Background
Quality of life (QoL) is an important outcome in schizophrenia research and was the primary outcome of the UK CUtLASS trials (Cost Utility of the Latest Antipsychotics in Severe Schizophrenia) [
1]-[
3]. Self- or patient-rated measures of QoL in schizophrenia can be viewed with suspicion by the clinician [
4] due to the perceived impact of depressive and psychotic symptoms, poor insight, and cognitive deficits [
5],[
6]. The alternative is clinician- or assessor-rated measures of QoL but these show only moderate correlation with self-ratings [
7]-[
9].
Existing literature converges on extensive differences between the predictors of subjective and objective QoL. Previous researchers have found, for example, with the use of factor analysis, that subjective and objective QoL components cluster separately [
10]. The majority of studies have examined the predictors of patient-rated and assessor-rated QoL separately, for example, research on outpatients with schizophrenia concluded that objective and subjective QoL measures have different clinical predictors and should be considered separate outcomes [
11]-[
13]. Similarly, a recent review highlighted that few studies had investigated the relationship between subjective and objective QoL assessment, concluding that researchers ought to use both approaches as complementary outcome measures in schizophrenia [
14]. This message was re-iterated by Awad and Voruganti in their 2012 update on QoL measurement in schizophrenia [
15].
Two major QoL measures used for schizophrenia are the objective QLS (Quality of Life Scale) [
16] and the LQOLP (Lancashire Quality Of Life Profile) [
17]. The QLS consists of a series of questions regarding common social achievements (see Methods section). The LQOLP consists of a series of questions including two visual analogue scales of subjective life satisfaction (see Methods section).
It is not enough, however, to know that the two types of measure differ. An important issue to examine is what makes them differ. QoL measurement discrepancy serves to highlight fundamental differences between the priorities of the patient and those of the clinician, which may drive patient dissatisfaction with treatment. In addition, knowledge of the drivers of QoL discrepancy can inform research using QoL as an outcome measure.
Despite a large QoL evidence base, few researchers have specifically examined predictors of discrepancy between the two methods of QoL measurement in the same individual. One such study did suggest that patients with QoL measurement concordance were more likely to be young, male and have high symptom scores [
18]. Patient ratings in this study, however, were measured using a subjective measure of function; the Subjective Well-being on Neuroleptics (SWN) scale [
19], rather than with a specific quality of life measure [
18]. Other studies have employed specific QoL measures. Poor insight was linked to greater discrepancy between objective (Standard of Living Interview SOL-I) [
20] and subjective (LQOLP) [
17] measures of QoL by Doyle and colleagues [
6]. Similarly, Whitty et al. [
21] suggest that better insight may serve to lower both subjective (WHO Quality of Life Scale Brief Version) [
22] and objective QoL (QLS) [
16]. A further study using the patient's global QoL assessment and the interviewer's global QoL assessment from the same measure; the LQOLP [
17] highlighted an association between greater QoL measurement discrepancy and fewer affective symptoms [
7].
During the UK CUtLASS trials [
1]-[
3] QoL of participants at one trial site was assessed using the patient-rated LQOLP [
17] in addition to the observer-assessed QLS [
16]. This provided the ideal opportunity to assess in detail what factors affect each construct and (most importantly) the discrepancy between the two.
Discussion
A significant but moderate correlation (
r = 0.38) was found between patient-rated and assessor-rated measures of QoL in patients with schizophrenia (N = 80) participating in a clinical trial. This confirms previous findings [
7],[
8],[
39], whilst pointing to the existence of differing constructs of the concept of QoL on the part of the patient and the assessor, or clinician. Separate analyses of the two QoL measures allowed us to understand the drivers of the discrepancy between the two; a previously under-researched topic.
As in previous studies, for example, Narvaez et al. [
11], good insight and depression predicted low patient-rated QoL [
7],[
8],[
11],[
12],[
39]-[
46] but not assessor-rated QoL, driving the discrepancy between the two measures. Previous work has highlighted the relationship between good insight and depression [
47]. Increased insight will serve to decrease the patient’s rating of their own QoL, to bring it more in line with the rating given by an observer, thereby reducing QoL measurement discrepancy. Good insight could be toxic, in that it can erode self-esteem and promote self-stigmatisation; but it may also promote a positive, integrated sense of self [
48],[
49]. Of course, rating one’s quality of life as worse is not the same as experiencing a life of poorer quality.
Although QLS and subjective QoL global ratings were correlated, this could conceal substantial variation in the relationship between different aspects of objective QoL and satisfaction. Edmondson and colleagues [
50] discovered this when comparing change in various measures of social functioning and satisfaction with them. They argued that this could be explained by Zissi and Barry’s [
51] suggestion of a shift in aspirations as function improves.
Negative attitudes to medication, as measured by DAI, predicted lower objective QoL but not self-rated QoL, increasing the discrepancy. Negative attitude toward prescribed treatment is associated by the observer with reduced QoL in a way that patients do not perceive.
Good adherence (based on openly expressed attitudes and medication-related behaviour) predicted low concurrent patient-rated QoL but not assessor-rated QoL nor QoL measurement discrepancy. Other research has suggested that direct links between adherence and QoL are weak or absent [
52],[
53], although this pattern was observed using self-report, rather than externally-assessed, adherence. This type of study is not designed to address the meaning of adherence to individuals and hence understand how it affects QoL [
54]. One possibility is that good adherence, if it does not spring from positive attitudes to treatment and a sense of self-efficacy, reflects a sense of burden by illness or services. Positive attitude to medication (rather than broader service relationships) against a complex background of emotive attitudes to services and illness might not be enough.
As in previous studies, greater negative symptomatology predicted lower assessor-rated QoL [
5],[
6],[
8],[
11]-[
13],[
39],[
40],[
55] but not patient-rated QoL, suggesting that to a clinician, or assessor, extent of negative symptomatology is seen as the prime driver of quality of life. This corresponds with previous survey research indicating that psychiatrists’ concept of QoL is more illness-orientated [
56]. Negative symptoms may have little subjective impact on the patient, whereas they are apparent to an observer, or assessor [
8]. Others have argued that negative symptoms linked to the deficit syndrome in schizophrenia can serve to protect the individual's self-esteem via social withdrawal [
57], although this perspective would suggest that there ought to be a relationship between negative symptoms and subjective QoL, which was not found here. QLS score is influenced by social function, which is linked to negative symptoms. Hence, negative symptoms will impair one form of QoL measurement without affecting the other; enhancing the discrepancy.
Increased experience of non-neurological side effects was predictive of higher self-reported QoL but of neither assessor-rated QoL nor the discrepancy. Patients with a higher QoL rating may report side effects as more severe; these experiences may have a greater impact on the lives of better-functioning individuals. This relationship was only observed for non-neurological side effect scores, not extrapyramidal side effects (EPS); others have found an association between patient-rated QoL and EPS [
58]. A significant positive correlation between ANNSERS score and depression score suggests the experience of non-neurological treatment side effects contributes to depressive symptoms.
The study had a number of limitations. Data were collected during baseline assessments of a RCT; the cross-sectional design therefore means that assumptions of causality in the statistical models must be viewed with caution. In addition, the sample analysed here was comprised of patients entered into a RCT, because of poor clinical response or intolerance, although the trial did have broad inclusion criteria. Although conclusions were derived from a relatively small sample (N = 80) QoL predictive factors highlighted by this study are consistent with the existing evidence base.
In terms of measures used, a number of variables found to be associated with QoL in previous studies, for example, neurocognitive deficits, social support, coping style and anxiety, were not assessed in this sample. It is not known whether the statistical determinants of QoL measurement discrepancy observed here will apply to discrepancies between other patient- and assessor-rated QoL measures. Analysis did include demographic variables previously associated with QoL, such as gender, although other demographic data, which may also be associated with QoL, such as social class and educational level, were not available. In terms of the possible effects of medication, the main outcome paper of the CUtLASS study reported no significant effect of antipsychotic drug class on QoL [
1]. The CUtLASS study used a depression scale developed for use specifically in schizophrenia and used measures of QoL with proven reliability and validity. The QLS is argued to be largely determined by the deficit symptoms of schizophrenia [
59] leading to the strong relationship between the QLS and negative symptoms reported in previous studies [
60]. Potentially, the QLS and negative syndrome overlap, as both are reflective of poor social function. However, we found that items for anhedonia, alogia and blunting were, though conceptually distinct, as predictive of the QLS score.
Competing interests
KPH has received assistance to attend educational events from Novartis and Lilly. SWL has received speaker fees from AstraZeneca and advisory board fees from Janssen-Cilag. JAM, RJD and GD have no conflicts of interest to declare.
Authors' contributions
KPH, JAM, SWL and RJD conceived the design of the study. KPH conducted the literature search and was responsible for the first draft of the manuscript. JAM obtained the study data. SWL acquired funding for the study. KPH, RJD and GD carried out the statistical analyses. KPH, RJD and SWL contributed to the interpretation of the study findings. All authors contributed to and approved the final manuscript.