Impaired self-assessment in schizophrenia: Why patients misjudge their cognition and functioning

Lack of insight or “unawareness of illness” occurs within a set of self-assessment problems commonly seen in schizophrenia.¹ In the clinical domain, people who do not realize they are ill typically are unwilling to accept treatment, including medication, with potential for worsened illness. They also may have difficulty self-assess­ing everyday function and functional potential, cognition, social cognition, and attitude, often to a variable degree across these domains (Table 1).^1-3

Self-assessment of performance can be clinically help­ful whether performance is objectively good or bad. Those with poor performance could be helped to attempt to match their aspirations to accomplishments and improve over time. Good performers could have their functioning bolstered by recognizing their competence. Thus, even a population whose performance often is poor could ben­efit from accurate self-assessment or experience additional challenges from inaccurate self-evaluation.

This article discusses patient characteristics associated with impairments in self-assessment and the most accu­rate sources of information for clinicians about patient functioning. Our research shows that an experienced psy­chiatrist is well positioned to make accurate judgments of functional potential and cognitive abilities for people with schizophrenia.

Patterns in patients with impaired self-assessment
Healthy individuals routinely overestimate their abilities and their attractiveness to others.⁴ Feedback that deflates these exag­gerated estimates increases the accuracy of their self-assessments. Mildly depressed individuals typically are the most accurate judges of their true functioning; those with more severe levels of depression tend to underestimate their competence. Thus, sim­ply being an inaccurate self-assessor is not “abnormal.” These response biases are con­sistent and predictable in healthy people.

People with severe mental illness pose a different challenge. As in the following cases, their reports manifest minimal cor­relation with other sources of information, including objective information about performance.

CASE 1
JR, age 28, is referred for occupational therapy because he has never worked since graduat­ing from high school. He tells the therapist his cognitive abilities are average and intact, although his scores on a comprehensive cog­nitive assessment suggest performance at the first percentile of normal distribution or less. His self-reported Beck Depression Inventory (BDI) score is 4. He says he would like to work as a certified public accountant, because he believes he has an aptitude for math. He admits he has no idea what the job entails, but he is quite motivated to set up an interview as soon as possible.

CASE 2
LM, age 48, says his “best job” was managing an auto parts store for 18 months after he earned an associate’s degree and until his second psychotic episode. His most recent work was approximately 12 years ago at an oil-change facility. He agrees to discuss employment but feels his vocational skills are too deteriorated for him to succeed and requests an assessment for Alzheimer’s disease. His cognitive perfor­mance averages in the 10th percentile of the overall population, and his BDI score is 18. Tests of his ability to perform vocational skills sug­gest he is qualified for multiple jobs, including his previous technician position.

Individuals with schizophrenia who report no depression and no work history routinely overestimate their functional potential, whereas those with a history of unsuccessful vocational attempts often underestimate their functional potential. Inaccurate self-assessment can contrib­ute to reduced functioning—in JR’s case because of unrealistic assessment of the match between skills and vocational poten­tial, and in LM’s case because of overly  pessimistic self-evaluation. For people with schizophrenia, inability to self-evaluate can have a bidirectional adverse impact on func­tioning: overestimation may lead to trying tasks that are too challenging, and under­estimation may lead to reduced effort and motivation to take on functional tasks.

Metacognition and introspective accuracy
“Metacognition” refers to self-assessment of the quality and accuracy of performance on cognitive tests.^5-7 Problem-solving tests— such as the Wisconsin Card Sorting test (WCST), in which the person being assessed needs to solve the test through performance feedback—are metacognition tests. When errors are made, the strategy in use needs to be discarded; when responses are correct, the strategy is retained. People with schizo­phrenia have disproportionate difficulties with the WCST, and deficits are especially salient when the test is modified to measure self-assessment of performance and ability to use feedback to change strategies.

“Introspective accuracy” is used to describe the wide-ranging self-assessment impairments in severe mental illness. Theories of metacognition implicate a broad spectrum, of which self-assessment is 1 component, whereas introspective accuracy more specifically indicates judg­ments of accuracy. Because self-assessment is focused on the self, and hence is intro­spective, this conceptualization can be applied to self-evaluations of:
   • achievement in everyday functioning (“Did I complete that task well?”)
   • potential for achievement in everyday functioning (“I could do that job”)
   • cognitive performance (“Yes, I remembered all of those words”)
   • social cognition (“He really is angry”).

 

Domains of impaired introspective accuracy
Everyday functioning. The 3 global domains of everyday functioning are social outcomes, productive/vocational out­comes, and everyday activities, including residential independence/support for peo­ple with severe mental illness. Two areas of inquiry are used in self-assessing everyday functioning: (1) what are you doing now and (2) what could you do in the future? For people with schizophrenia, a related question is how perceived impairments in everyday functioning are associated with subjective illness burden.

People with schizophrenia report illness burden consistent with their self-reported disability, suggesting their reports in these domains are not random.⁸ Studies have consistently found, however, that these patients report:
   • less impairment on average in their everyday functioning than observed by clinicians
   • less subjective illness burden com­pared with individuals with much less severe illnesses.

Their reports also fail to correlate with cli­nicians’ observations.9 Patients with schizo­phrenia who have never been employed may report greater vocational potential than those employed full-time. Interestingly, patients who were previously—but not currently—employed reported the least vocational potential.¹⁰ These data suggest that experience may be a factor: individuals who have never worked have no context for their self-assessments, whereas people who are persistently unemployed may have a perspective on the challenges associated with employment.

In our research,⁹ high-contact clinicians (ie, case manager, psychiatrist, therapist, or residential facility manager) were bet­ter able than family or friends to generate ratings from an assessment questionnaire that correlated with performance-based measures of patients’ ability to perform everyday functional skills. The ratings were generated across multiple functional status scales, suggesting that the rater was more important than the specific scale. We concluded that high-contact clinicians can generate ratings of everyday functioning that are convergent with patients’ abilities, even when they have no information about actual performance scores.

Cognitive performance. When self-reported cognitive abilities are correlated with the results of performance on neuro­psychological assessments, the results are quite consistent. Patients provide reports that do not correlate with their objective performance.¹¹ Interestingly, when clini­cians were asked to use the same strategies as patients to generate ratings of cognitive impairment, clinician ratings had consider­ably greater evidence of validity. In several studies, patients’ ratings of their cognitive performance did not correlate with their neuropsychological test performance, even though they had just been tested on the assessment battery. Ratings by clinicians or other high-contact informants (who were unaware of patients’ test performance) were much more strongly related to patients’ objective test performance, compared with patient self-reports.¹²

The convergence of clinician ratings of cognitive performance with objective test data has been impressive. Correlation coef­ficients of at least r = 0.5, reflecting a moder­ate to large relationships between clinician ratings and objective performance, have been detected. Individual cognitive test domains, such as working memory and processing speed, often do not correlate with each other or with aspects of everyday functioning to that extent.¹³ These data sug­gest that a clinician assessment of cognitive performance, when focused on the correct aspects of cognitive functioning, can be a highly useful proxy for extensive neuropsy­chological testing.

Social cognitive performance. Introspective accuracy for social cognitive judg­ments can be assessed similarly to the strategies used to assess the domains of everyday functioning and cognitive perfor­mance. Patients are asked to complete a typi­cal social cognitive task, such as determining emotions from facial stimuli or examin­ing the eye region of the face, to determine the mental state of the depicted person. Immediately after responding to each stimu­lus, participants rate their confidence in the correctness of that response.

Consistent with the pattern of introspec­tive accuracy for everyday functioning, patients with schizophrenia tend to make more high-confidence errors than healthy individuals on social cognitive tasks. That is, the patients are less likely to realize when they are wrong in their judgments of social stimuli. A similar pattern has been found for mental state attribution,¹⁴ rec­ognition of facial emotion from the self,¹⁵ and recognition of facial emotion from others.¹⁶ These high-confidence errors also are more likely to occur for more difficult stimuli, such as faces that display only mildly emotional expressions. These dif­ficulties appear to be specific to judgments in an immediate evaluation situation. When asked to determine if the behavior of another individual is socially appropri­ate, individuals with schizophrenia are as able as healthy individuals to recognize social mistakes.¹⁷ This work suggests that, at least within the domain of social cogni­tion, introspective accuracy impairment is not caused by generalized poor judgment, just as self-assessments of disability and illness burden are generated at random.

Choosing a reliable informant
If a clinician has not had adequate time or exposure to a patient to make a cogni­tive or functional judgment, what should the strategy be? If asking the patient is uninformative, who should be asked? Our group has gathered information that may help clinicians identify informants who can provide ratings of cognitive perfor­mance and everyday functioning that are convergent with objective evidence.

In a systematic study of validity of reports of various informants, we com­pared correlations between reports of competence of everyday functioning with objective measures of cognitive test per­formance and ability to perform everyday functional skills. Our findings:
   • Patient reports of everyday function­ing were not correlated with performance-based measures for any of 6 rating scales.⁹
   • Clinician reports of everyday func­tioning were correlated with objective per­formance across 4 of 6 rating scales.
   • Correlations between ratings gener­ated by friend or relative informants and other information were almost shocking in their lack of validity (Table 2).⁹

 

We concluded that ratings generated by a generic informant—someone who simply knows the patient and is willing to provide ratings—are highly likely to be uninformative. If a friend or relative provides information of limited useful­ness, the report could easily lead to clinical decisions with high potential for bad out­comes. For example, attempts could fail to transition someone with impaired every­day living skills to independent living, or a patient whose potential is underesti­mated might not be offered opportunities to achieve attainable functional goals.

We found that the closer the rater was to a full caregiver role, the better and more accurate the information obtained. Caregivers who had regular contact with patients had much more valid rat­ings when performance on functionally relevant objective measures was consid­ered. Patients with caregivers had greater impairments in everyday outcomes, how­ever, suggesting that this subset was more impaired than the overall sample. For patients without caregivers, other sources of information—including careful obser­vation by high-contact clinicians—seem to be required to generate a valid assessment of functioning.

Direct functional implications of impaired introspective accuracy
Clinical effects of reduced awareness of ill­ness include reduced adherence to medi­cation, followed by relapse, disturbed behavior, leading to emergency room treat­ments or acute admissions, and—more rarely—disturbed behavior associated with violence or self-harm. Relapses such as these can adversely affect brain structure and function, with declines in cognitive functioning early in the illness.

Our recent study¹⁸ quantifies the direct impact of impairments in introspective accuracy on everyday functioning. We asked 214 individuals with schizophrenia to self-evaluate their cognitive ability with a systematic rating scale and to self-report their everyday functioning in social, voca­tional, and everyday activities domains. We used performance-based measures to assess their cognitive abilities and everyday functional skills. Concurrently, high-contact clinicians rated these same abilities with the same rating scales. We then predicted everyday functioning, as rated by the cli­nicians, with the discrepancies between self-assessed and clinician-assessed func­tioning, and patients’ scores on the perfor­mance-based measures.

Impaired introspective accuracy, as indexed by difference scores between cli­nician ratings and self-reports, was a more potent predictor of everyday functional deficits in social, vocational, and every­day activities domains than scores on performance-based measures of cognitive abilities and functional skills. Even when we analyzed only deficits in introspective accuracy for cognition as the predictor of everyday outcomes in these 3 real-world functional domains, the results were the same. Impaired introspective accuracy was the single best predictor of everyday func­tioning in all 3 domains, with actual abili­ties considerably less important.

Patient characteristics that predict introspective accuracy
Patient characteristics associated with impairments in introspective accuracy (Table 3)^19,20 are easy to identify and assess. Subjective reports of depression have a bell-shaped relationship with introspec­tive accuracy. A self-reported score of 0 by a disabled schizophrenia patient suggests some unawareness of an unfortunate life situation; mild to moderate scores are asso­ciated with more accurate self-assessment; and more severe scores, as seen in other conditions, often predict overestimation of disability.¹⁹

Psychosis and negative symptoms are associated with reduced introspective accu­racy and global over-reporting of functional competence.²⁰ Patients who have never worked have no way to comprehend the specific challenges associated with obtain­ing and sustaining employment. Patients who had a job and have not been able to return work may perceive barriers as more substantial than they are.

Tips to manage impairments in introspective accuracy
Ensure that assessment information is valid. If a patient has limited ability to self-assess, seek other sources of data. If a patient has psychotic symptoms, denies being depressed, or has limited life experi­ence, the clinician should adjust her (his) interpretation of the self-report accord­ingly, because these factors are known to adversely affect the accuracy of self-assessment. Consider informants’ level and quality of contact with the patient, as well as any motivation or bias that might influence the accuracy of their reports. Other professionals, such as occupational therapists, can provide useful information as reference points for treatment planning.

Consider treatments aimed at increasing introspective accuracy, such as structured training and exposure to self-assessment situations,⁶ and interventions aimed at increasing organization and skills perfor­mance. Cognitive remediation therapies, although not widely available, have poten­tial to improve functioning, with excellent persistence over time.²¹

Related Resources
• Harvey PD, ed. Cognitive impairment in schizophrenia: characteristics, assessment and treatment. Cambridge, United Kingdom: Cambridge University Press; 2013.
• Gould F, McGuire LS, Durand D, et al. Self-assessment in schizophrenia: accuracy of assessment of cognition and everyday functioning [published online February 2, 2015]. Neuropsychology.
• Dunning D. Self-insight: detours and roadblocks on the path to knowing thyself. New York, NY: Psychology Press; 2012.

 

Acknowledgment
This paper was supported by Grants MH078775 to Dr. Harvey and MH093432 to Drs. Harvey and Pinkham from the National Institute of Mental Health.

Disclosures
Dr. Harvey has received consulting fees from AbbVie, Boehringer Ingelheim, Forum Pharmaceuticals, Genentech, Otsuka America Pharmaceuticals, Roche, Sanofi, Sunovion Pharmaceuticals, and Takeda Pharmaceuticals. Dr. Pinkham has served as a consultant for Otsuka America Pharmaceuticals.

Lack of insight or “unawareness of illness” occurs within a set of self-assessment problems commonly seen in schizophrenia.¹ In the clinical domain, people who do not realize they are ill typically are unwilling to accept treatment, including medication, with potential for worsened illness. They also may have difficulty self-assess­ing everyday function and functional potential, cognition, social cognition, and attitude, often to a variable degree across these domains (Table 1).^1-3

Self-assessment of performance can be clinically help­ful whether performance is objectively good or bad. Those with poor performance could be helped to attempt to match their aspirations to accomplishments and improve over time. Good performers could have their functioning bolstered by recognizing their competence. Thus, even a population whose performance often is poor could ben­efit from accurate self-assessment or experience additional challenges from inaccurate self-evaluation.

This article discusses patient characteristics associated with impairments in self-assessment and the most accu­rate sources of information for clinicians about patient functioning. Our research shows that an experienced psy­chiatrist is well positioned to make accurate judgments of functional potential and cognitive abilities for people with schizophrenia.

Patterns in patients with impaired self-assessment
Healthy individuals routinely overestimate their abilities and their attractiveness to others.⁴ Feedback that deflates these exag­gerated estimates increases the accuracy of their self-assessments. Mildly depressed individuals typically are the most accurate judges of their true functioning; those with more severe levels of depression tend to underestimate their competence. Thus, sim­ply being an inaccurate self-assessor is not “abnormal.” These response biases are con­sistent and predictable in healthy people.

People with severe mental illness pose a different challenge. As in the following cases, their reports manifest minimal cor­relation with other sources of information, including objective information about performance.

CASE 1
JR, age 28, is referred for occupational therapy because he has never worked since graduat­ing from high school. He tells the therapist his cognitive abilities are average and intact, although his scores on a comprehensive cog­nitive assessment suggest performance at the first percentile of normal distribution or less. His self-reported Beck Depression Inventory (BDI) score is 4. He says he would like to work as a certified public accountant, because he believes he has an aptitude for math. He admits he has no idea what the job entails, but he is quite motivated to set up an interview as soon as possible.

CASE 2
LM, age 48, says his “best job” was managing an auto parts store for 18 months after he earned an associate’s degree and until his second psychotic episode. His most recent work was approximately 12 years ago at an oil-change facility. He agrees to discuss employment but feels his vocational skills are too deteriorated for him to succeed and requests an assessment for Alzheimer’s disease. His cognitive perfor­mance averages in the 10th percentile of the overall population, and his BDI score is 18. Tests of his ability to perform vocational skills sug­gest he is qualified for multiple jobs, including his previous technician position.

Individuals with schizophrenia who report no depression and no work history routinely overestimate their functional potential, whereas those with a history of unsuccessful vocational attempts often underestimate their functional potential. Inaccurate self-assessment can contrib­ute to reduced functioning—in JR’s case because of unrealistic assessment of the match between skills and vocational poten­tial, and in LM’s case because of overly  pessimistic self-evaluation. For people with schizophrenia, inability to self-evaluate can have a bidirectional adverse impact on func­tioning: overestimation may lead to trying tasks that are too challenging, and under­estimation may lead to reduced effort and motivation to take on functional tasks.

Metacognition and introspective accuracy
“Metacognition” refers to self-assessment of the quality and accuracy of performance on cognitive tests.^5-7 Problem-solving tests— such as the Wisconsin Card Sorting test (WCST), in which the person being assessed needs to solve the test through performance feedback—are metacognition tests. When errors are made, the strategy in use needs to be discarded; when responses are correct, the strategy is retained. People with schizo­phrenia have disproportionate difficulties with the WCST, and deficits are especially salient when the test is modified to measure self-assessment of performance and ability to use feedback to change strategies.

“Introspective accuracy” is used to describe the wide-ranging self-assessment impairments in severe mental illness. Theories of metacognition implicate a broad spectrum, of which self-assessment is 1 component, whereas introspective accuracy more specifically indicates judg­ments of accuracy. Because self-assessment is focused on the self, and hence is intro­spective, this conceptualization can be applied to self-evaluations of:
   • achievement in everyday functioning (“Did I complete that task well?”)
   • potential for achievement in everyday functioning (“I could do that job”)
   • cognitive performance (“Yes, I remembered all of those words”)
   • social cognition (“He really is angry”).

 

Domains of impaired introspective accuracy
Everyday functioning. The 3 global domains of everyday functioning are social outcomes, productive/vocational out­comes, and everyday activities, including residential independence/support for peo­ple with severe mental illness. Two areas of inquiry are used in self-assessing everyday functioning: (1) what are you doing now and (2) what could you do in the future? For people with schizophrenia, a related question is how perceived impairments in everyday functioning are associated with subjective illness burden.

People with schizophrenia report illness burden consistent with their self-reported disability, suggesting their reports in these domains are not random.⁸ Studies have consistently found, however, that these patients report:
   • less impairment on average in their everyday functioning than observed by clinicians
   • less subjective illness burden com­pared with individuals with much less severe illnesses.

Their reports also fail to correlate with cli­nicians’ observations.9 Patients with schizo­phrenia who have never been employed may report greater vocational potential than those employed full-time. Interestingly, patients who were previously—but not currently—employed reported the least vocational potential.¹⁰ These data suggest that experience may be a factor: individuals who have never worked have no context for their self-assessments, whereas people who are persistently unemployed may have a perspective on the challenges associated with employment.

In our research,⁹ high-contact clinicians (ie, case manager, psychiatrist, therapist, or residential facility manager) were bet­ter able than family or friends to generate ratings from an assessment questionnaire that correlated with performance-based measures of patients’ ability to perform everyday functional skills. The ratings were generated across multiple functional status scales, suggesting that the rater was more important than the specific scale. We concluded that high-contact clinicians can generate ratings of everyday functioning that are convergent with patients’ abilities, even when they have no information about actual performance scores.

Cognitive performance. When self-reported cognitive abilities are correlated with the results of performance on neuro­psychological assessments, the results are quite consistent. Patients provide reports that do not correlate with their objective performance.¹¹ Interestingly, when clini­cians were asked to use the same strategies as patients to generate ratings of cognitive impairment, clinician ratings had consider­ably greater evidence of validity. In several studies, patients’ ratings of their cognitive performance did not correlate with their neuropsychological test performance, even though they had just been tested on the assessment battery. Ratings by clinicians or other high-contact informants (who were unaware of patients’ test performance) were much more strongly related to patients’ objective test performance, compared with patient self-reports.¹²

The convergence of clinician ratings of cognitive performance with objective test data has been impressive. Correlation coef­ficients of at least r = 0.5, reflecting a moder­ate to large relationships between clinician ratings and objective performance, have been detected. Individual cognitive test domains, such as working memory and processing speed, often do not correlate with each other or with aspects of everyday functioning to that extent.¹³ These data sug­gest that a clinician assessment of cognitive performance, when focused on the correct aspects of cognitive functioning, can be a highly useful proxy for extensive neuropsy­chological testing.

Social cognitive performance. Introspective accuracy for social cognitive judg­ments can be assessed similarly to the strategies used to assess the domains of everyday functioning and cognitive perfor­mance. Patients are asked to complete a typi­cal social cognitive task, such as determining emotions from facial stimuli or examin­ing the eye region of the face, to determine the mental state of the depicted person. Immediately after responding to each stimu­lus, participants rate their confidence in the correctness of that response.

Consistent with the pattern of introspec­tive accuracy for everyday functioning, patients with schizophrenia tend to make more high-confidence errors than healthy individuals on social cognitive tasks. That is, the patients are less likely to realize when they are wrong in their judgments of social stimuli. A similar pattern has been found for mental state attribution,¹⁴ rec­ognition of facial emotion from the self,¹⁵ and recognition of facial emotion from others.¹⁶ These high-confidence errors also are more likely to occur for more difficult stimuli, such as faces that display only mildly emotional expressions. These dif­ficulties appear to be specific to judgments in an immediate evaluation situation. When asked to determine if the behavior of another individual is socially appropri­ate, individuals with schizophrenia are as able as healthy individuals to recognize social mistakes.¹⁷ This work suggests that, at least within the domain of social cogni­tion, introspective accuracy impairment is not caused by generalized poor judgment, just as self-assessments of disability and illness burden are generated at random.

Choosing a reliable informant
If a clinician has not had adequate time or exposure to a patient to make a cogni­tive or functional judgment, what should the strategy be? If asking the patient is uninformative, who should be asked? Our group has gathered information that may help clinicians identify informants who can provide ratings of cognitive perfor­mance and everyday functioning that are convergent with objective evidence.

In a systematic study of validity of reports of various informants, we com­pared correlations between reports of competence of everyday functioning with objective measures of cognitive test per­formance and ability to perform everyday functional skills. Our findings:
   • Patient reports of everyday function­ing were not correlated with performance-based measures for any of 6 rating scales.⁹
   • Clinician reports of everyday func­tioning were correlated with objective per­formance across 4 of 6 rating scales.
   • Correlations between ratings gener­ated by friend or relative informants and other information were almost shocking in their lack of validity (Table 2).⁹

 

We concluded that ratings generated by a generic informant—someone who simply knows the patient and is willing to provide ratings—are highly likely to be uninformative. If a friend or relative provides information of limited useful­ness, the report could easily lead to clinical decisions with high potential for bad out­comes. For example, attempts could fail to transition someone with impaired every­day living skills to independent living, or a patient whose potential is underesti­mated might not be offered opportunities to achieve attainable functional goals.

We found that the closer the rater was to a full caregiver role, the better and more accurate the information obtained. Caregivers who had regular contact with patients had much more valid rat­ings when performance on functionally relevant objective measures was consid­ered. Patients with caregivers had greater impairments in everyday outcomes, how­ever, suggesting that this subset was more impaired than the overall sample. For patients without caregivers, other sources of information—including careful obser­vation by high-contact clinicians—seem to be required to generate a valid assessment of functioning.

Direct functional implications of impaired introspective accuracy
Clinical effects of reduced awareness of ill­ness include reduced adherence to medi­cation, followed by relapse, disturbed behavior, leading to emergency room treat­ments or acute admissions, and—more rarely—disturbed behavior associated with violence or self-harm. Relapses such as these can adversely affect brain structure and function, with declines in cognitive functioning early in the illness.

Our recent study¹⁸ quantifies the direct impact of impairments in introspective accuracy on everyday functioning. We asked 214 individuals with schizophrenia to self-evaluate their cognitive ability with a systematic rating scale and to self-report their everyday functioning in social, voca­tional, and everyday activities domains. We used performance-based measures to assess their cognitive abilities and everyday functional skills. Concurrently, high-contact clinicians rated these same abilities with the same rating scales. We then predicted everyday functioning, as rated by the cli­nicians, with the discrepancies between self-assessed and clinician-assessed func­tioning, and patients’ scores on the perfor­mance-based measures.

Impaired introspective accuracy, as indexed by difference scores between cli­nician ratings and self-reports, was a more potent predictor of everyday functional deficits in social, vocational, and every­day activities domains than scores on performance-based measures of cognitive abilities and functional skills. Even when we analyzed only deficits in introspective accuracy for cognition as the predictor of everyday outcomes in these 3 real-world functional domains, the results were the same. Impaired introspective accuracy was the single best predictor of everyday func­tioning in all 3 domains, with actual abili­ties considerably less important.

Patient characteristics that predict introspective accuracy
Patient characteristics associated with impairments in introspective accuracy (Table 3)^19,20 are easy to identify and assess. Subjective reports of depression have a bell-shaped relationship with introspec­tive accuracy. A self-reported score of 0 by a disabled schizophrenia patient suggests some unawareness of an unfortunate life situation; mild to moderate scores are asso­ciated with more accurate self-assessment; and more severe scores, as seen in other conditions, often predict overestimation of disability.¹⁹

Psychosis and negative symptoms are associated with reduced introspective accu­racy and global over-reporting of functional competence.²⁰ Patients who have never worked have no way to comprehend the specific challenges associated with obtain­ing and sustaining employment. Patients who had a job and have not been able to return work may perceive barriers as more substantial than they are.

Tips to manage impairments in introspective accuracy
Ensure that assessment information is valid. If a patient has limited ability to self-assess, seek other sources of data. If a patient has psychotic symptoms, denies being depressed, or has limited life experi­ence, the clinician should adjust her (his) interpretation of the self-report accord­ingly, because these factors are known to adversely affect the accuracy of self-assessment. Consider informants’ level and quality of contact with the patient, as well as any motivation or bias that might influence the accuracy of their reports. Other professionals, such as occupational therapists, can provide useful information as reference points for treatment planning.

Consider treatments aimed at increasing introspective accuracy, such as structured training and exposure to self-assessment situations,⁶ and interventions aimed at increasing organization and skills perfor­mance. Cognitive remediation therapies, although not widely available, have poten­tial to improve functioning, with excellent persistence over time.²¹

Related Resources
• Harvey PD, ed. Cognitive impairment in schizophrenia: characteristics, assessment and treatment. Cambridge, United Kingdom: Cambridge University Press; 2013.
• Gould F, McGuire LS, Durand D, et al. Self-assessment in schizophrenia: accuracy of assessment of cognition and everyday functioning [published online February 2, 2015]. Neuropsychology.
• Dunning D. Self-insight: detours and roadblocks on the path to knowing thyself. New York, NY: Psychology Press; 2012.

 

Acknowledgment
This paper was supported by Grants MH078775 to Dr. Harvey and MH093432 to Drs. Harvey and Pinkham from the National Institute of Mental Health.

Disclosures
Dr. Harvey has received consulting fees from AbbVie, Boehringer Ingelheim, Forum Pharmaceuticals, Genentech, Otsuka America Pharmaceuticals, Roche, Sanofi, Sunovion Pharmaceuticals, and Takeda Pharmaceuticals. Dr. Pinkham has served as a consultant for Otsuka America Pharmaceuticals.

Lack of insight or “unawareness of illness” occurs within a set of self-assessment problems commonly seen in schizophrenia.¹ In the clinical domain, people who do not realize they are ill typically are unwilling to accept treatment, including medication, with potential for worsened illness. They also may have difficulty self-assess­ing everyday function and functional potential, cognition, social cognition, and attitude, often to a variable degree across these domains (Table 1).^1-3

Self-assessment of performance can be clinically help­ful whether performance is objectively good or bad. Those with poor performance could be helped to attempt to match their aspirations to accomplishments and improve over time. Good performers could have their functioning bolstered by recognizing their competence. Thus, even a population whose performance often is poor could ben­efit from accurate self-assessment or experience additional challenges from inaccurate self-evaluation.

This article discusses patient characteristics associated with impairments in self-assessment and the most accu­rate sources of information for clinicians about patient functioning. Our research shows that an experienced psy­chiatrist is well positioned to make accurate judgments of functional potential and cognitive abilities for people with schizophrenia.

Patterns in patients with impaired self-assessment
Healthy individuals routinely overestimate their abilities and their attractiveness to others.⁴ Feedback that deflates these exag­gerated estimates increases the accuracy of their self-assessments. Mildly depressed individuals typically are the most accurate judges of their true functioning; those with more severe levels of depression tend to underestimate their competence. Thus, sim­ply being an inaccurate self-assessor is not “abnormal.” These response biases are con­sistent and predictable in healthy people.

People with severe mental illness pose a different challenge. As in the following cases, their reports manifest minimal cor­relation with other sources of information, including objective information about performance.

CASE 1
JR, age 28, is referred for occupational therapy because he has never worked since graduat­ing from high school. He tells the therapist his cognitive abilities are average and intact, although his scores on a comprehensive cog­nitive assessment suggest performance at the first percentile of normal distribution or less. His self-reported Beck Depression Inventory (BDI) score is 4. He says he would like to work as a certified public accountant, because he believes he has an aptitude for math. He admits he has no idea what the job entails, but he is quite motivated to set up an interview as soon as possible.

CASE 2
LM, age 48, says his “best job” was managing an auto parts store for 18 months after he earned an associate’s degree and until his second psychotic episode. His most recent work was approximately 12 years ago at an oil-change facility. He agrees to discuss employment but feels his vocational skills are too deteriorated for him to succeed and requests an assessment for Alzheimer’s disease. His cognitive perfor­mance averages in the 10th percentile of the overall population, and his BDI score is 18. Tests of his ability to perform vocational skills sug­gest he is qualified for multiple jobs, including his previous technician position.

Individuals with schizophrenia who report no depression and no work history routinely overestimate their functional potential, whereas those with a history of unsuccessful vocational attempts often underestimate their functional potential. Inaccurate self-assessment can contrib­ute to reduced functioning—in JR’s case because of unrealistic assessment of the match between skills and vocational poten­tial, and in LM’s case because of overly  pessimistic self-evaluation. For people with schizophrenia, inability to self-evaluate can have a bidirectional adverse impact on func­tioning: overestimation may lead to trying tasks that are too challenging, and under­estimation may lead to reduced effort and motivation to take on functional tasks.

Metacognition and introspective accuracy
“Metacognition” refers to self-assessment of the quality and accuracy of performance on cognitive tests.^5-7 Problem-solving tests— such as the Wisconsin Card Sorting test (WCST), in which the person being assessed needs to solve the test through performance feedback—are metacognition tests. When errors are made, the strategy in use needs to be discarded; when responses are correct, the strategy is retained. People with schizo­phrenia have disproportionate difficulties with the WCST, and deficits are especially salient when the test is modified to measure self-assessment of performance and ability to use feedback to change strategies.

“Introspective accuracy” is used to describe the wide-ranging self-assessment impairments in severe mental illness. Theories of metacognition implicate a broad spectrum, of which self-assessment is 1 component, whereas introspective accuracy more specifically indicates judg­ments of accuracy. Because self-assessment is focused on the self, and hence is intro­spective, this conceptualization can be applied to self-evaluations of:
   • achievement in everyday functioning (“Did I complete that task well?”)
   • potential for achievement in everyday functioning (“I could do that job”)
   • cognitive performance (“Yes, I remembered all of those words”)
   • social cognition (“He really is angry”).

 

Domains of impaired introspective accuracy
Everyday functioning. The 3 global domains of everyday functioning are social outcomes, productive/vocational out­comes, and everyday activities, including residential independence/support for peo­ple with severe mental illness. Two areas of inquiry are used in self-assessing everyday functioning: (1) what are you doing now and (2) what could you do in the future? For people with schizophrenia, a related question is how perceived impairments in everyday functioning are associated with subjective illness burden.

People with schizophrenia report illness burden consistent with their self-reported disability, suggesting their reports in these domains are not random.⁸ Studies have consistently found, however, that these patients report:
   • less impairment on average in their everyday functioning than observed by clinicians
   • less subjective illness burden com­pared with individuals with much less severe illnesses.

Their reports also fail to correlate with cli­nicians’ observations.9 Patients with schizo­phrenia who have never been employed may report greater vocational potential than those employed full-time. Interestingly, patients who were previously—but not currently—employed reported the least vocational potential.¹⁰ These data suggest that experience may be a factor: individuals who have never worked have no context for their self-assessments, whereas people who are persistently unemployed may have a perspective on the challenges associated with employment.

In our research,⁹ high-contact clinicians (ie, case manager, psychiatrist, therapist, or residential facility manager) were bet­ter able than family or friends to generate ratings from an assessment questionnaire that correlated with performance-based measures of patients’ ability to perform everyday functional skills. The ratings were generated across multiple functional status scales, suggesting that the rater was more important than the specific scale. We concluded that high-contact clinicians can generate ratings of everyday functioning that are convergent with patients’ abilities, even when they have no information about actual performance scores.

Cognitive performance. When self-reported cognitive abilities are correlated with the results of performance on neuro­psychological assessments, the results are quite consistent. Patients provide reports that do not correlate with their objective performance.¹¹ Interestingly, when clini­cians were asked to use the same strategies as patients to generate ratings of cognitive impairment, clinician ratings had consider­ably greater evidence of validity. In several studies, patients’ ratings of their cognitive performance did not correlate with their neuropsychological test performance, even though they had just been tested on the assessment battery. Ratings by clinicians or other high-contact informants (who were unaware of patients’ test performance) were much more strongly related to patients’ objective test performance, compared with patient self-reports.¹²

The convergence of clinician ratings of cognitive performance with objective test data has been impressive. Correlation coef­ficients of at least r = 0.5, reflecting a moder­ate to large relationships between clinician ratings and objective performance, have been detected. Individual cognitive test domains, such as working memory and processing speed, often do not correlate with each other or with aspects of everyday functioning to that extent.¹³ These data sug­gest that a clinician assessment of cognitive performance, when focused on the correct aspects of cognitive functioning, can be a highly useful proxy for extensive neuropsy­chological testing.

Social cognitive performance. Introspective accuracy for social cognitive judg­ments can be assessed similarly to the strategies used to assess the domains of everyday functioning and cognitive perfor­mance. Patients are asked to complete a typi­cal social cognitive task, such as determining emotions from facial stimuli or examin­ing the eye region of the face, to determine the mental state of the depicted person. Immediately after responding to each stimu­lus, participants rate their confidence in the correctness of that response.

Consistent with the pattern of introspec­tive accuracy for everyday functioning, patients with schizophrenia tend to make more high-confidence errors than healthy individuals on social cognitive tasks. That is, the patients are less likely to realize when they are wrong in their judgments of social stimuli. A similar pattern has been found for mental state attribution,¹⁴ rec­ognition of facial emotion from the self,¹⁵ and recognition of facial emotion from others.¹⁶ These high-confidence errors also are more likely to occur for more difficult stimuli, such as faces that display only mildly emotional expressions. These dif­ficulties appear to be specific to judgments in an immediate evaluation situation. When asked to determine if the behavior of another individual is socially appropri­ate, individuals with schizophrenia are as able as healthy individuals to recognize social mistakes.¹⁷ This work suggests that, at least within the domain of social cogni­tion, introspective accuracy impairment is not caused by generalized poor judgment, just as self-assessments of disability and illness burden are generated at random.

Choosing a reliable informant
If a clinician has not had adequate time or exposure to a patient to make a cogni­tive or functional judgment, what should the strategy be? If asking the patient is uninformative, who should be asked? Our group has gathered information that may help clinicians identify informants who can provide ratings of cognitive perfor­mance and everyday functioning that are convergent with objective evidence.

In a systematic study of validity of reports of various informants, we com­pared correlations between reports of competence of everyday functioning with objective measures of cognitive test per­formance and ability to perform everyday functional skills. Our findings:
   • Patient reports of everyday function­ing were not correlated with performance-based measures for any of 6 rating scales.⁹
   • Clinician reports of everyday func­tioning were correlated with objective per­formance across 4 of 6 rating scales.
   • Correlations between ratings gener­ated by friend or relative informants and other information were almost shocking in their lack of validity (Table 2).⁹

 

We concluded that ratings generated by a generic informant—someone who simply knows the patient and is willing to provide ratings—are highly likely to be uninformative. If a friend or relative provides information of limited useful­ness, the report could easily lead to clinical decisions with high potential for bad out­comes. For example, attempts could fail to transition someone with impaired every­day living skills to independent living, or a patient whose potential is underesti­mated might not be offered opportunities to achieve attainable functional goals.

We found that the closer the rater was to a full caregiver role, the better and more accurate the information obtained. Caregivers who had regular contact with patients had much more valid rat­ings when performance on functionally relevant objective measures was consid­ered. Patients with caregivers had greater impairments in everyday outcomes, how­ever, suggesting that this subset was more impaired than the overall sample. For patients without caregivers, other sources of information—including careful obser­vation by high-contact clinicians—seem to be required to generate a valid assessment of functioning.

Direct functional implications of impaired introspective accuracy
Clinical effects of reduced awareness of ill­ness include reduced adherence to medi­cation, followed by relapse, disturbed behavior, leading to emergency room treat­ments or acute admissions, and—more rarely—disturbed behavior associated with violence or self-harm. Relapses such as these can adversely affect brain structure and function, with declines in cognitive functioning early in the illness.

Our recent study¹⁸ quantifies the direct impact of impairments in introspective accuracy on everyday functioning. We asked 214 individuals with schizophrenia to self-evaluate their cognitive ability with a systematic rating scale and to self-report their everyday functioning in social, voca­tional, and everyday activities domains. We used performance-based measures to assess their cognitive abilities and everyday functional skills. Concurrently, high-contact clinicians rated these same abilities with the same rating scales. We then predicted everyday functioning, as rated by the cli­nicians, with the discrepancies between self-assessed and clinician-assessed func­tioning, and patients’ scores on the perfor­mance-based measures.

Impaired introspective accuracy, as indexed by difference scores between cli­nician ratings and self-reports, was a more potent predictor of everyday functional deficits in social, vocational, and every­day activities domains than scores on performance-based measures of cognitive abilities and functional skills. Even when we analyzed only deficits in introspective accuracy for cognition as the predictor of everyday outcomes in these 3 real-world functional domains, the results were the same. Impaired introspective accuracy was the single best predictor of everyday func­tioning in all 3 domains, with actual abili­ties considerably less important.

Patient characteristics that predict introspective accuracy
Patient characteristics associated with impairments in introspective accuracy (Table 3)^19,20 are easy to identify and assess. Subjective reports of depression have a bell-shaped relationship with introspec­tive accuracy. A self-reported score of 0 by a disabled schizophrenia patient suggests some unawareness of an unfortunate life situation; mild to moderate scores are asso­ciated with more accurate self-assessment; and more severe scores, as seen in other conditions, often predict overestimation of disability.¹⁹

Psychosis and negative symptoms are associated with reduced introspective accu­racy and global over-reporting of functional competence.²⁰ Patients who have never worked have no way to comprehend the specific challenges associated with obtain­ing and sustaining employment. Patients who had a job and have not been able to return work may perceive barriers as more substantial than they are.

Tips to manage impairments in introspective accuracy
Ensure that assessment information is valid. If a patient has limited ability to self-assess, seek other sources of data. If a patient has psychotic symptoms, denies being depressed, or has limited life experi­ence, the clinician should adjust her (his) interpretation of the self-report accord­ingly, because these factors are known to adversely affect the accuracy of self-assessment. Consider informants’ level and quality of contact with the patient, as well as any motivation or bias that might influence the accuracy of their reports. Other professionals, such as occupational therapists, can provide useful information as reference points for treatment planning.

Consider treatments aimed at increasing introspective accuracy, such as structured training and exposure to self-assessment situations,⁶ and interventions aimed at increasing organization and skills perfor­mance. Cognitive remediation therapies, although not widely available, have poten­tial to improve functioning, with excellent persistence over time.²¹

Related Resources
• Harvey PD, ed. Cognitive impairment in schizophrenia: characteristics, assessment and treatment. Cambridge, United Kingdom: Cambridge University Press; 2013.
• Gould F, McGuire LS, Durand D, et al. Self-assessment in schizophrenia: accuracy of assessment of cognition and everyday functioning [published online February 2, 2015]. Neuropsychology.
• Dunning D. Self-insight: detours and roadblocks on the path to knowing thyself. New York, NY: Psychology Press; 2012.

 

Acknowledgment
This paper was supported by Grants MH078775 to Dr. Harvey and MH093432 to Drs. Harvey and Pinkham from the National Institute of Mental Health.

Disclosures
Dr. Harvey has received consulting fees from AbbVie, Boehringer Ingelheim, Forum Pharmaceuticals, Genentech, Otsuka America Pharmaceuticals, Roche, Sanofi, Sunovion Pharmaceuticals, and Takeda Pharmaceuticals. Dr. Pinkham has served as a consultant for Otsuka America Pharmaceuticals.