- Email: [email protected]
Negative symptoms and psychosocial functioning in schizophrenia: Neglected but important targets for treatment
European Psychiatry 27 (2012) 432–436
Original article
Negative symptoms and psychosocial functioning in schizophrenia: Neglected but important targets for treatment R. Hunter a,b,*, S. Barry c a
Psychiatric Research Institute for Neuroscience in Glasgow (PsyRING), West Medical Building, University of Glasgow, Glasgow G12 8QQ, United Kingdom Gartnavel Royal Hospital, Glasgow G12 0XH, United Kingdom c Robertson Centre for Biostatistics, Boyd Orr Building, University of Glasgow, Glasgow G12 8QQ, United Kingdom b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 8 September 2010 Received in revised form 3 February 2011 Accepted 3 February 2011 Available online 23 May 2011
Background. – Many patients with schizophrenia suffer from poor social functioning, with high levels of unemployment being one particular consequence. Negative symptoms tend to persist during periods of clinical stability and may have a detrimental effect on function. This paper aims to investigate the relationship between negative symptoms and ability to function. Methods. – The EGOFORS study measured negative symptoms in 295 schizophrenia patients in 11 European sites using the PANSS Negative Subscale and assessment scales for psychosocial function: Global Assessment of Functioning (GAF), Personal and Social Performance (PSP), Quality of Life Scale (QLS), Functional Remission of General Schizophrenia (FROGS), Psychosocial Remission in Schizophrenia (PSRS) and Subjective Wellbeing under Neuroleptics (SWN). The relationships between the PANSS Negative Subscale and the functional scales were investigated, adjusting for differences between study sites. Being in work, duration of illness, age of onset and number of years of education were also investigated for a relationship with function. Results. – There were strong, statistically significant correlations between PANSS Negative Subscale and all of the function scales (95% confidence intervals for the correlation coefficients: PSRS 0.77–0.91; FROGS 0.74–0.89; QLS 0.74–0.92; GAF 0.64–0.78; PSP 0.63–0.80) except the SWN. All of the functional scales except SWN were at least moderately related to one another. All of the items in each of the PANSS Negative Subscale and the function scales contributed to the relationships between them. Better functioning correlated strongly with participants being in work. Conclusion. – This study shows a strong and significant relationship between negative symptoms and psychosocial functioning. Given the impact of negative symptoms on psychosocial function, much more emphasis should be placed on developing effective treatments for negative symptoms, given that most patients with schizophrenia now live in community settings and require to function adequately to support their quality of life. ß 2011 Elsevier Masson SAS. All rights reserved.
Keywords: Schizophrenia Negative symptoms Psychosocial function Drug target
1. Introduction A significant proportion of people with schizophrenia continue to experience disabling symptoms (positive, negative, affective) and cognitive difficulties that affect personal, social and occupational functioning [13]. In one typical study based in six European countries, over 80% of adults with schizophrenia had persistent problems with social functioning [19]. Moreover, poor functioning in the first three years after diagnosis tended to predict long-term poor functioning. This is reflected in the high levels of unemployment among people with schizophrenia [7,17]. There has been renewed interest recently in the importance of ‘‘real world’’
* Corresponding author. Tel.: +44 141 2113550; fax: +44 141 3305175. E-mail address: [email protected] (R. Hunter). 0924-9338/$ – see front matter ß 2011 Elsevier Masson SAS. All rights reserved. doi:10.1016/j.eurpsy.2011.02.015
functioning and the development of treatment strategies for negative symptoms and cognitive impairments. In parallel with these developments in neuroscience, the ‘‘recovery model’’ has emphasized the importance of supporting individuals with schizophrenia to develop meaningful lives, with good functioning, despite continuing symptoms [14]. Negative symptoms were first described by Kraepelin in 1919 [10] as the ‘‘avolitional syndrome’’ and the term now refers to the absence or reduction in normal behaviours and functions (PANSS Negative Subscale symptoms are listed in Table 1). The classification of negative symptoms [2] is beyond the scope of this paper but persistent negative symptoms can be defined as symptoms that are either primary or secondary to schizophrenia, persist during periods of clinical stability, have not responded to treatment and which interfere with the ability to perform normal functions. Research is underway to develop new drugs that will improve
R. Hunter, S. Barry / European Psychiatry 27 (2012) 432–436
negative symptoms and the National Institutes of Health (NIH) website (www.ClinicalTrials.gov) lists many active trials in phases 1–3. Given this interest, the primary aim of the present study was to investigate the relationship between negative symptoms and psychosocial functioning in the EGOFORS cohort of people with schizophrenia. In particular, to explore the strength of the relationships between item and total scores of the PANSS Negative Subscale and items in a number of different scales used to measure psychosocial function. 2. Methods Recruitment of the 295 participants to the EGOFORS cohort in the 11 European study sites, and the selection and use of the assessment measures has been described in the accompanying paper [5]. The relationship between data from the functional scales (PSRS, FROGS, QLS, GAF and PSP see Table 1 for details of the scales used and individual scale items) and the PANSS Negative Subscale was examined (scales described in Gorwood et al. 2011 [5]), as was the relationship between the Subjective Wellbeing under Neuroleptics Scale (SWN) and the PANSS Negative Subscale. The SWN is not a psychosocial functioning scale per se but provides information on the subjective experience of patients during treatment with antipsychotic medication [3]. Data from the UPSA scale was not included given its more limited use in the cohort and is discussed elsewhere in the accompanying EGOFORS subanalysis [6]. As the PSRS items N1, N4 and N6 are shared with the PANSS scale, we have therefore considered the PSRS scale without these, in order to avoid artefactual significance. We also removed one patient from the dataset because they had missing data for both the PANSS N3 and N7 items. Correlations were calculated using the Spearman correlation coefficient because of some minor deviations from normality of the PANSS negative and total scores. The relationships between the PANSS negative subscale and the functional scales were assessed using canonical correlation analysis (CCA). This method finds the linear combinations of the items in the PANSS Negative Subscale and each functional scale that have the highest correlation with one another [11]. The item combinations are presented as so-called ‘‘canonical variables’’, in a similar way to factors in factor analysis, thereby allowing us to identify the items in each functional scale that contribute most to the relationship with the items in the PANSS Negative Subscale, and vice versa. Items with a ‘‘loading’’ (the correlation with the original variable) of more than 0.5 on a canonical variable were considered ‘‘important.’’ To account for age, duration of illness, years of education and whether the participant was in work, these characteristics were included in the canonical variables for the PANSS items. Like factor analysis, CCA finds several canonical variables for each set of functional items and PANSS items, with decreasing correlation between them, and we present the first two in each case here as these are the most important (see Table 2). Finally, we calculated the values of the first and second canonical variables for each scale on each patient, centred on the mean and standardised by the standard deviation. We fitted linear mixed effects models [4] with the first or second canonical variable for the respective functional scale as the response, and the corresponding canonical variable for PANSS as the covariate, accounting for differences between investigators (and therefore sites) by including investigator as a random effect [4]. In these models the regression coefficients correspond to the correlations between the canonical variables. MCMC sampling [1] was used to calculate the confidence intervals for these correlations. The R programming language [15] was used for all statistical analyses.
433
Table 1 The scales. PANSS Negative Subscale N1 Blunted affect N2 Emotional withdrawal N3 Poor rapport N4 Passive / apathetic social withdrawal N5 Difficulty in abstract thinking N6 lack of spontaneity and flow of conversation N7 Stereotyped thinking PsychoSocial Remission in Schizophrenia Scale (PSRS): (items scored from 1 = no difficulty to 7 = extreme difficulty) P1 Delusions P2 Conceptual disorganization P3 Hallucinatory behaviour G5 Mannerisms and posturing G9 Unusual thought content Q1 Impaired familial relations Q2 Impaired understanding and self-awareness Q3 Impaired energy Q4 Impaired interest in daily life F1 Impaired self-care F2 Impaired activism F3 Impaired responsibility medical treatment F4 Impaired use of community services Functional Remission of General Schizophrenia (FROGS): (5 subscales; items scored from 1 = low level of recovery to 5 = high level of recovery) Daily life (DL) 5 criteria Activities (AC) 3 criteria Relationships (R) 5 criteria Quality of adaptation (QA) 3 criteria Health and treatments (HT) 3 criteria Quality of Life Scale (QLS) Interpersonal relations (Household, Friends, Acquaintances, Social activity, Social network, Social initiative, Withdrawal, Socio-sexual) Instrumental role (Occupational role, Work functioning, Work level, Work satisfaction) Intrapsychic foundations (Sense of purpose, Motivation, Curiosity, Anhedonia, Aimless inactivity, Empathy, Emotional interaction) Commonplace objects and activities Global Assessment of Functioning (GAF): continuous scale from 1 = poor functioning to 90 = good functioning
Personal and Social Performance Scale (PSP): continuous scale from 1 = poor function to 100 = good function Subjective Well-Being under Neuroleptics (SWN): see reference [5] for more information about this self rated scale
3. Results All 11 investigators used the PANSS scale to assess their patients; however, each investigator utilized a different set of functional scales to assess the patients they recruited [5]. The summary statistics for each scale used in EGOFORS for each investigator’s patients are presented in an accompanying paper [5]. The GAF and PSRS scales were used by all investigators, though one investigator only used the PSRS on half of their sample of patients; QLS was used by five investigators, PSP by eight and SWN by seven, though no investigator used the SWN to assess all their patients. There are some differences between investigators in the mean and median scale scores. Fig. 1 shows a ‘‘scatterplot matrix’’ [18] of rating scale scores as an initial exploration of the relationships between negative symptoms and function, displaying the values for each individual of their total PANSS Negative score versus their total scores for each of the functional scales and SWN. The PANSS total scores have also been included for completeness. In the corresponding boxes on the upper diagonal panels of the figure, the correlations coefficients for each pair of variables have been displayed.
R. Hunter, S. Barry / European Psychiatry 27 (2012) 432–436
434
Table 2 Items and characteristics on which the canonical correlations analysis (CCA) puts most importance, along with the correlation between the negative symptoms and functional canonical variables with 95% confidence intervals (CI), for the first and second canonical variables. GAF and PSP each have one canonical variable only because there is only one item for each of these functional scales. See the methods section for the definition of a canonical variable. ‘‘% Variance explained’’ represents the amount of variation between individuals in the original scale measurements that is explained by its relationship with the other scale (the ‘‘redundancy index’’). PSRS 1st canonical variable 2nd canonical variable
PANSS PSRS PANSS PSRS
N1, N2, N3, N4, N5, N6, N7, work All items N7, duration of illness G5, Q3, Q4, F2
% Variance explained
Correlation (95% CI)
27.6% 34.4% 1.5% 2.2%
0.83 (0.76,0.90)
Correlation (95% CI)
FROGS 1st canonical variable 2nd canonical variable
PANSS FROGS PANSS FROGS
N1, N2, N3, N4, N5, N6, N7, work All items work AC (5 items)
26.1% 54.6% 2.7% 1.8%
2nd canonical variable
PANSS QLS PANSS QLS
N1, N2, N3, N4, N5, N6, N7, work, education All items N6, work Instrumental role (4 items)
26.5% 49.6% 2.1% 4.6%
PANSS GAF
N1, N2, N3, N4, N5, N6, N7, work GAF
22.7% 54.8%
PSP 1st canonical variable
0.57 (0.45,0.68)
0.84 (0.74,0.92) 0.47 (0.36,0.59)
Correlation (95% CI)
GAF 1st canonical variable
0.82 (0.74,0.89)
Correlation (95% CI)
QLS 1st canonical variable
0.33 (0.21,0.45)
0.78 (0.70,0.85)
Correlation (95% CI) PANSS PSP
N1, N2, N3, N4, N5, N6, N7, work PSP
There is a strong positive linear relationship between the PANSS Negative Subscale and PSRS, and similarly strong negative linear relationships between the PANSS Negative Subscale and FROGS, QLS, GAF and PSP. There are also strong linear relationships between some of the functional variables, particularly between
19.1% 53.8%
0.78 (0.67,0.86)
PSRS, FROGS and QLS, and between QLS, GAF and PSP, suggesting that these scales measure similar aspects of function. There is little relationship evident between PANSS Negative and SWN scores and this is borne out by the correlation coefficients between the PANSS Negative Subscale items and SWN items (not
Fig. 1. Scatter-plots of all of the functional scales (PSP, GAF, QLS, FROGS, PSRS), and SWN scale versus each other and the PANSS Negative Subscale and total scores, with correlation coefficients in the corresponding boxes on the upper diagonal panel. Read horizontally from one scale and vertically from another to find the relationship between them.
R. Hunter, S. Barry / European Psychiatry 27 (2012) 432–436
shown) as none are above 0.4. The correlation between total SWN and total PANSS is slightly stronger than 0.4, but total PANSS is not our interest here. There is also little evidence of any relationship between SWN and the other functional scales. We therefore have not considered the SWN scale further in this analysis. The results from the CCA and corresponding mixed effects models are presented in Table 2, identifying the functional and negative symptom scale items and other characteristics that were most important in each case. FROGS and QLS items were grouped into appropriate domains by taking the mean of the item scores contributing to them due to a large number of highly correlated items on these scales. The results are presented as correlations between the canonical variables, with 95% confidence intervals. For all functional scales, there are high correlations between the items on the PANSS Negative Subscale and the functional items, for at least the first canonical variable, and the correlations are statistically significantly different from zero in all cases. For all scales whether or not the patient was in work is also important in, at least for the first canonical variable. Duration of illness and age of onset generally did not contribute to the relationship with function. The results for each scale are summarized below. 3.1. PsychoSocial Remission in Schizophrenia Scale (PSRS) The correlation between the first canonical variables is strong, while for the second canonical variables, although statistically significant, is considerably weaker, suggesting that the items constituting the first canonical variable are the most important. Furthermore the amount of variance between subjects on the scales that is explained by the second canonical variable is very low. All of the PANSS Negative Subscale items and PSRS items contribute to the relationship between them for the first canonical variable, while only stereotyped thinking and duration of illness correlate with a small number of PSRS items in the second. 3.2. Functional Remission of General Schizophrenia (FROGS) The correlations between the first and second canonical variables are both reasonably strong, though once again little variance in the original variables is explained by the second canonical variable. All PANSS Negative Subscale items and FROGS items contribute to the first canonical variable, while only work and the FROGS item Activities (AC) appear in the second variable. 3.3. Quality of Life Scale (QLS) Each of the PANSS Negative Subscale items and QLS items appears in the first canonical variable, with a high correlation between the scales and, for QLS at least, a high amount of variance explained. Length of education is important in the first canonical variable along with the participant being in work. Being in work, lack of spontaneity and flow of conversation moderately correlate with the item ‘‘instrumental role’’ in QLS, though this explains little of the variance. 3.4. Global Assessment of Functioning scale; Personal and Social Performance scale (GAF and PSP) All of the PANSS Negative Subscale items are given importance in the canonical variables with GAF and PSP, with approximately half of the variance in each of these scales is explained by their relationship with negative symptoms. 4. Discussion The functional scales PSRS, FROGS, QLS, GAF and PSP have all been shown to have high inverse correlation with the PANSS
435
Negative Subscale, suggesting that the relationship between negative symptoms and functioning is strong. Thus, this analysis provides strong evidence that higher negative symptom scores are associated with poorer psychosocial functioning. This finding emphasizes the potential importance of negative symptoms as a treatment target given that the success of community based care systems are predicated to a large extent on adequate levels of functioning by patients. All of the items of the PANSS Negative Subscale appear important in the relationship with the different measures of function. All of the items in each of these functional scales contribute to the relationships with the PANSS Negative Subscale, suggesting that the negative symptoms have an important effect on them. As expected there is a strong relationship between being in work and psychosocial functioning. In this study the number of years in education has also been shown to have some impact on post morbid functioning, with improved functioning associated with longer educational experience. Perhaps surprisingly however, age of onset of psychosis and duration of illness did not appear to correlate with psychosocial functioning. These findings, if confirmed may have important therapeutic implications. Interestingly, neither individual items of the Subjective Wellbeing under Neuroleptics (SWN) scale, nor total SWN score were strongly related to the PANSS Negative Subscale items or total score, or to any of the other functional scales. Unlike the other measures reported, SWN is a patient-completed questionnaire with only a few of the items asking about psychosocial functioning per se [3]. Importantly SWN provides information from the patient’s perspective, and needs to be interpreted carefully, given that negative, positive and cognitive symptoms may profoundly influence the responses given by patients when completing scales such as SWN. There is also evidence using different instruments, that the views of patients and healthcare staff are often more divergent with respect to psychosocial functioning than other areas of need [7,8]. 4.1. Limitations The different scales for psychosocial functioning were measured on different sets of patients from a variety of settings, each of which involved different investigators. We have attempted to adjust for the effects of investigator and for other characteristics in our analysis, but there may be other unknown factors that we have not taken into account. Furthermore, the relationships between negative symptoms and the different functional scales may vary across different settings. We believe, however, that by considering all of the scales and all of the patients on whom they have each been measured and arriving at similar conclusions for each, we have minimized the risk of this causing any bias. 5. Conclusion The social and economic costs of mental illness are considerable; for example in Scotland this was recently estimated at 8% of GDP [16]. Despite the introduction of second-generation antipsychotics, drug treatment for schizophrenia still has a considerable way to go to address the challenges presented by the range of psychopathology involved [12,13]. It is highly unlikely that any single drug will be able to address all the symptom domains of schizophrenia. In particular the negative and cognitive symptoms are important areas of unmet need [9,13]. The current work shows that negative symptoms as assessed by PANSS are strongly related to psychosocial functioning as assessed by a number of different scales across eleven different European centres. Given the nature of such negative symptoms and their frequent occurrence in
436
R. Hunter, S. Barry / European Psychiatry 27 (2012) 432–436
schizophrenia, it is reasonable to assume that negative symptoms are an important causal contribution to reduced psychosocial functioning in schizophrenia. This has several important implications for the care of people with schizophrenia. Firstly, given that successful community care will require good or at least, adequate psychosocial functioning, improving negative symptomatology is an essential prerequisite. Secondly, physicians, psychologists, researchers and the pharmaceutical industry need to refocus on this area in order to develop effective treatments. There is some evidence that this is starting to occur (see www.ClinicalTrials.gov) although hampered by a still largely incomplete understanding of the pathogenesis and neurobiology of schizophrenia. Lastly, progress will also require that clinicians undertake more regular and systematic assessments of patients, in order to assess negative symptoms, cognition and psychosocial functioning. Disclosure of interest The authors declare that they have no conflicts of interest concerning this article. Acknowledgement The authors would like to thank the participants who took part in the study. References [1] Bates D, Maechler M. Linear mixed effects models using S4 classes; 2010. [2] Buchanan RW. Persistent negative symptoms in schizophrenia: an overview. Schizophr Bull 2007;33(4):1013–22. [3] de Haan L, Weisfelt M, Dingemans PMAJ, Linszen DH, Wouters L. Psychometric properties of the subjective well-being under neuroleptics scale and the subjective deficit syndrome scale. Psychopharmacology 2002;162(1):24–8. [4] Diggle PJ, Heagerty P, Liang K-Y, Zeger SL. Analysis of longitudinal data, Vol. 2. Clarendon: Oxford University Press Inc; 2002.
[5] Peuskens J, Gorwood P. On behalf of the EGOFORS initiative. How are we assessing functioning in schizophrenia? A need for a consensus approach. Eur Psychiatry 2011 (this issue). [6] Helldin L, Cavallaro R, Galderisi S. On behalf of the EGOFORS initiative. To measure functional outcome - what does a proper finding need to be based on? Eur Psychiatry 2011 (this issue). [7] Hunter R, Cameron R, Norrie J. Using Patient-Reported Outcomes in Schizophrenia: The Scottish Schizophrenia Outcomes Study. Psychiatr Serv 2009;60(2):240–5. [8] Hunter R, McLean J, Peck D, Pullen I, Greenfield A, Macri T, et al. The Scottish 700 Outcomes Study: a Comparative Evaluation of the Health of the Nation Outcome Scale (HoNOS), the Avon Mental Health Measure (AVON), and an Idiographic Scale. J Ment Health 2004;13(1):93–105. [9] Kirkpatrick B, Fenton WS, Carpenter WT, Marder SR. The NIMHMATRICS consensus statement on negative symptoms. Schizophr Bull 2006;32(2):214–9. [10] Kraepelin E. Dementia praecox and paraphrenia translated by Barclay RM. 1919. New York: RE Krieger; 1971. [11] Krzanowski WJ. Principles of multivariate analysis. New York: Oxford University Press; 1988. [12] Leucht S, Corves C, Arbter D, Engel RR, Li CB, Davis JM. Second-generation versus first-generation antipsychotic drugs for schizophrenia: a meta-analysis. Lancet 2009;373(9657):31–41. [13] NICE. Schizophrenia: Core interventions in the treatment and management of schizophrenia in primary and secondary care (update). National Clinical Practice Guideline Number 82, National Collaborating Centre for Mental Health. 2009, National Institute for Health and Clinical Excellence: London, UK. [14] Peyser H. What is recovery? A commentary. Psychiatr Serv 2001;52(4): 486–7. [15] Development Core Team R: a language and environment for statistical computing, version 2.11.1 (2010-05-31). R Foundation for Statistical Computing. Austria: Vienna; 2010. [16] SAMH, Scottish Association for Mental Health. What’s it worth? The social and economic costs of mental health problems in Scotland.. UK: Sainsbury Centre for Mental Health; 2007. [17] Thornicroft G, Tansella M, Becker T, Knapp M, Leese M, Schene A, et al. The personal impact of schizophrenia in Europe. Schizophr Res 2004;69(2– 3):125–32. [18] Venables WN, Ripley BD. Modern Applied Statistics with S, 4th ed., New York: Springer-Verlag; 2002. [19] Wiersma D, Wanderling J, Dragomirecka E, Ganev K, Harrison G, an der Heiden W, et al. Social disability in schizophrenia: its development and prediction over 15 years in incidence cohorts in six European centres. Psychol Med 2000;30(5):1155–67.
Original article
Negative symptoms and psychosocial functioning in schizophrenia: Neglected but important targets for treatment R. Hunter a,b,*, S. Barry c a
Psychiatric Research Institute for Neuroscience in Glasgow (PsyRING), West Medical Building, University of Glasgow, Glasgow G12 8QQ, United Kingdom Gartnavel Royal Hospital, Glasgow G12 0XH, United Kingdom c Robertson Centre for Biostatistics, Boyd Orr Building, University of Glasgow, Glasgow G12 8QQ, United Kingdom b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 8 September 2010 Received in revised form 3 February 2011 Accepted 3 February 2011 Available online 23 May 2011
Background. – Many patients with schizophrenia suffer from poor social functioning, with high levels of unemployment being one particular consequence. Negative symptoms tend to persist during periods of clinical stability and may have a detrimental effect on function. This paper aims to investigate the relationship between negative symptoms and ability to function. Methods. – The EGOFORS study measured negative symptoms in 295 schizophrenia patients in 11 European sites using the PANSS Negative Subscale and assessment scales for psychosocial function: Global Assessment of Functioning (GAF), Personal and Social Performance (PSP), Quality of Life Scale (QLS), Functional Remission of General Schizophrenia (FROGS), Psychosocial Remission in Schizophrenia (PSRS) and Subjective Wellbeing under Neuroleptics (SWN). The relationships between the PANSS Negative Subscale and the functional scales were investigated, adjusting for differences between study sites. Being in work, duration of illness, age of onset and number of years of education were also investigated for a relationship with function. Results. – There were strong, statistically significant correlations between PANSS Negative Subscale and all of the function scales (95% confidence intervals for the correlation coefficients: PSRS 0.77–0.91; FROGS 0.74–0.89; QLS 0.74–0.92; GAF 0.64–0.78; PSP 0.63–0.80) except the SWN. All of the functional scales except SWN were at least moderately related to one another. All of the items in each of the PANSS Negative Subscale and the function scales contributed to the relationships between them. Better functioning correlated strongly with participants being in work. Conclusion. – This study shows a strong and significant relationship between negative symptoms and psychosocial functioning. Given the impact of negative symptoms on psychosocial function, much more emphasis should be placed on developing effective treatments for negative symptoms, given that most patients with schizophrenia now live in community settings and require to function adequately to support their quality of life. ß 2011 Elsevier Masson SAS. All rights reserved.
Keywords: Schizophrenia Negative symptoms Psychosocial function Drug target
1. Introduction A significant proportion of people with schizophrenia continue to experience disabling symptoms (positive, negative, affective) and cognitive difficulties that affect personal, social and occupational functioning [13]. In one typical study based in six European countries, over 80% of adults with schizophrenia had persistent problems with social functioning [19]. Moreover, poor functioning in the first three years after diagnosis tended to predict long-term poor functioning. This is reflected in the high levels of unemployment among people with schizophrenia [7,17]. There has been renewed interest recently in the importance of ‘‘real world’’
* Corresponding author. Tel.: +44 141 2113550; fax: +44 141 3305175. E-mail address: [email protected] (R. Hunter). 0924-9338/$ – see front matter ß 2011 Elsevier Masson SAS. All rights reserved. doi:10.1016/j.eurpsy.2011.02.015
functioning and the development of treatment strategies for negative symptoms and cognitive impairments. In parallel with these developments in neuroscience, the ‘‘recovery model’’ has emphasized the importance of supporting individuals with schizophrenia to develop meaningful lives, with good functioning, despite continuing symptoms [14]. Negative symptoms were first described by Kraepelin in 1919 [10] as the ‘‘avolitional syndrome’’ and the term now refers to the absence or reduction in normal behaviours and functions (PANSS Negative Subscale symptoms are listed in Table 1). The classification of negative symptoms [2] is beyond the scope of this paper but persistent negative symptoms can be defined as symptoms that are either primary or secondary to schizophrenia, persist during periods of clinical stability, have not responded to treatment and which interfere with the ability to perform normal functions. Research is underway to develop new drugs that will improve
R. Hunter, S. Barry / European Psychiatry 27 (2012) 432–436
negative symptoms and the National Institutes of Health (NIH) website (www.ClinicalTrials.gov) lists many active trials in phases 1–3. Given this interest, the primary aim of the present study was to investigate the relationship between negative symptoms and psychosocial functioning in the EGOFORS cohort of people with schizophrenia. In particular, to explore the strength of the relationships between item and total scores of the PANSS Negative Subscale and items in a number of different scales used to measure psychosocial function. 2. Methods Recruitment of the 295 participants to the EGOFORS cohort in the 11 European study sites, and the selection and use of the assessment measures has been described in the accompanying paper [5]. The relationship between data from the functional scales (PSRS, FROGS, QLS, GAF and PSP see Table 1 for details of the scales used and individual scale items) and the PANSS Negative Subscale was examined (scales described in Gorwood et al. 2011 [5]), as was the relationship between the Subjective Wellbeing under Neuroleptics Scale (SWN) and the PANSS Negative Subscale. The SWN is not a psychosocial functioning scale per se but provides information on the subjective experience of patients during treatment with antipsychotic medication [3]. Data from the UPSA scale was not included given its more limited use in the cohort and is discussed elsewhere in the accompanying EGOFORS subanalysis [6]. As the PSRS items N1, N4 and N6 are shared with the PANSS scale, we have therefore considered the PSRS scale without these, in order to avoid artefactual significance. We also removed one patient from the dataset because they had missing data for both the PANSS N3 and N7 items. Correlations were calculated using the Spearman correlation coefficient because of some minor deviations from normality of the PANSS negative and total scores. The relationships between the PANSS negative subscale and the functional scales were assessed using canonical correlation analysis (CCA). This method finds the linear combinations of the items in the PANSS Negative Subscale and each functional scale that have the highest correlation with one another [11]. The item combinations are presented as so-called ‘‘canonical variables’’, in a similar way to factors in factor analysis, thereby allowing us to identify the items in each functional scale that contribute most to the relationship with the items in the PANSS Negative Subscale, and vice versa. Items with a ‘‘loading’’ (the correlation with the original variable) of more than 0.5 on a canonical variable were considered ‘‘important.’’ To account for age, duration of illness, years of education and whether the participant was in work, these characteristics were included in the canonical variables for the PANSS items. Like factor analysis, CCA finds several canonical variables for each set of functional items and PANSS items, with decreasing correlation between them, and we present the first two in each case here as these are the most important (see Table 2). Finally, we calculated the values of the first and second canonical variables for each scale on each patient, centred on the mean and standardised by the standard deviation. We fitted linear mixed effects models [4] with the first or second canonical variable for the respective functional scale as the response, and the corresponding canonical variable for PANSS as the covariate, accounting for differences between investigators (and therefore sites) by including investigator as a random effect [4]. In these models the regression coefficients correspond to the correlations between the canonical variables. MCMC sampling [1] was used to calculate the confidence intervals for these correlations. The R programming language [15] was used for all statistical analyses.
433
Table 1 The scales. PANSS Negative Subscale N1 Blunted affect N2 Emotional withdrawal N3 Poor rapport N4 Passive / apathetic social withdrawal N5 Difficulty in abstract thinking N6 lack of spontaneity and flow of conversation N7 Stereotyped thinking PsychoSocial Remission in Schizophrenia Scale (PSRS): (items scored from 1 = no difficulty to 7 = extreme difficulty) P1 Delusions P2 Conceptual disorganization P3 Hallucinatory behaviour G5 Mannerisms and posturing G9 Unusual thought content Q1 Impaired familial relations Q2 Impaired understanding and self-awareness Q3 Impaired energy Q4 Impaired interest in daily life F1 Impaired self-care F2 Impaired activism F3 Impaired responsibility medical treatment F4 Impaired use of community services Functional Remission of General Schizophrenia (FROGS): (5 subscales; items scored from 1 = low level of recovery to 5 = high level of recovery) Daily life (DL) 5 criteria Activities (AC) 3 criteria Relationships (R) 5 criteria Quality of adaptation (QA) 3 criteria Health and treatments (HT) 3 criteria Quality of Life Scale (QLS) Interpersonal relations (Household, Friends, Acquaintances, Social activity, Social network, Social initiative, Withdrawal, Socio-sexual) Instrumental role (Occupational role, Work functioning, Work level, Work satisfaction) Intrapsychic foundations (Sense of purpose, Motivation, Curiosity, Anhedonia, Aimless inactivity, Empathy, Emotional interaction) Commonplace objects and activities Global Assessment of Functioning (GAF): continuous scale from 1 = poor functioning to 90 = good functioning
Personal and Social Performance Scale (PSP): continuous scale from 1 = poor function to 100 = good function Subjective Well-Being under Neuroleptics (SWN): see reference [5] for more information about this self rated scale
3. Results All 11 investigators used the PANSS scale to assess their patients; however, each investigator utilized a different set of functional scales to assess the patients they recruited [5]. The summary statistics for each scale used in EGOFORS for each investigator’s patients are presented in an accompanying paper [5]. The GAF and PSRS scales were used by all investigators, though one investigator only used the PSRS on half of their sample of patients; QLS was used by five investigators, PSP by eight and SWN by seven, though no investigator used the SWN to assess all their patients. There are some differences between investigators in the mean and median scale scores. Fig. 1 shows a ‘‘scatterplot matrix’’ [18] of rating scale scores as an initial exploration of the relationships between negative symptoms and function, displaying the values for each individual of their total PANSS Negative score versus their total scores for each of the functional scales and SWN. The PANSS total scores have also been included for completeness. In the corresponding boxes on the upper diagonal panels of the figure, the correlations coefficients for each pair of variables have been displayed.
R. Hunter, S. Barry / European Psychiatry 27 (2012) 432–436
434
Table 2 Items and characteristics on which the canonical correlations analysis (CCA) puts most importance, along with the correlation between the negative symptoms and functional canonical variables with 95% confidence intervals (CI), for the first and second canonical variables. GAF and PSP each have one canonical variable only because there is only one item for each of these functional scales. See the methods section for the definition of a canonical variable. ‘‘% Variance explained’’ represents the amount of variation between individuals in the original scale measurements that is explained by its relationship with the other scale (the ‘‘redundancy index’’). PSRS 1st canonical variable 2nd canonical variable
PANSS PSRS PANSS PSRS
N1, N2, N3, N4, N5, N6, N7, work All items N7, duration of illness G5, Q3, Q4, F2
% Variance explained
Correlation (95% CI)
27.6% 34.4% 1.5% 2.2%
0.83 (0.76,0.90)
Correlation (95% CI)
FROGS 1st canonical variable 2nd canonical variable
PANSS FROGS PANSS FROGS
N1, N2, N3, N4, N5, N6, N7, work All items work AC (5 items)
26.1% 54.6% 2.7% 1.8%
2nd canonical variable
PANSS QLS PANSS QLS
N1, N2, N3, N4, N5, N6, N7, work, education All items N6, work Instrumental role (4 items)
26.5% 49.6% 2.1% 4.6%
PANSS GAF
N1, N2, N3, N4, N5, N6, N7, work GAF
22.7% 54.8%
PSP 1st canonical variable
0.57 (0.45,0.68)
0.84 (0.74,0.92) 0.47 (0.36,0.59)
Correlation (95% CI)
GAF 1st canonical variable
0.82 (0.74,0.89)
Correlation (95% CI)
QLS 1st canonical variable
0.33 (0.21,0.45)
0.78 (0.70,0.85)
Correlation (95% CI) PANSS PSP
N1, N2, N3, N4, N5, N6, N7, work PSP
There is a strong positive linear relationship between the PANSS Negative Subscale and PSRS, and similarly strong negative linear relationships between the PANSS Negative Subscale and FROGS, QLS, GAF and PSP. There are also strong linear relationships between some of the functional variables, particularly between
19.1% 53.8%
0.78 (0.67,0.86)
PSRS, FROGS and QLS, and between QLS, GAF and PSP, suggesting that these scales measure similar aspects of function. There is little relationship evident between PANSS Negative and SWN scores and this is borne out by the correlation coefficients between the PANSS Negative Subscale items and SWN items (not
Fig. 1. Scatter-plots of all of the functional scales (PSP, GAF, QLS, FROGS, PSRS), and SWN scale versus each other and the PANSS Negative Subscale and total scores, with correlation coefficients in the corresponding boxes on the upper diagonal panel. Read horizontally from one scale and vertically from another to find the relationship between them.
R. Hunter, S. Barry / European Psychiatry 27 (2012) 432–436
shown) as none are above 0.4. The correlation between total SWN and total PANSS is slightly stronger than 0.4, but total PANSS is not our interest here. There is also little evidence of any relationship between SWN and the other functional scales. We therefore have not considered the SWN scale further in this analysis. The results from the CCA and corresponding mixed effects models are presented in Table 2, identifying the functional and negative symptom scale items and other characteristics that were most important in each case. FROGS and QLS items were grouped into appropriate domains by taking the mean of the item scores contributing to them due to a large number of highly correlated items on these scales. The results are presented as correlations between the canonical variables, with 95% confidence intervals. For all functional scales, there are high correlations between the items on the PANSS Negative Subscale and the functional items, for at least the first canonical variable, and the correlations are statistically significantly different from zero in all cases. For all scales whether or not the patient was in work is also important in, at least for the first canonical variable. Duration of illness and age of onset generally did not contribute to the relationship with function. The results for each scale are summarized below. 3.1. PsychoSocial Remission in Schizophrenia Scale (PSRS) The correlation between the first canonical variables is strong, while for the second canonical variables, although statistically significant, is considerably weaker, suggesting that the items constituting the first canonical variable are the most important. Furthermore the amount of variance between subjects on the scales that is explained by the second canonical variable is very low. All of the PANSS Negative Subscale items and PSRS items contribute to the relationship between them for the first canonical variable, while only stereotyped thinking and duration of illness correlate with a small number of PSRS items in the second. 3.2. Functional Remission of General Schizophrenia (FROGS) The correlations between the first and second canonical variables are both reasonably strong, though once again little variance in the original variables is explained by the second canonical variable. All PANSS Negative Subscale items and FROGS items contribute to the first canonical variable, while only work and the FROGS item Activities (AC) appear in the second variable. 3.3. Quality of Life Scale (QLS) Each of the PANSS Negative Subscale items and QLS items appears in the first canonical variable, with a high correlation between the scales and, for QLS at least, a high amount of variance explained. Length of education is important in the first canonical variable along with the participant being in work. Being in work, lack of spontaneity and flow of conversation moderately correlate with the item ‘‘instrumental role’’ in QLS, though this explains little of the variance. 3.4. Global Assessment of Functioning scale; Personal and Social Performance scale (GAF and PSP) All of the PANSS Negative Subscale items are given importance in the canonical variables with GAF and PSP, with approximately half of the variance in each of these scales is explained by their relationship with negative symptoms. 4. Discussion The functional scales PSRS, FROGS, QLS, GAF and PSP have all been shown to have high inverse correlation with the PANSS
435
Negative Subscale, suggesting that the relationship between negative symptoms and functioning is strong. Thus, this analysis provides strong evidence that higher negative symptom scores are associated with poorer psychosocial functioning. This finding emphasizes the potential importance of negative symptoms as a treatment target given that the success of community based care systems are predicated to a large extent on adequate levels of functioning by patients. All of the items of the PANSS Negative Subscale appear important in the relationship with the different measures of function. All of the items in each of these functional scales contribute to the relationships with the PANSS Negative Subscale, suggesting that the negative symptoms have an important effect on them. As expected there is a strong relationship between being in work and psychosocial functioning. In this study the number of years in education has also been shown to have some impact on post morbid functioning, with improved functioning associated with longer educational experience. Perhaps surprisingly however, age of onset of psychosis and duration of illness did not appear to correlate with psychosocial functioning. These findings, if confirmed may have important therapeutic implications. Interestingly, neither individual items of the Subjective Wellbeing under Neuroleptics (SWN) scale, nor total SWN score were strongly related to the PANSS Negative Subscale items or total score, or to any of the other functional scales. Unlike the other measures reported, SWN is a patient-completed questionnaire with only a few of the items asking about psychosocial functioning per se [3]. Importantly SWN provides information from the patient’s perspective, and needs to be interpreted carefully, given that negative, positive and cognitive symptoms may profoundly influence the responses given by patients when completing scales such as SWN. There is also evidence using different instruments, that the views of patients and healthcare staff are often more divergent with respect to psychosocial functioning than other areas of need [7,8]. 4.1. Limitations The different scales for psychosocial functioning were measured on different sets of patients from a variety of settings, each of which involved different investigators. We have attempted to adjust for the effects of investigator and for other characteristics in our analysis, but there may be other unknown factors that we have not taken into account. Furthermore, the relationships between negative symptoms and the different functional scales may vary across different settings. We believe, however, that by considering all of the scales and all of the patients on whom they have each been measured and arriving at similar conclusions for each, we have minimized the risk of this causing any bias. 5. Conclusion The social and economic costs of mental illness are considerable; for example in Scotland this was recently estimated at 8% of GDP [16]. Despite the introduction of second-generation antipsychotics, drug treatment for schizophrenia still has a considerable way to go to address the challenges presented by the range of psychopathology involved [12,13]. It is highly unlikely that any single drug will be able to address all the symptom domains of schizophrenia. In particular the negative and cognitive symptoms are important areas of unmet need [9,13]. The current work shows that negative symptoms as assessed by PANSS are strongly related to psychosocial functioning as assessed by a number of different scales across eleven different European centres. Given the nature of such negative symptoms and their frequent occurrence in
436
R. Hunter, S. Barry / European Psychiatry 27 (2012) 432–436
schizophrenia, it is reasonable to assume that negative symptoms are an important causal contribution to reduced psychosocial functioning in schizophrenia. This has several important implications for the care of people with schizophrenia. Firstly, given that successful community care will require good or at least, adequate psychosocial functioning, improving negative symptomatology is an essential prerequisite. Secondly, physicians, psychologists, researchers and the pharmaceutical industry need to refocus on this area in order to develop effective treatments. There is some evidence that this is starting to occur (see www.ClinicalTrials.gov) although hampered by a still largely incomplete understanding of the pathogenesis and neurobiology of schizophrenia. Lastly, progress will also require that clinicians undertake more regular and systematic assessments of patients, in order to assess negative symptoms, cognition and psychosocial functioning. Disclosure of interest The authors declare that they have no conflicts of interest concerning this article. Acknowledgement The authors would like to thank the participants who took part in the study. References [1] Bates D, Maechler M. Linear mixed effects models using S4 classes; 2010. [2] Buchanan RW. Persistent negative symptoms in schizophrenia: an overview. Schizophr Bull 2007;33(4):1013–22. [3] de Haan L, Weisfelt M, Dingemans PMAJ, Linszen DH, Wouters L. Psychometric properties of the subjective well-being under neuroleptics scale and the subjective deficit syndrome scale. Psychopharmacology 2002;162(1):24–8. [4] Diggle PJ, Heagerty P, Liang K-Y, Zeger SL. Analysis of longitudinal data, Vol. 2. Clarendon: Oxford University Press Inc; 2002.
[5] Peuskens J, Gorwood P. On behalf of the EGOFORS initiative. How are we assessing functioning in schizophrenia? A need for a consensus approach. Eur Psychiatry 2011 (this issue). [6] Helldin L, Cavallaro R, Galderisi S. On behalf of the EGOFORS initiative. To measure functional outcome - what does a proper finding need to be based on? Eur Psychiatry 2011 (this issue). [7] Hunter R, Cameron R, Norrie J. Using Patient-Reported Outcomes in Schizophrenia: The Scottish Schizophrenia Outcomes Study. Psychiatr Serv 2009;60(2):240–5. [8] Hunter R, McLean J, Peck D, Pullen I, Greenfield A, Macri T, et al. The Scottish 700 Outcomes Study: a Comparative Evaluation of the Health of the Nation Outcome Scale (HoNOS), the Avon Mental Health Measure (AVON), and an Idiographic Scale. J Ment Health 2004;13(1):93–105. [9] Kirkpatrick B, Fenton WS, Carpenter WT, Marder SR. The NIMHMATRICS consensus statement on negative symptoms. Schizophr Bull 2006;32(2):214–9. [10] Kraepelin E. Dementia praecox and paraphrenia translated by Barclay RM. 1919. New York: RE Krieger; 1971. [11] Krzanowski WJ. Principles of multivariate analysis. New York: Oxford University Press; 1988. [12] Leucht S, Corves C, Arbter D, Engel RR, Li CB, Davis JM. Second-generation versus first-generation antipsychotic drugs for schizophrenia: a meta-analysis. Lancet 2009;373(9657):31–41. [13] NICE. Schizophrenia: Core interventions in the treatment and management of schizophrenia in primary and secondary care (update). National Clinical Practice Guideline Number 82, National Collaborating Centre for Mental Health. 2009, National Institute for Health and Clinical Excellence: London, UK. [14] Peyser H. What is recovery? A commentary. Psychiatr Serv 2001;52(4): 486–7. [15] Development Core Team R: a language and environment for statistical computing, version 2.11.1 (2010-05-31). R Foundation for Statistical Computing. Austria: Vienna; 2010. [16] SAMH, Scottish Association for Mental Health. What’s it worth? The social and economic costs of mental health problems in Scotland.. UK: Sainsbury Centre for Mental Health; 2007. [17] Thornicroft G, Tansella M, Becker T, Knapp M, Leese M, Schene A, et al. The personal impact of schizophrenia in Europe. Schizophr Res 2004;69(2– 3):125–32. [18] Venables WN, Ripley BD. Modern Applied Statistics with S, 4th ed., New York: Springer-Verlag; 2002. [19] Wiersma D, Wanderling J, Dragomirecka E, Ganev K, Harrison G, an der Heiden W, et al. Social disability in schizophrenia: its development and prediction over 15 years in incidence cohorts in six European centres. Psychol Med 2000;30(5):1155–67.