Introduction
Between 50 and 80% of patients with schizophrenia are reported to have impaired insight into their illness [
1‐
3]. In the context of a life-altering disease such as schizophrenia, retaining insight helps patients make sense of the meaning of life events. In contrast, poor insight could be understood as a failure to construct a coherent account of complex and potentially traumatic life experiences that can result in loss of functioning and hospitalization [
4]. Insight is a multidimensional concept that includes awareness of illness, the capacity to re-label psychotic experiences as abnormal, and adherence to treatment, which vary along a continuum [
5,
6]. Poor insight is a feature of schizophrenia across different cultures and across all stages of the illness, and it persists even after symptoms have remitted [
2,
4]. From a treating clinician’s perspective, impaired insight is one of the most vexing aspects of the illness because of the challenges it poses for therapeutic engagement and medication adherence [
7,
8].
The relationship between insight and health is complex. Poor clinical insight in schizophrenia has been associated with poorer medication adherence, which can lead to an increase in positive symptoms and relapse risk [
2,
4,
9‐
12]. Conversely, poor insight also may exert a protective effect. In patients with schizophrenia or other psychotic illnesses, small but positive associations have been reported between insight and depression or suicidal thoughts or actions [
9,
13‐
15]. However, more complex relationships between insight and suicidal thoughts or actions, mediated by other symptoms and changing over the course of illness, have been observed in several analyses [
16,
17]. This dual nature of insight in schizophrenia outcomes has been called the insight paradox [
4,
14,
18,
19]. Because symptoms of depression affect health-related quality of life (QoL), there may be an interaction between poor insight and better health-related QoL arising from a lower level of depressive symptoms. Negative impacts of depression on health-related QoL in patients with good insight may prevent these individuals from attaining personal goals and increase the risk of suicide [
4,
19]. However, such determinations may be limited if insight is viewed by category (i.e., good versus poor insight) rather than as a continuum or assessed using only population means, which is typically how such data have been reported [
20‐
22].
Neurocognition may also play a role in potential interactions between insight and mental health status, depression, and health-related QoL, although the specific connections between these domains and global insight are unclear. For example, one study in patients with psychotic disorders found that improvement in insight over 3 years was significantly associated both with fewer clinical symptoms and with better neurocognitive performance at baseline. Gradual improvement in insight over 3 years was associated with symptom improvement but not with improved neurocognition [
23]. Additionally, a meta-analysis identified positive correlations between neurocognition and objective health-related QoL (i.e., observable life conditions) but negative or no association between neurocognition and subjective health-related QoL (e.g., patient-reported satisfaction with life conditions) [
24].
To date, one of the major limitations of insight research is that insight generally has been conceptualized and reported as a categorical variable, whereas it seems more appropriate to consider insight as a dimensional phenomenon, lying along a continuum with gradations in the severity of lack of awareness [
4]. As a result, it remains unknown whether there is a particular level of impaired insight (e.g., mild versus moderate versus severe) at which a lack of agreement with others’ appraisal of the patient’s well-being, neurocognitive abilities, and depression becomes apparent. The current post hoc analysis used data from the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) [
25] study to explore how insight, assessed both as a categorical variable (more versus less impairment) and based on gradations along a continuum from none to extreme, is related to patient-reported mental health status, depression, and neurocognition in patients with schizophrenia.
Discussion
This post hoc analysis of CATIE study data provides a greater understanding of how the degree of insight may influence responses of individuals with schizophrenia on patient-reported outcomes, including health-related QoL measures. At baseline, statistically significant associations were observed between CATIE participants’ levels of insight, based on PANSS Item G12 rating, and most outcomes assessed, including patient-reported health-related QoL outcomes, illness severity, PANSS component scores, and depression scale scores. Among CATIE participants, insight that was moderately severe or worse (i.e., PANSS Item G12 rating ≥ 5) was associated with better mental/emotional health status and better health-related QoL (based on SF-12 MCS and PCS scores) compared with those with less impairment (PANSS Item G12 rating < 5). Patients with more impaired insight (i.e., PANSS Item G12 ≥ 5) accounted for 10% of CATIE participants and had lower levels of baseline depression compared with patients with less impaired insight (i.e., PANSS Item G12 rating < 5). Also, among the group with poor insight, there was a discrepancy in appraisal of schizophrenia severity, with lower severity reported by patients than by physicians. These results are consistent with the finding that, compared with the overall CATIE population, patients with substantially greater insight impairment were less likely to adhere to their medication [
39]. For the 90% of CATIE patients with baseline PANSS Item G12 rating < 5, physician- and patient-reported severity ratings were generally more consistently close in their agreement with one another.
The validity of patient-reported outcomes to predict health outcomes in schizophrenia has been questioned because of possible confounding due to impaired insight, such that patients with poor insight might lack the ability to appraise their own health-related QoL accurately [
40,
41]. Some studies have shown that poor insight has a weaker correlation with self- and reviewer-rated health-related QoL measurements than good insight [
42,
43]. However, other studies have found that self-reported health-related QoL measures in patients with schizophrenia are reliable [
44], although this appears to be most reproducible among those with better insight (and potentially less severe symptoms) [
45]. The current results indicate that although severe insight impairment can be associated with significant divergence between physician- and patient-reported ratings of illness, the proportion of patients in which this occurs is likely small. For a large majority of patients with schizophrenia—90% in the current analysis—physician and patient ratings aligned. These findings have important implications for the use of patient-reported outcomes in the clinical trials enrolling patients with schizophrenia. Because successful treatment of schizophrenia requires functional recovery in addition to remission of symptoms [
46], assessments of health-related QoL and daily functioning are essential for evaluating treatment efficacy in patients with schizophrenia [
47]. Impairment in insight would not be expected to have a substantial effect on measurement of those outcomes at a study population level, as patient-reported outcomes diverge from physician-reported outcomes only with moderate-severe to extreme impairment in insight. However, on the individual patient level, patients with very poor insight may report better mental health status relative to those with fair or good insight, and this should be taken into consideration when interpreting such data.
In the current post hoc analysis, self-reported mental health (SF-12 MCS) status was found to be linearly related to symptom severity (PANSS total score) when PANSS was <90, but no relation was found in patients with greater symptom severity (≥90). Linear regression modeling demonstrated a negative association between self-reported mental health and PANSS total scores, as expected given that higher SF-12 MCS scores indicate better health-related QoL, whereas higher PANSS scores indicate greater symptom severity. For both baseline and longitudinal data, an approximate 2-point change in SF-12 MCS was associated with a 10-point change in PANSS total score. No relationship was observed between self-reported physical health (SF-12 PCS score) and PANSS total score.
In this analysis, we found that depression scores (based on Calgary total score) decreased with increasing impairment in insight. The change across the insight continuum was small but consistent with the findings from a recent meta-analysis, in which a significant association between global clinical insight and depression was observed, with better insight associated with higher levels of depressive symptoms [
48]. While the current post hoc analysis (in line with other studies [
3,
13,
31]) reports better health-related QoL among patients with schizophrenia with impaired insight relative to those with fair or good insight, other studies have found that patients with impaired insight have decreased health-related QoL [
42,
49] or have found no significant association between the two [
28,
50‐
52]. Several factors may explain the discrepancies between these findings, including differences in assessments used to measure insight [
28,
31,
52]. In addition, characteristics of the study population, such as treatment history, symptom stability, status of social and living situations, and objective QoL [
42,
49] may also be operant. The single measure in this analysis that did not vary significantly with insight was the Neurocognitive Composite Score; however, an association was observed between insight and the PANSS cognitive component score. Significant associations between measures of insight and cognition have been reported in meta-analyses of patients with schizophrenia (11 studies) and in patients with schizophrenia or psychosis (35 studies) [
53]. However, the authors observed small and inconsistent correlations in a number of included studies and posited that the relationship between insight and cognitive deficits is nonlinear [
53]. Results of published studies suggest that the relation between insight and cognition may vary with age, severity of disease, or number of previous episodes [
54,
55].
Treatment of poor insight in schizophrenia has been approached using both pharmacological and psychological therapies. Clozapine and second-generation antipsychotics have been associated with improvements in insight in schizophrenia [
20,
56]. However, the association was not observed in an analysis that controlled for other clinical factors, indicating that effects on impaired insight were mediated by overall improvement in symptoms [
20]. Several psychological therapies have been assessed for effect on insight in schizophrenia, including cognitive behavioral therapy, psycho-education, adherence therapy, social skills training, and metacognitive training [
8,
57]. Overall, meta-analyses have shown small to moderate effect sizes for these interventions on insight, although the approaches vary in effectiveness [
8,
57]. Cognitive behavioral therapy for psychosis, which is specifically adapted for individuals with psychosis, has been associated with improvements in insight in several studies [
58‐
61] and is recommended in the American Psychiatric Association practice guideline for the treatment of patients with schizophrenia [
62]. Metacognitive training for psychotic illnesses and metacognitive reflection and insight therapy (MERIT) are emerging therapies that address insight in schizophrenia and aim to help patients strengthen their understanding of their distorted mental processes (through metacognitive training) or their self-appraisal abilities (through MERIT) [
63‐
65]. One study assessing MERIT found that patients with first-episode psychosis and poor clinical insight who received 6 months of MERIT had statistically significant improvements in objective measures of insight without any increases in hopelessness or emotional distress relative to those who had standard meetings with therapists [
66]. A systematic review in patients with schizophrenia spectrum disorders concluded that metacognitive training improved cognitive insight, illness awareness, and awareness of delusions and hallucinations, while MERIT was found to be less effective [
67]. These types of therapy hold promise for patients with poor insight.
Limitations of the current study include the post hoc nature of the analysis and the use of a single-item measurement for insight, PANSS Item G12. In addition, while the goal of the current analysis was to better understand how impairment in insight might affect responses on patient-reported measures of symptom severity and mental health, observed associations between the outcomes assessed do not necessarily indicate a causal relationship. Finally, the results observed here may not generalize to patients with schizophrenia who differ from those who enrolled in the CATIE study. CATIE study participants were predominantly male and white and notably were willing to enroll in a clinical trial for the treatment of schizophrenia; such patients may differ in their level of insight from those who are unwilling to enroll in a clinical study. Our analysis is strengthened by use of a nonparametric regression method (LOESS) to examine the relationship between insight and subjective health-related QoL. By characterizing relatively large groups for each of the levels of insight, this approach revealed a spectrum of insight levels based on PANSS Item G12 ratings.
In conclusion, this post hoc analysis assessed the relationship between degrees of impairment in insight and physician- and patient-reported outcomes. Results indicate that the inverse relationships between insight and self-perceived mental health and mood are most robust once moderate-severe levels of impairment of insight are reached. In this analysis, agreement about health status, depression, and neurocognition differed between patients with different degrees of insight impairment, but most notably between the 10% of patients with moderate-severe to extreme lack of insight and the 90% with lower levels of impairment. For most patients with schizophrenia enrolled in the CATIE study, there was little effect of insight impairment on the convergence of ratings for patient- versus physician-reported outcomes. By providing a greater understanding of the relationship between insight impairment and patient-reported outcomes in schizophrenia, the results from this analysis may be informative for clinicians seeking to address impairment in insight to help patients achieve their therapeutic goals.
Acknowledgments
Medical writing and editorial support for the preparation of this article was provided by Susan Bartko-Winters, PhD, and Esther Tazartes, MS, of Global Outcomes Group, and Gina Daniel, PhD and Kathleen Dorries, PhD of Peloton Advantage, LLC (Parsippany, NJ), an OPEN Health company, and funded by Alkermes, Inc.
The data used in the preparation of this manuscript were obtained from the limited-access datasets distributed from the NIH-supported Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) study (NCT00014001). This was a multisite, clinical trial of persons with schizophrenia comparing the effectiveness of randomly assigned medication treatment. The study was supported by National Institute of Mental Health (NIMH) Contract #N01MH90001 to the University of North Carolina at Chapel Hill. The CATIE study was conducted by investigators from the University of North Carolina, Duke University, Columbia University, the University of Southern California, the University of Rochester, and Yale University along with program staff of the Division of Interventions and Services Research, NIMH, and investigators from each of the 57 research sites in the United States. This manuscript reflects the views of the authors and may not reflect the opinions or views of the CATIE Study Investigators or the NIMH.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.