Today, lack of insight into illness is considered to be a core symptom of schizophrenia and is most commonly defined as a multidimensional construct including awareness of having a mental disorder, awareness of the need of treatment, understanding the social consequences of the disorder, awareness of specific signs and symptoms of the disorder, and attribution of symptoms to the disorder [Reference Amador and David2].
Two different theories are discussed in the literature concerning the onset of insight. The first is a neuropsychological theory hypothesising that insight is related to an underlying cerebral pathology [Reference Thompson, McGorry and Harrigan56]. It is based on findings of associations between impairments in the neurocognitive performance of schizophrenia patients as well as on structural brain abnormalities associated with reduced insight in neurological disorders. Lately, imaging studies in schizophrenia have tried to identify neuronal correlates of insight into illness and found prefrontal, inferior and orbitofrontal regions to be in relation with lack of insight [Reference Smith, Hull, Israel and Willson52]. Alternatively, the second theory in terms of insight comes from a psychological point of view suggesting that a lack of insight may be a psychological defence [Reference Startup53]. Therefore, schizophrenia patients deny being mentally ill in order to protect themselves from stigma [Reference Tait, Birchwood and Trower55].
It is believed that between 50–80% of patients suffering from schizophrenia do not believe that they have a mental disorder which makes insight into illness a wide spread phenomenon [Reference Keshavan, Rabinowitz, DeSmedt, Harvey and Schooler25]. Given the fact that several authors reported a lack of insight in schizophrenia to lead to poor treatment outcome, impairments of insight in schizophrenia patients turn to be a major challenge especially in every day clinical practice. This is why identifying influencing factors and predictors of insight into illness has been of great scientific interest and importance. For the identification of such variables could result in the development of effective and efficient treatment strategies to improve insight and with it the outcome of schizophrenia patients [Reference Schwartz47].
Several factors have emerged to be associated with insight such as the neurocognitive performance [Reference Simon, Berger, Giacomini, Ferrero and Mohr51], the patients’ psychopathology [Reference Sevy, Nathanson, Visweswaraiah and Amador48] or the level of functioning [Reference Francis and Penn14]. Impairments in insight are believed to be mediated by deficiencies in conceptual disorganization and flexibility in abstract thinking [Reference Shad, Tamminga, Cullum, Haas and Keshavan49] and Aleman et al. reported in their meta-analysis that impairments of set-shifting and error monitoring also contribute to poor insight [Reference Aleman, Agrawal, Morgan and David1]. Greater insight on the other hand was found to be linked to adherence and satisfaction with antipsychotic treatment indicating that improving insight is one of the major goals in the treatment of schizophrenia [Reference Lysaker, Buck, Salvatore, Popolo and Dimaggio29].
However, today literature on insight in schizophrenia yields inconsistent results due to differing and small patient samples and individual assessment tools [Reference Lincoln, Lullmann and Rief26]. Besides, lots of studies have been either performed in symptomatic or stable patients reflecting state measures rather than trait factors [Reference Varga, Magnusson, Flekkoy, David and Opjordsmoen57]. Also, most often studies on insight in schizophrenia group patients into those with “good” and those with “bad” insight which seems to be short-handed given the complex construct of insight [Reference Saeedi, Addington and Addington43].
Therefore, in order to overcome limitations of past research we chose to descriptively examine insight into illness in schizophrenia patients with acute symptoms and to evaluate the change of insight during treatment. A non-dichotomic statistical approach was chosen to evaluate insight into illness in this patient sample. Aims of this analysis were to:
(I) examine insight into illness at admission and during the course of the study;
(II) identify influencing factors of insight at every assessment time-point;
(III) analyse baseline variables predicting insight at discharge.
Data were collected in a multicenter follow-up programme (German Research Network on Schizophrenia) [Reference Wolwer, Buchkremer, Hafner, Klosterkotter, Maier and Moller59] at eleven psychiatric university hospitals and three psychiatric district hospitals. All patients admitted to one of the above mentioned hospitals between January 2001 and December 2004 with the diagnosis of schizophrenia, schizophreniform disorder, delusional disorder and schizoaffective disorder according to DSM-IV criteria were selected for inclusion using a random generator. This was done to limit the number of patients included concurrently assuring a representative patient sample. Subjects were aged between 18 and 65 years. Exclusion criteria were a head injury, a history of major medical illness and alcohol or drug dependency. An informed written consent had to be provided to participate in the study. The study protocol was approved by the local ethics committees [Reference Jager, Riedel, Messer, Laux, Pfeiffer and Naber20].
DSM-IV diagnoses were established by clinical researchers on the basis of the German version of the Structured Clinical Interview for DSM-IV . Sociodemographic, clinical and course-related variables such as age at onset, age at first hospitalization, suicidality or episodes of illness were collected using a standardized documentation system (BADO) [Reference Cording11] during interviews with patients, relatives and care providers.
Insight into illness was evaluated using the item G12 “lack of insight and judgement” of the general psychopathology subscale of the Positive and Negative Syndrome Scale [Reference Kay, Opler and Lindenmayer23]. To assess symptom severity the Positive and Negative Syndrome Scale for Schizophrenia (PANSS) [Reference Kay, Opler and Lindenmayer23] was used. Depressive symptoms were evaluated via the Hamilton Depression Rating Scale [Reference Hamilton19]. Adverse events were assessed using the Udvalg for Klinske Undersogelser (UKU) Side Effect Rating Scale [Reference Lingjaerde, Ahlfors, Bech, Dencker and Elgen27]. The Global Assessment of Functioning Scale (GAF)  and the Social and Occupational Functioning Scale (SOFAS)  were applied to evaluate the level of functioning. Subjective well-being was evaluated using the Subjective Well-being Under Neuroleptic Treatment Scale, short version (SWN-K) [Reference Naber, Moritz, Lambert, Pajonk, Holzbach and Mass35]. To assess the patient's attitude towards adherence the Compliance Rating Scale (CRS) was used [Reference Kemp and David24] and dichotomised according to Schennach-Wolff et al. [Reference Schennach-Wolff, Jager, Seemuller, Obermeier, Messer and Laux46]. To evaluate the patient's premorbid adjustment the short-scale for premorbid social-personal adjustment of the Phillips Scale was applied [Reference Phillips37]. All raters had been trained using the applied scales. A high inter-rater reliability was achieved (ANOVA-ICC>0.8).
2.3. Statistical analysis
The primary criterion of this analysis, insight into illness, was assessed using the PANSS-G12 item. To overcome limitations of the literature, this item was kept in its original form as an ordinal variable. In a first step, univariate tests were applied to investigate the association of insight into illness with different sociodemographic, illness-related and clinical variables using the Wilcoxon rank-sum test, the Kruskal-Wallis test or Spearman's correlation as appropriate. In order to examine the change of insight into illness during the course of the study, patients were grouped as follows: patients with stable and good insight, patients with an improvement of insight, patients with no change of insight, and patients with a worsening of insight. Here, Fisher's exact test, the Kruskal-Wallis test, and the ANOVA F-test were applied to analyse the association between different sociodemographic, illness-related and clinical variables and the four different subgroups concerning the change of insight.
In a second step, a multivariate regression approach identified predictors of insight at discharge while adjusting for the baseline level of insight at admission. Specifically, proportional odds logistic regression [Reference Phillips37] was applied, which uses the full information on the patients’ level of insight. Starting with a model incorporating main effects for all possible predictors, backwards selection based on Akaike's Information Criterion (AIC) was performed to find the best model trading off goodness-of-fit and complexity. The proportional odds assumption of the final model was verified by estimating more complex models with nominal effects of the predictors. As a measure of goodness-of-fit compared to the null model, Nagelkerke's R2 is reported. To visualize the mean course of insight the last observation carried forward (LOCF) method was applied. Throughout the present analyses, the general psychopathology subscore of the PANSS excludes the “lack of insight” item G12. All analyses were performed using the statistical software environment R 2.11.1 .
3.1. Patients, psychopathology and treatment
In the entire multicenter study 474 patients were enrolled. 46 patients had to drop out for different reasons (e.g. retrospective violation of inclusion criteria, withdrawal of informed consent). Another 29 patients were excluded from analysis: 28 patients because they were discharged from the hospital within 7 days after admission and 1 patient due to incomplete PANSS ratings. Therefore, the sample available for analysis comprised 399 subjects, 174 (44%) female. The mean age was 35.4 (±11.1) and the median duration of illness was 3 [Reference Aleman, Agrawal, Morgan and David1] years. The mean number of previous hospitalizations was 2.95 (±5.41) and the mean duration of the current hospitalization was 64.7 (±46.5) days. 31% of the patients suffered from their first illness episode. 50% of patients followed a fulltime activity and 16% lived with a partner. 48% of the patients were thought to be adherent to treatment. At admission, 20% of the patients suffered from suicidality. 207 patients were available for follow-up. For a drop-out analysis and further sample description please see Schennach et al. .
The patients’ psychopathological, functioning and subjective well-being data at admission are displayed in Table 1. Also, DSM-IV diagnoses are listed there. Patients were treated under naturalistic conditions as follows: 81% of the patients received first-generation antipsychotic, 80% of patients second-generation antipsychotic treatment and 64% of the patients were treated with first - as well as second-generation antipsychotics. Tranquilizers were administered in 79% of patients and mood stabilizers in 16%. 27% of the patients were also treated with antidepressants.
3.2. Insight into illness at admission and the change of insight during treatment
The mean PANSS rating at admission on the “lack of insight”-item was 3.1 improving significantly (p<0.0001) to a score of 2.2 at discharge (Fig. 1). At admission, 32% of the patients were found to have good insight into illness, 8% had a satisfying insight, 15% of the patients suffered from a mild lack of insight, 22% showed a moderate level of insight and 13% had poor insight into their illness. 9% of the patients had almost no insight. Two patients (0.5%) were rated to have an extreme lack of insight.
For the course of the patients’ insight see Fig. 2. 76% of the patients with good insight at admission remained stable regarding their insight into illness and were still found to have good insight at discharge (Fig. 3). For the change of the different insight ratings see Fig. 4. We were able to follow 207 patients up one year after their discharge from hospital and found no significant change of the level of insight between discharge and follow-up (p = 0.20) underlining the importance of the improvement of insight during the index episode.
3.3. Insight and its association with sociodemographic and clinical variables at admission
The level of insight into illness was compared regarding different sociodemographic, illness-related and clinical variables shown in Table 2. Patients with poorer insight were found to be significantly more likely first-episode patients (p = 0.014). These patients were found to suffer significantly more often from suicidality (p = 0.0027) concurrently being more adherent to treatment (p<0.0001). Also, patients with poorer insight scored significantly lower on the PANSS total score (p<0.0001) and the PANSS positive (p<0.0001) and PANSS negative subscore (p = 0.0037) while having a significantly better functioning (GAF: p<0.0001; SOFAS: p<0.0001).
a PANSS general psychopathology subscore was calculated excluding the PANSS G12 item.
3.4. Change of insight during the study and its association with baseline sociodemographic and clinical variables
Patients suffering from their first illness episode (p<0.0001) and from a shorter duration of illness (p = 0.0004) were found to be significantly more likely amongst the patients improving regarding insight or having a stable level of good insight (Table 3). A significant improvement of insight was found in patients suffering from suicidality (p = 0.049) and those adherent at admission (p<0.0001). Furthermore, patients improving in terms of their insight into illness or keeping their rating of good insight were found to be significantly more often in a working position (p = 0.009) with significantly better functioning (p = 0.0024) and a more satisfying subjective well-being (p = 0.047) at admission. These patients also scored significantly lower on the PANSS total score (p<0.0001), the PANSS positive (p<0.0001), PANSS negative (p = 0.0022) and PANSS general psychopathology subscore (p = 0.0044) as well as on the HAMD total score (p = 0.0096) at admission.
a PANSS general psychopathology subscore was calculated excluding the PANSS G12 item.
3.5. Predictors of poor insight at discharge
Results of the ordinal regression model revealed that being non-suicidal (p = 0.0218), suffering from multiple illness episodes (<0.0001), being non-adherent (p = 0.0012) and suffering from only few depressive symptoms (p = 0.0004) at admission were significant predictors of poor insight at discharge (Table 4). The goodness-of-fit as measured by Nagelkerke's R2 was 0.40.
4.1. Insight in schizophrenia patients and its course during treatment
Our result finding 32% of the patients to have good insight into their illness and 23% of the patients to have only minimal or mild impairments in their insight might be surprisingly positive on the first glance when keeping earlier literature reports in mind finding up to 80% of schizophrenia patients not believing that they have a disorder [Reference Amador and Gorman3]. One possible explanation for this result might be associated with the rather moderate illness severity of the examined patients’ psychopathological symptoms (PANSS TOTAL admission mean score: 71.2 (±19.2)) because several authors found an association between the severity of psychopathological symptoms and the degree of impairments of insight. Wiffen et al., for example, discuss that the relatively high level of insight in their sample (29% of their study sample with no impairments in insight) was almost expected given the psychopathological condition of the patients with a mean PANSS total score of 66.3 points (SD 18.46) mirroring similar illness severity as in the present study [Reference Wiffen, Rabinowitz, Fleischhacker and David58].
One strength of this study is the non-dichotomic analysis of the deficits in insight and their course during the treatment. We were able to show that most patients with good insight at admission had similarly satisfying insight ratings at discharge (Figure 3). Starting with an insight rating of “moderate lack of insight” at admission to a rating of “severe/extreme lack of insight” some patients improved only minimally (1 point on the 7-point likert scale) or even remained to have the same impairments at discharge than at admission. This is interesting for it is generally believed that the patients’ insight improves during inpatient treatment suggesting that with the individual's psychopathological improvement impairments of insight into illness decrease [Reference Mintz, Addington and Addington31]. We were able to show that insight was stable in most patients after discharge underlining the importance of improving insight into illness during the acute treatment phase given that there is no further significant improvement of insight after discharge.
When discussing these results, it should be kept in mind that the evaluation of insight into illness is difficult and a gold standard measuring this multidimensional construct is still not yet available. Many authors have proposed that given the fact that insight is a multidimensional concept it should be examined applying multidimensional rating scales [Reference Raffard, Bayard, Capdevielle, Garcia, Boulenger and Gely-Nargeot41]. However, previous research shows that single-item measures of insight into illness provide a good estimate of insight strongly correlating with more detailed schedules [Reference Sanz, Constable, Lopez-Ibor, Kemp and David44] suggesting that our results might adequately mirror the patient's level of insight.
4.2. Insight and its influencing factors at admission and during inpatient treatment
4.2.1. Illness-related variables
It has been reported that patients suffering from their first episode of schizophrenia are less aware of having a mental illness than multiple-episode patients and that insight improves after the first episode [Reference Parellada, Boada, Fraguas, Reig, Castro-Fornieles and Moreno36]. Our results finding first-episode patients to have a significantly poorer insight than multiple-episode patients at admission with a significant improvement of insight during the treatment confirm these findings. However, in contrast to the literature we found first-episode patients to feature a significantly greater insight at discharge compared to the multiple-episode patients and suffering from multiple illness episodes was furthermore identified to be a significant predictor of having poor insight at discharge. From a clinical point of view this result is not surprising given that it is well known that first-episode patients respond very well to treatment [Reference Gaebel15] resulting in a satisfying improvement of symptoms and with this possibly improvement of insight. Also, it asks a lot of a patient in his first episode to be able to label his symptoms correctly and understand that he is suffering from a mental illness when this is a totally new experience. The improvement of insight in these patients might therefore also mirror psychoeducational strategies provided during inpatient treatment. Besides, it should be kept in mind that first-episode schizophrenia is thought to constitute a very heterogeneous group including different psychosis categories suggesting that sample characteristics might have an even greater influence that in other patient populations [Reference Flashman, Roth, Amador and David13].
In agreement with our finding of first-episode patients suffering from less impairments in insight at discharge we found a shorter duration of illness and fewer numbers of previous hospitalizations to be significantly associated with good insight and greater improvement of insight during the treatment. Just recently Gastal and Januel reported of a link between better insight and a shorter evolution of the disease on the one hand and of a taboo status concerning their insight into illness in patients suffering from schizophrenia for many years [Reference Gastal and Januel16].
4.2.2. Psychopathological variables
In the present study the severity of psychopathological symptoms, especially positive symptoms, was found to be significantly associated with the patients’ insight in that sense that patients with more psychopathological symptoms had poorer insight. Somewhat in contrast to that finding is our result of suicidal and depressed patients having better insight even though suicidal patients are usually the ones suffering from more severe psychopathological symptoms. Depressive symptoms and suicidality were also identified to be significant predictors of good insight at discharge in our study.
Comparative literature gives no clear guidance in terms of the association between insight and psychopathological symptoms. Some authors believe that symptomatology may only play a small role in the level of insight [Reference Mintz, Dobson and Romney30] and other clinical and premorbid factors might be more important. However, in a specific analysis on symptomatic determinants of insight in schizophrenia spectrum disorders Buchy et al. also suggested that depression/anxiety is associated with better insight and that psychomotor excitation and delusions are associated with poorer insight [Reference Buchy, Torres, Liddle and Woodward8]. And when exploring insight, functioning and psychopathology amongst schizophrenia inpatients Stefanopoulou et al. stated that based on their result of a positive correlation between insight and the severity of psychopathology insight is at least partly cross-related to the psychotic illness process itself [Reference Stefanopoulou, Lafuente, Saez Fonseca and Huxley54]. Furthermore, when evaluating insight in early schizophrenia Mintz et al. found insight to have improved as positive symptoms improved [Reference Mintz, Addington and Addington31].
Regarding the relationship between insight, depressive symptoms and suicidality most researchers have shown a link between better insight and greater severity of depressive symptoms [Reference Carroll, Pantelis and Harvey9] believing that the relationship between depressive symptoms and insight might be explained by the absence of self-deception as a denial defense [Reference Moore, Cassidy, Carr and O’Callaghan33]. Similarly, based on the so called demoralization theory stating that “as patients become more acutely aware of symptoms, demoralization and fear of their future lives with this illness may develop” greater insight is believed to lead to depression and hopelessness resulting in an increased risk of suicidality [Reference Restifo, Harkavy-Friedman and Shrout42]. In a recent qualitative analysis on suicide attempts and schizophrenia Crumlish et al. suggested that self-reported recognition of having a mental illness may be predictive of depression and suicide attempts [Reference Crumlish, Whitty, Kamali, Clarke, Browne and McTigue12]. And in a recently published study on suicidal behavior and insight into illness in schizophrenia spectrum disorders Kao & Liu were able to show a significant association between insight into illness and current suicidal ideations even after controlling for depressive symptoms .
4.2.3. Adherence to treatment
The link between insight into illness and adherence to treatment seems obvious and a clear association between poor insight and impaired adherence has been consistently reported. This is confirmed by present results also identifying non-adherence at admission to be among the few predictors of poor insight at discharge. In a cluster analytical approach evaluating how clinical factors interact with the adherence profile in schizophrenia patients Beck et al. encouraged care givers to consider the patient's subjective adherence attitude profile and specific risk factors for nonadherence including lack of insight . It has been discussed that the strong association between insight and adherence might fade over time [Reference Yen, Chen, Ko, Yeh, Yang and Yen60] possibly because insight is a necessary but not a sufficient precondition of adherence and other factors of the illness might gain more influence on the patients’ adherence over time. However, in their study on cross-sectional and longitudinal relationships between insight and the patients’ attitude towards adherence in schizophrenia Mohamed et al. also found a consistent association between insight and adherence in their longitudinal analyses [Reference Mohamed, Rosenheck, McEvoy, Swartz, Stroup and Lieberman32]. The authors discuss a causal relationship between these variables emphasizing the potential value for interventions improving the patient's awareness of both the nature of the illness and the potential benefits of medical treatment improving outcome [Reference Mohamed, Rosenheck, McEvoy, Swartz, Stroup and Lieberman32].
4.2.4. Functional variables
Traditionally, impairments in insight have been linked to poorer functioning which is confirmed by present results [Reference Baier6]. In a systematic review Lincoln et al. pointed out that insight had a particular impact on improvements in functioning and less association at baseline which was also the case in the study at hand [Reference Lincoln, Lullmann and Rief26]. This might be due to the fact that the relationship between functioning and insight seems to be mediated by symptom severity suggesting that with the improvement of psychopathological symptoms the patient's functioning improves and with it impairments in insight. The strong association between insight and functioning has just recently been pointed out by Cechnicki & Wojciechowska who examined social networks and outcome in schizophrenia patients seven years after the first hospitalization and found large extra-familial networks with high level of support to be correlated with better insight into illness [Reference Cechnicki and Wojciechowska10]. Such results underline the importance of including specific social and networking treatment strategies in the multidimensional concept of treating schizophrenia in order to positively influence and enhance insight into illness.
4.2.5. Other variables
At discharge patients with greater insight into their illness suffered from less side effects. There are several reports in the literature finding a significant association between side effects and insight, however, mostly additionally assessing and linking adherence to treatment to side effects and insight [Reference Goldberg, Green-Paden, Lehman and Gold18]. Goff et al. for example showed that negative attitudes towards adherence reflect the presence of side effects and lack of insight [Reference Goff, Hill and Freudenreich17]. However, the question of whether or not a patient needs insight into his illness to detect and complain about side effects as suggested by present results and as expected from a clinical point of view remains unclear. Future studies are warranted to shed more light on this potential causal relationship.
Regarding the influencing factor subjective well-being we were not able to find a linear relationship between insight and subjective well-being as described in the literature. Karow et al. reported that patients with better insight into illness possibly realize their restrictions more clearly resulting in a decrease in their quality of life [Reference Karow, Pajonk, Reimer, Hirdes, Osterwald and Naber22]. However, in contrast to our study evaluating subjective well-being using the SNW-K, Karow et al. applied the modular system of quality of life questionnaire (MSQoL) which might at least partly explain the differing results [Reference Karow, Pajonk, Reimer, Hirdes, Osterwald and Naber22].
Another significant association observed in this study is the link between insight and full-time activity. Generally, it is believed that impairments in insight go along with deficits in work function even though only few studies explicitly focused on evaluating the association between insight and work performance. One study performed by Lysaker et al. examined schizophrenia patients having impaired/unimpaired insight and showed that patients with impaired insight had significantly poorer ratings of work quality, work habits, cooperativeness and personal presentation [Reference Lysaker, Bryson and Bell28] The significant association between occupation and insight in schizophrenia patients was just recently underlined by results of Poo et al. who analysed an occupational therapy program and found patients with better insight to be the ones with a more favorable vocational outcome [Reference Poon, Siu and Ming39]. Possibly these results go back to findings from imaging studies reporting of global and regional grey matter deficits in the cingulate gyrus associated with symptom self-appraisal [Reference Morgan, Dazzan, Morgan, Lappin, Hutchinson and Suckling34] and insight function being significantly linked to both, the volume and activation of the dorsolateral prefrontal cortex [Reference Shad, Keshavan, Tamminga, Cullum and David50,Reference Pia and Tamietto38] suggesting impairments in working tasks and executive function in patients with poorer insight in turn explaining impairments in the work performance of the patients.
4.3. Strengths and limitations
The strength of this study is the non-dichotomic analysis of insight and its course during treatment which has not been performed so far. Also, the sample size is very satisfying and only standardized rating scales were applied. A limitation of this study is the fact that a single item was used to evaluate insight into illness therefore possibly not adequately complying with the multidimensional approach of insight. Another limitation is that neurocognition, a well known influencing factor of insight, was not evaluated in this study. Furthermore, the ratings of the patients’ psychopathology and insight were performed by the same rater making them not independent, possibly biasing present results. The naturalistic study design is both a strength and a limitation. On the one hand such a design does not allow a sufficient control of study results for the effect of different pharmacological treatments but mirrors a real-world situation. Also, due to the liberal inclusion and exclusion criteria, findings of this study on treatment seeking patients might be more generalizable and exhibit higher external validity than results from RCTs.
Almost 70% of schizophrenia patients showed some kind of impairment in their insight into illness at admission significantly improving during the course of the study. Poor insight at admission as well as the improvement of impairments in insight were primarily associated with the patients’ psychopathology, functioning and treatment adherence. Fewer depressive symptoms, less suicidality and worse adherence at admission were identified to be significant predictors of poor insight at discharge. Specific psychotherapeutic programs as well as greater awareness regarding insight and strengthening the therapeutic alliance might help improving insight without the risk of deteriorating mood or suicidality.
Disclosure of interest
The authors declare that they have no conflicts of interest concerning this article.