting that the combination of these factors may be a particularly important marker of high cognitive symptom endorsement in the clinic.Contrary to some published studies (Brown & Patterson, 2020;Margolis et al., 2021;Stein et al., 2021;Taylor et al., 2020;Zapater-Fajari et al., 2022), perceived cognitive impairment in our sample was not significantly related to demographics or the breadth of medical or neuropsychiatric comorbidities, after controlling for objective cognitive performance.
Current findings may highlight differences in the utility of assessing perceived cognitive impairment across various clinical samples and practice settings.emerging clinical research from geriatric settings suggests that among older adults, cognitive concerns may signal early signs of neurodegeneration before cognitive impairment is detected by objective neurocognitive tests (e.g.Burmester et al., 2016).Our findings suggest that among younger adults seeking neuropsychological evaluation, subjective cognitive concerns may signal the presence of internalizing psychopathology independent from neurocognitive dysfunction or underlying medical or neuropsychiatric comorbidities.While the current findings are uninformative regarding the mechanisms linking internalizing psychopathology with cognitive concerns, they do suggest that cognitive concerns may not be as reliable indicators of cognitive impairment among younger adults as assumed by many clinicians.
notably, performance on objective measures of cognition administered in a controlled clinical environment may have limited sensitivity and ecological validity as an index of everyday cognition.The possibility that some of the participants in our study may have reported subtle symptoms that were not captured by standardized neurocognitive measures, cannot be ruled out.Prior research has indeed suggested that adults in the prodromal and early stages of neurodegenerative disease may report symptoms before they are detected by objective measures (Burmester et al., 2016;Jessen et al., 2014).Although the current study excluded participants with an established diagnosis of probable neurodegenerative disease, some participants in the sample may have had neurodegenerative pathology given that the sample included participants up to age 74.Therefore, it is possible that some adults in our sample may have had undetectable neurodegenerative pathology causing changes in cognition that were not observed using traditional neuropsychological assessments.However, it should be noted that the current study did not find any association between perceived cognitive impairment and age, a primary risk factor for neurodegeneration (Hou et al., 2019).Furthermore, internalizing psychopathology explained a substantial and similar amount of variance in both the ratings of perceived cognitive impairment and the discrepancy between perceived symptoms and objective testing across participants of all ages, symptom severity, and cognitive ability.Internalizing psychopathology also accounted for a substantial proportion of variance in perceived cognitive impairment above and beyond the variance accounted by cognitive test scores across patients of all levels of cognitive impairment.
A more plausible explanation for the discrepancy between cognitive concerns and cognitive test performance in our sample is that subjective and objective measures of cognition may index different constructs in younger adults.This notion is supported by research showing that even healthy individuals tend to endorse some cognitive symptoms on self-report measures, which are not observed on objective measures (Mäntylä et al., 2010;Martin, 1983;Tucker-Drob, 2011).Our findings suggest that self-report measures of cognitive impairment may capture psychological processes that are uniquely associated with internalizing psychopathology.For instance, prior research has suggested that internalizing psychopathology may cause patients to over-pathologize cognitive deficits or innocuous cognitive lapses (Beck, 1987;Dux et al., 2008;Hill et al., 2016;Pearman et al., 2014).Another explanation may be that patients mistake functionally disruptive symptoms of depression or anxiety with cognitive symptoms (Hill et al., 2016).
Internalizing psychopathology tends to co-occur with variables that are also thought to influence perceived cognitive concerns, such as pain and/or fatigue, suggesting that the relationship between cognitive concerns and internalizing symptoms may be mediated by somatic symptoms stemming from physical illness or somatic reactivity.However, our findings evidenced no significant interaction between internalizing psychopathology and somatic symptoms or medical comorbidities, suggesting that clinically elevated somatic concerns and internalizing psychopathology may increase perceived cognitive impairment through discernable, if not independent, mechanisms.This conceptualization is consistent with prior research demonstrating a discordance between subjective and objective measures of cognition among clinically relevant populations such as patients with chronic fatigue syndrome (Cockshell & Mathias, 2014).Researchers have speculated that somatization may preoccupy one's attentional ability and make mentally taxing tasks more effortful on a daily basis, which may not be observed on objective testing or in depressed/anxious individuals (Siciliano et al., 2021;Teodoro et al., 2018).
Our results did not support previously reported associations between medical comorbidities and perceived cognitive impairment among individuals who report experiencing internalizing symptoms.One explanation may be that cognitive concerns among younger adults are influenced primarily by mechanisms that are unique to depression or anxiety, and prior associations with medical diagnoses may have been due to methodological differences or reporting bias.For example, the majority of prior studies assessed medical comorbidities via self-report, whereas we harvested diagnostic data directly from medical records.It is possible that patients who over-endorse cognitive symptoms also over-report the extent of their medical histories.neither did our analyses evidence previously reported associations between perceived cognitive impairment and demographic factors.This may in part reflect the relatively demographically homogeneous makeup of our sample; this is addressed further in the limitations section.

Clinical and research implications
The current study findings contribute to the growing body of literature regarding the limited utility and validity of cognitive symptom ratings in clinical practice and social-behavioral research.Taken together, findings suggest that perceived cognitive impairment may not be a good indicator of underlying neurocognitive pathology.Researchers may wish to rely on objective measures of cognition when investigating factors associated with cognitive impairment rather than relying on self-reported cognitive concerns as a proxy for cognitive status (Stewart, 2012).Indeed, perceived cognitive impairment is not sufficient to denote the presence of a neurocognitive disorder or perhaps any degree of cognitive disability.For clinical neuropsychologists, these findings should not discourage the solicitation of cognitive concerns.Rather, these findings deemphasize the value in using subjective symptom report to guide decisions regarding etiologic rule-outs during the clinical interview and the subsequent selection of evaluation procedures.neuropsychologists may wish to rely more heavily upon other aspects of clinical history and observable disease signs to guide their case conceptualization and test selection.However, an improved understanding of the role of perceived cognitive impairment in patients who also present with internalizing psychopathology may help clinicians educate patients and their families regarding the differences between symptoms experienced in everyday life and objective neurocognitive assessment findings.
These findings may have additional implications for primary care, mainly encouraging physicians to consider brief objective assessments of mental status (e.g.Mini Mental Status examination, Montreal Cognitive Assessment) in lieu of self-reported cognitive concerns, when deciding whether to refer young adults for comprehensive neuropsychological evaluation.A downstream benefit of cognitive screening at the level of primary care is the reduction of financial and/or emotional burden on patients and families who seek extensive medical workup (Horsfall et al., 2010;Ton et al., 2017).

Limitations and directions for future research
The current study is not without limitations.First, our study sample comprised predominately non-Hispanic White participants with a high education attainment level who sought elective outpatient care.It is likely that these results may not generalize to more diverse populations, including individuals of non-Hispanic Black and Hispanic/ latinx descent who are disproportionately affected by social inequalities and comorbid conditions, yet less likely to discuss their cognitive condition with a medical provider (Gupta, 2021).Future research should corroborate these findings with a more diverse sample to better clarify how practitioners can account for sex, education, race, and ethnicity when interpreting perceived cognitive impairment.Individuals with lower educational attainment will be an important cohort to study given reports of higher prevalence of cognitive concerns in this group (e.g. Brown & Patterson, 2020).
The current investigation utilized a broad composite index of internalizing psychopathology that did not allow for a detailed construct-level exploration of potential associations between specific features of depression and anxiety and perceived cognitive impairment.Future studies may elucidate the relationships between cognitive symptom endorsement and the affective, cognitive, and physiological features of depression and anxiety.This may also help parse out any overlapping variance between internalizing symptomology and somatic symptoms.
The measure of perceived cognitive impairment utilized in the current study (i.e.WHODAS 2.0), while extensively validated and widely used, does not allow for a detailed analysis of perceived dysfunction in the different domains of cognition (e.g.memory, attention, language).There is some evidence that certain types of reported cognitive symptoms are differentially related to objective performance (Cockshell & Mathias, 2014;Hahm et al., 2020;Rasouli et al., 2019).Future research may explore the relationship between domain-specific cognitive concerns and cognitive test performance.There are other existing measures of perceived cognitive impairment (e.g.Chelune et al., 1986;Farias et al., 2008;Rami et al., 2014) that may be better suited for these additional analyses.
Because the current study utilized a clinical dataset, the cognitive assessment battery was not uniform, resulting in cases with missing tests.It is possible that excluding participants who were not administered the tests used in our objective cognitive composite score could bias our findings.However, there were no systematic differences in cases with missing tests, and these data were missing completely at random based on formal statistical analysis (little's Test) and visual inspection.There were no significant differences between the subset of participants who completed the measure of somatic symptoms and the entire sample.However, this may have been a biased subsample because practitioners may be more inclined to administer a personality inventory rating for patients with prominent mood or behavioral symptoms.
Similarly, because the main aim of the current study design was to maximize clinical relevance to the neuropsychology practitioner, we utilized demographically corrected neurocognitive test scores in the current analyses, with the choice of demographic corrections having been clinically determined.In order to uniformly correct for the influence of all theoretically relevant demographic factors, we included demographics in Step 1 of the regression models, which may have resulted in over-adjustment and slightly biased estimates of the coefficients.nonetheless, we thought that the advantage of accounting for all relevant demographics in our models of cognitive scores outweighed the risk of model saturation.
lastly, the strongest combination of intrapersonal factors in our analyses explained 44% of the variance in perceived cognitive impairment ratings, meaning that the majority of the variance is explained by other unelucidated factors.Future researchers may wish to expand upon the current findings by examining the role of insight, perceived fatigue and pain, specific types of medical and psychiatric comorbidities, and other theoretically relevant variables (Cockshell & Mathias, 2014).

Conclusions
Patients commonly present for a neuropsychological evaluation with comorbid cognitive concerns and internalizing psychopathology, regardless of the degree of objectively measured cognitive impairment.This patient population has traditionally been regarded as complex and challenging in terms of diagnosis and treatment.The marked increase in neuropsychology referrals over the past several years driven in part by patients with persistent post-acute sequelae of COVID-19 has further highlighted the need for an improved understanding of the role of perceived cognitive dysfunction in neuropsychological care.Current study findings provide further evidence for a lack of significant association between perceived cognitive impairment and objectively measured cognitive performance in nongeriatric adults presenting for outpatient neuropsychological evaluation.Instead, findings suggest that the endorsement of cognitive concerns is predominately influenced by internalizing psychopathology, and to a lesser extent somatization, even when controlling for cognitive ability, medical and neuropsychiatric comorbidities, and demographics.Among a subsample of patients who endorsed particularly high levels of perceived cognitive impairment, the discrepancy between subjective and objective indices of cognition was again influenced by internalizing psychopathology more than any other factor, further supporting the role of depression and anxiety as important determinants of cognitive concerns in this population.We hope that these findings will help improve the neuropsychological evaluation process as well as referrals to specialty clinics for younger adults seeking outpatient care.We also hope these findings will spur further research regarding the associations between subjective and objective indices of cognition in this population.

Table 1 .
sample characteristics and descriptive statistics.

Table 2 .
hierarchical linear regression analysis of intrapersonal variables and perceived cognitive impairment.

Table 3 .
hierarchical linear regression analysis of intrapersonal variables and perceived-objective discrepancy in high perceived cognitive impairment reporters.
1. Boston naming Test, Delis-Kaplan executive Function System (D-KeFS) number Sequencing, letter-number Sequencing, letter Fluency, and Category Fluency, California Verbal learning Test Total and long Delay Free Recall, Controlled Auditory Word Association Test letter and Category Fluency, neuropsychological Assessment Battery naming, Trail Making Test Part A and B, and the Wechsler Adult Intelligence Scale, 4 th edition (WAIS-IV) Digit Span Total, Coding, Similarities, and Matrix Reasoning.Generally speaking, processing speed was measured via the WAIS-IV Coding, Trail Making Test Part A, and D-KeFS number Sequencing; attention and working memory was measured via the WAIS-IV Digit Span; executive functioning was measured via the Trail Making Test Part B, D-KeFS letter-number Sequencing, and WAIS-IV Similarities; language was measured via the neuropsychological Assessment Battery naming, Boston naming Test, and Verbal Fluency tests from the Controlled Auditory Word Association Test and D-KeFS; visuospatial ability was measured via the WAIS-IV Matrix Reasoning; learning and memory was measured via the California Verbal learning Test Total and long Delay Free Recall, respectively (lezak et al., 2012).