The concept of subjective well-being in schizophrenia is the subject of an ongoing and controversial debate. Ranging from incomprehension to enthusiasm the positioning of quality of life in the treatment and evaluation of outcome in schizophrenia is still evolving Reference Bobes, Garcia-Portilla, Saiz, Bascaran and Bousono. Interestingly, studies often found high levels of general life satisfaction despite deprived living conditions in patients with schizophrenia.
This is why some researchers doubt whether schizophrenia patients are capable of self-assessment of their quality of life because of cognitive deficits and lack of insight into their illness or illness-associated reduced expectations Reference Bobes, Garcia-Portilla, Saiz, Bascaran and Bousono. Franz et al. believe that experiences of restricted and deprived living conditions might induce accommodation processes and response-shifts, which should be taken into account when interpreting quality of life data Reference Franz, Meyer, Reber and Gallhofer. Other authors were able to demonstrate convergent validity in the perception of quality of life between patients and clinicians and showed that schizophrenic patients feel and experience their deficits, supporting the thesis that quality of life can be assessed subjectively in schizophrenics Reference Lehman, Postrado and Rachuba[26,Reference Skantze, Malm, Dencker, May and Corrigan42].
Since then, several trials have been performed to assess and classify patients’ subjective well-being in schizophrenia. Psychopathological symptoms Reference gtsson-Tops, Hansson, Sandlund, Bjarnason, Korkeila and Merinder[15,Reference Sim, Mahendran, Siris, Heckers and Chong41] and side effects of antipsychotic treatment have among other factors been found to significantly influence the patient's well-being Reference Voruganti, Heslegrave and Awad[45,Reference Young, Sullivan, Burnam and Brook49]. Most important, as impairments in well-being often result in higher rates of non-compliance the patients’ subjective well-being has become a frequent additional outcome criterion Reference Rettenbacher, Hofer, Eder, Hummer, Kemmler and Weiss[36,Reference Wehmeier, Kluge, Schneider, Schacht, Wagner and Schreiber46]. Also, attempts have been made to predict subjective well-being in order to enhance treatment outcome and patients’ compliance Reference Caron, Lecomte, Stip and Renaud.
However, studies evaluating predictors of well-being are rare and results have not shown any consistent picture Reference Ritsner, Gibel and Ratner. This is mainly due to the fact that only sparse sets of predictor variables were used, additionally varying between studies. Recently, as one of the first, Lambert et al. examined response and remission of subjective well-being and found remission mainly predicted by low negative symptoms, better social functioning and early response in subjective well-being Reference Lambert, Naber, Eich, Schacht, Linden and Schimmelmann. However, the authors focused on outcome of severely impaired subjective well-being in patients with schizophrenia, thereby possibly overlooking valuable information regarding the heterogeneity of schizophrenia and the multidimensional construct of well-being.
Therefore, the aims of the present study were to:
• assess subjective well-being in a heterogeneous schizophrenia inpatient sample;
• to find clinical and sociodemographic variables predictive for response and remission of subjective well-being.
2.1. Study design
A multicenter follow-up programme (German Research Network on Schizophrenia) Reference Wolwer, Buchkremer, Hafner, Klosterkotter, Maier and Moller was carried out at 11 psychiatric university hospitals and three psychiatric district hospitals. Patients with the diagnosis of schizophrenia (paranoid, disorganized, catatonic or undifferentiated subtype), schizophreniform disorder, delusional disorder and schizoaffective disorder according to DSM-IV criteria who were admitted between January 2001 and December 2004 to one of the above-mentioned hospitals were randomly selected. The patient selection was carried out with a randomisation software. Patients included in the study were aged between 18 and 65 years. The exclusion criteria were a head injury, a history of major medical illness and alcohol or drug dependency. To participate in the study, an informed written consent had to be provided to. The study protocol was approved by the local ethics committees Reference Jager, Riedel, Messer, Laux, Pfeiffer and Naber.
Clinical researchers assessed DSM-IV diagnoses on the basis of the German version of the Structured Clinical Interview for DSM-IV . During interviews with patients, relatives and care providers, sociodemographic variables (partnership, employment state) and course-related variables such as age at onset, duration of untreated psychosis or episodes of illness were collected using a standardized documentation system (BADO) Reference Cording.
The Subjective well-being under neuroleptic treatment scale, short version (SWN-K) was used to assess well-being. Reference Naber. Symptom severity was assessed via the Positive and Negative Syndrome Scale for schizophrenia (PANSS) Reference Kay, Opler and Lindenmayer. To quantify antipsychotic side effects the Udvalg for Klinske Undersogelser (UKU) side effect rating scale Reference Lingjaerde, Ahlfors, Bech, Dencker and Elgen was used comprising a total of 48 symptoms classified into four groups: psychic, autonomic, neurological and other side effects.
Ratings were performed within the first 3 days after admission and biweekly during hospital stay until discharge. Since in the present analysis, possible predictor variables at baseline were examined, only PANSS baseline data is presented. To better detect possible side effects, the UKU total score was not only examined at baseline, but additionally at week 2. The observation period in the presented analyses was 8 weeks of treatment. All raters had been trained using the applied scales. A high inter-rater reliability was achieved (ANOVA-ICC > 0.8).
2.3. Statistical analysis
Definition of SWN-K early response, SWN-K response and SWN-K remission following Lambert et al. Reference Lambert, Naber, Eich, Schacht, Linden and Schimmelmann:
• early response was defined as a SWN-K total score improvement of 20% and by at least 10 points within the first two treatment weeks;
• the authors defined subjective well-being response as improvement in SWN-K total score of at least 20% and by at least 10 points from admission to discharge;
• remission in subjective well-being was defined as a total score of more or equal to 80 points at discharge.
Symptomatic response was defined as a PANSS total score reduction of 40% from baseline to discharge Reference Kinon, Chen, scher-Svanum, Stauffer, Kollack-Walker and Sniadecki and remission according to the Remission in Schizophrenia Working Group's criteria Reference Andreasen, Carpenter, Kane, Lasser, Marder and Weinberger.
2.3.1 Statistical methods
The Wilcoxon test, a non-parametric test for comparing two independent random variables (responder/non responder, remitter/non remitter) and for at least ordinal scaled variables, the t-test was applied. The t-test was applied when the population was assumed to be normally distributed. For categorical data the Fisher's exact test was used. Furthermore, the Pearson's correlation coefficient was applied to estimate the linear relationship between two random variables.
Two different methods were used to select the most important predictors of response and remission:
• regression model;
• classification and regression trees (CART) analysis.
Starting with a mixed regression model including clinical and sociodemographic variables, a backward-forward method based on the Akaike Information Criterion was used to identify the relevant predictors. Eight BADO items representing the patients’ clinical, sociodemographic and functional status were investigated with respect to their predictive value (age, gender, duration of illness, length of current episode, length of hospital stay, suicidal behaviour before hospitalisation, living situation (with partner/single), employment status (employed, unemployed, retired). Also, SWN total score at baseline and SWN early improvement, PANSS positive, negative and global subscores at baseline, the UKU total score from baseline, week 2, the percentaged change from baseline to week 2 and the kind of medical treatment applied (typical antipsychotics, atypical antipsychotics, typical and atypical antipsychotics) were considered potential predictors. The final model was computed with only these predictors. The discriminative ability of the regression model was also evaluated using a receiver-operating characteristic (ROC) curve. The area under the curve (AUC) is a measure of the overall discriminative power.
A value of 0.5 for the AUC does not represent discriminative ability, whereas a value of 1.0 indicates a perfect power. A value of 0.7 to 0.8 is considered as a reasonable value and a value greater than 0.8 as a good discriminative capacity Reference Weinstein and Fineberg. Additionally, the CART were modelled to confirm the results of the linear model.
All statistical analyses were performed using the statistical program R2.6.1 .
The entire multicenter study comprised 474 patients. Forty-six patients dropped out for different reasons (e.g. retrospective violation of inclusion criteria, withdrawal of informed consent, incomplete information). Another 196 patients had to be excluded from analysis: 28 patients because they were discharged from the hospital within 7 days after admission, 168 patients due to missing SWN-K data. Therefore, the sample available for analyses enrolled 232 patients, with 135 male and 97 female patients. A comparison between dropped-out patients and study participants is shown in Table 1. The mean number of clinical treatments was 2.97 (±3.86) and the length of current period was less than 1 month for 82 patients, less than 3 months for 36 patients, less than 6 months for 38 patients, less than 2 years for 25 patients and less than 10 years for 30 patients. For 15 patients, the length of current period was more than 10 years and six patients could not be assigned. The length of hospital stay was 74.26 days (± 47.13) and the mean age at first treatment 28.15 years (±9.32). 204 patients reported to be living on their own, 27 lived together with a partner/family. 176 patients had graduated from school with a certificate, whereas 56 patients were without a graduation certificate. One hundred and seven patients were employed, 43 unemployed and 77 not working on a regular basis. Suicide attempts before hospitalization were documented for 53 patients. DSM-IV diagnoses are listed in Table 2.
Patients were treated under naturalistic conditions as follows: 46.23% of the patients received typical antipsychotics, 68.75% atypical antipsychotics and 41.62% of the patients were treated with typical as well as atypical antipsychotics. Furthermore, in 63.27% of patients tranquilizers were administered, in 9.46% mood stabilizers and in 26.54% of the patients antidepressants.
3.2.1 Clinical characteristics and improvement of subjective well-being
Psychopathological characteristics and the UKU scale are presented in Table 2. The SWN-K total score at admission and from week 2 to week 8 in biweekly intervals up to discharge is shown in Tables 3 and 4.
3.2.2. Subjective well-being of responders and remitters
Of the 232 patients, 67 patients (29%) were detected to be SWN-K early responders. Ninety-three patients (40%) fulfilled criteria for response in subjective well-being, 139 patients were non responders (60%). One hundred and fifty-three patients (66%) fulfilled criteria for remission and 79 patients (34%) were non remitters concerning subjective well-being at discharge. The Phi-coefficient between early response and response was 0.47, suggesting a strong association between the two measures. Subjective well-being responders/non responders and remitters/non remitters are compared in Tables 5 and 6.
a Subjective well-being under neuroleptic treatment scale.
a Subjective well-being under neuroleptic treatment scale.
3.2.3. Association of subjective well-being and symptomatic change and outcome
No significant association was found between SWN-K and PANSS responders (p = 0.85), nor between SWN-K and PANSS remitters (p = 0.32). The correlation between the subjective improvement measured by the SWN-K and an objective improvement measured by the PANSS was again not found to be significant (Pearson's correlation coefficient: −0.07).
3.2.4. Regression models for response and remission
22.214.171.124 SWN-K response
Among the investigated predictors, SWN-K early improvement and the educational status were significantly associated with SWN-K response. The SWN-K total score at baseline showed a significant negative predictive value for response. Receiver operating characteristics (ROC) revealed an AUC of 0.80, indicating very reasonable predictability for this set of predictors (Table 7, Fig. 1).
126.96.36.199 SWN-K response
CART-analysis confirmed all three predictors identified by the regression model (SWN-K early improvement, SWN-K total score, educational status). CART-analysis further found a low PANSS global subscore and fewer side effects at baseline to be significant predictors of SWN-K response (Fig. 3).
188.8.131.52. SWN-K remission
SWN-K total score at baseline was confirmed significant predictor for remission by CART-analysis (Fig. 4).
4.1. Sociodemographic and clinical baseline variables
Out of all evaluated sociodemographic and clinical baseline variables, the existence of a school degree was found to be significantly predictive for response and remission of subjective well-being via both statistical methods. One explanation for this result might be that patients with a graduation certificate are the ones cognitively less impaired, as they were able to complete school successfully. Therefore, these patients might show greater insight into their illness, as other authors were able to demonstrate a negative association between cognitive dysfunction and insight Reference Karow, Pajonk, Reimer, Hirdes, Osterwald and Naber. We believe that this patient cohort is more capable of self-assessment and able to evaluate changes in their well-being better than patients with less insight, cognitively more impaired and maybe without a graduation certificate. Regarding comparative literature, a higher educational level has often been associated with greater general well-being in other trials Reference Carpiniello, Lai, Pariante, Carta and Rudas[5,Reference Vandiver44].
Results of the univariate tests found patients in subjective well-being remission to be significantly less suicidal before admission than patients without subjective well-being remission. The importance of quality of life in the context of suicidality and suicidal attempts in patients with schizophrenia has been of high scientific interest in the past few years, and impairments in subjective well-being have been associated with repeated suicide attempts Reference Ponizovsky, Grinshpoon, Levav and Ritsner. This is also underlined by long-term study results finding life satisfaction in the general population to have a long-term effect on the risk of suicide Reference Koivumaa-Honkanen, Honkanen, Viinamaki, Heikkila, Kaprio and Koskenvuo. Therefore, the need for clinicians to include aspects of quality of life in the evaluation of patients with schizophrenia that are suspected of being suicidal should be highlighted Reference Pedrelli, McQuaid, Granholm, Patterson, McClure and Beck.
4.2. Subjective well-being
We found the SWN-K total score at baseline to be significantly predictive for response and remission of subjective well-being Reference Lambert, Schimmelmann, Naber, Schacht, Karow and Wagner. Keeping both outcome definitions in mind, this result is almost to be expected and is in good accordance with the so-called law of initial value (LIV), which suggests that the magnitude of any psychobiological response is dependent on the initial baseline level Reference Myrtek and Foerster. The influence of baseline well-being for its course and outcome is furthermore underlined by long-term data. When evaluating influencing factors of well-being in a 6-year follow-up period using different assessment tools, Hannson and Björkman showed that the only baseline predictor of subjective quality of life was baseline quality of life Reference Hansson and Bjorkman. They believe that the degree of satisfaction with emotional and social relations represses symptoms of unmet needs and illness, especially over the course of treatment. In a 16-month-follow-up, Ritsner et al. found the baseline levels of emotional distress, oriented coping and self-esteem to explain 41% of the variability in the general subjective well-being index 16 months later Reference Ritsner, Gibel and Ratner.
In addition, we found SWN-K early response to be a significant predictor of outcome in subjective well-being. The importance of early subjective well-being response as a major determinant of long-term satisfying quality of life in schizophrenia has already been highlighted by Lambert et al. Reference Lambert, Naber, Eich, Schacht, Linden and Schimmelmann. In their study, the authors found only 9% of patients without early response of subjective well-being to achieve remission in well-being after 12 weeks of treatment, emphasizing the necessity to improve well-being early in order to resolve with moderate and stable subjective well-being. Given the strong association (Phi-coefficient = 0.47) between early response and response, our results are not surprising. However, we consider the significant influence of early response on subsequent response not as an innovative and unexpected predictor revealed to be important for subjective well-being outcome, but more as an adjusting variable in the statistical model applied. Additionally, against the backdrop of the importance of early symptomatic response for subsequent symptomatic outcome, the necessity to apply treatment radically and early in the course of the illness is also highlighted considering the course of subjective well-being Reference Emsley, Rabinowitz and Medori[10,Reference Emsley, Rabinowitz and Medori11].
Regarding psychopathological predictors, we found the PANSS global subscore to be significantly predictive for response and remission in subjective well-being. Also, considering the univariate tests, patients with remission of subjective well-being featured significantly fewer negative symptoms than patients not achieving remission criteria. Symptoms of depression, which are among other means assessed via the PANSS global subscore, and negative symptoms have most consistently and most strongly been related to subjective well-being Reference Corrigan and Buican[8,Reference Mueser, Douglas, Bellack and Morrison29,Reference Sim, Mahendran, Siris, Heckers and Chong41]. The PANSS negative and global subscores have also been found to be significantly correlated with the SWN total score, underlining the strong relation between subjective well-being and psychopathology Reference Putzhammer, Perfahl, Pfeiff and Hajak. Findings that the level of general psychopathology strongly influences the patients’ well-being also highlight the importance of the patients’ symptomatic status Reference gtsson-Tops and Hansson[14,Reference Sullivan, Wells and Leake43]. This suggests that an early and adequate symptom control might be the precondition for experiencing satisfying subjective well-being. The important influence of psychopathology on well-being is furthermore underlined when study drop-outs and study participants are compared. One hundred and sixty-eight had to be excluded due to missing SWN-K data, and the only difference between these patients and those who had filled in the SWN-K scale is that study drop-outs suffered from significantly more psychotic symptoms.
In addition, we found patients achieving remission to score significantly higher on the PANSS insight into illness item. Previous studies found good insight into having a mental disorder significantly related to better quality of life Reference Schwartz. On the other hand, good insight has been associated with greater compliance and satisfaction with antipsychotic treatment Reference Mintz, Addington and Addington[28,Reference Pedrelli, McQuaid, Granholm, Patterson, McClure and Beck32]. Therefore, consequently improved insight is one of the major goals in the treatment of patients suffering from schizophrenia Reference Cooke, Peters, Kuipers and Kumari. However, there are also recently published studies finding a negative relationship between insight and subjective well-being Reference Ritsner[37,Reference Sim, Mahendran, Siris, Heckers and Chong41]. There are some possible explanations for these differing results. Some studies evaluated patients including diagnoses other than schizophrenia spectrum disorders, and the psychopathological stability also seems to influence results on insight and well-being Reference Karow, Pajonk, Reimer, Hirdes, Osterwald and Naber. Furthermore, in the present study, insight was assessed using the PANSS G12 item only instead of using a multidimensional measure of the construct of illness insight Reference Karow, Pajonk, Reimer, Hirdes, Osterwald and Naber.
4.4. Side effects
We found that fewer side effects at baseline predicted SWN-K response and remission, and that patients with remission status suffered from significantly fewer side effects already at baseline. Antipsychotic side effects have often been associated with a substantial reduction of subjective well-being, suggesting that patients with fewer side effects feel less impaired in their well-being Reference Ritsner, Ponizovsky, Endicott, Nechamkin, Rauchverger and Silver. Especially, the distress resulting from akathisia and its far-reaching impact on clinical stability, well-being and even the rate of suicide have been noted by a number of different authors Reference Drake and Ehrlich.
In view of the strong relationship between dopamine D2 receptor occupancy, extrapyramidal side effects (EPS), and neuroleptic dysphoria, one might reasonably presume a different impact on well-being comparing typical and atypical antipsychotics Reference Fleischhacker, Maj and Sartorius. This assumption is underlined by reports in the literature that did not find antipsychotic side effects to be predictive for subjective well-being when patients were treated with atypical antipsychotics Reference Fleischhacker, Maj and Sartorius[12,Reference Lambert, Naber, Eich, Schacht, Linden and Schimmelmann24]. Interestingly, the type of treatment (comparing typical/atypical/and a combination of typical and atypical antipsychotics) was not revealed to be a significant predictor by the statistical model in the present study. And keeping in mind that in the present study, 70% of the patients were treated with atypical antipsychotics, it is surprising that we found a significant association between side effects and subjective well-being. However, this might be due to the heterogeneity of the atypical antipsychotic treatment applied. All antipsychotics available could be administered, including antipsychotics such as risperidone or ziprasidone that are known to induce EPS such as akathisia Reference Kane, Fleischhacker, Hansen, Perlis, Pikalov and ssuncao-Talbott. This suggests that the favorable influence of atypical antipsychotics on subjective well-being might depend on the individual compound prescribed. Although atypical antipsychotics cause fewer EPS, they have been found to cause other disturbing side effects such as sedation, body weight gain or sexual dysfunctions which can also influence the patient's well-being Reference Hofer, Kemmler, Eder, Edlinger, Hummer and Fleischhacker. Taken together, it should be kept in mind that this was a naturalistic trial limiting a sufficient control of psychopharmacological effects. No treatment regimen was recommended, and clinicians were able to adjust the medical treatment according to their assessment of the patient's psychopathological status and well-being.
4.5. Strengths and limitations
In this multicenter trial, patients were treated under naturalistic conditions, and such a design lacks a sufficient control of study results for the effect of different psychological and pharmacological treatments. Also, patients were treated with different typical as well as atypical antipsychotics, so that homogeneity of antipsychotic treatment and potential side effects might be limited. Furthermore, as neurocognition was not examined, the potential link between cognition and insight into illness cannot be confirmed by the present data. Altogether, this study has a hypothesis-generating rather than a confirmatory character. Another limitation is that depression, in this study, was assessed via the PANSS scale and without using a depression-specific rating instrument.
The strengths of this study include the rather large sample in the assessment of well-being. In addition, due to the liberal inclusion and exclusion criteria, findings of this study on treatment-seeking patients might be more generalizable and have a greater external validity than in randomised controlled trials.
Considering that the Remission in Schizophrenia Working Group phrased the importance of the patients’ functional remission comprising quality of life for treatment outcome, the need for an early identification of patients with an impaired subjective well-being or without early improvement is highlighted by the study's results Reference Andreasen, Carpenter, Kane, Lasser, Marder and Weinberger. Depressive and negative symptoms should be radically treated and side effects closely monitored. Regarding the important influence of subjective well-being on overall treatment outcome, clinicians should be encouraged to improve the patients’ subjective well-being in order to positively influence their overall long-term outcome.
The study was performed within the framework of the German Research Network on Schizophrenia, which is funded by the German Federal Ministry for Education and Research BMBF (grant 01 GI 0233).