Can we combine symptom scales for collaborative research projects?

  • Published on

  • View

  • Download

Embed Size (px)


  • lla


    Collaborative research

    jectourhizo, Scnd the Brief Psychiatric Rating Scale (BPRS) being among the most commonlyh emetudi

    comparison of the three scales there were considerable clinical differences between them. The paper

    o sympgs leadmentaars toelianceects isout th

    Negative Syndrome Scale (PANSS) (Andreasen, 1984; Andreasen,1989; Kay et al., 1987; Overall & Gorham, 1962). The aim of thispaper is to investigate the challenges of combining these three

    ative symptoms (Kirkpatrick et al., 2006), with sound psychometricproperties, such as interrater reliability, internal consistency, andpredictive validity (Andreasen, 1990a; Andreasen, 1990b;Andreasen, 1982). The scale is divided into 19 item ratings, and 5global factor ratings. Like the SAPS, each item is scored on a sixpoint scale (0e5). The SANS global total is the sum of the ve globalratings, while the SANS composite total is the sum of the nineteenitem ratings. Andreasen has reported that summary scores based

    * Corresponding author. DETECT Services, Avila House, Block 5 Blackrock Busi-ness Park, Co. Dublin, Ireland. Tel.: 353 01 2791700; fax: 353 01 2791799.

    Contents lists available at

    Journal of Psych


    Journal of Psychiatric Research 46 (2012) 233e238E-mail address: (J.P. Lyne).collection. This has the potential to increase statistical power foranswering research hypotheses for which sample size is an issue,for example when investigating long-term outcomes of psychosisor where large sample sizes are necessary for genetic studies.

    If different scales have been used in research projects, somemethod for combining scales is necessary to allow collaboration toproceed. Three of the most commonly used scales for psychoticsymptoms include: (i) Scale for Assessment of Positive Symptoms(SAPS) and the Scale for Assessment of Negative Symptoms (SANS),(ii) Brief Psychiatric Rating Scale (BPRS), and (iii) Positive and

    test-retest reliability (Moscarelli et al., 1987; Norman et al., 1996;Andreasen & Flaum, 1991), and is divided into 31 item ratings andfour global factor ratings. Each item is scored on a six point scale(0e5). TheSAPSglobal total is thesumof the fourglobal ratings,whilethe SAPS composite total is the sum of the thirty one item ratings.

    2.2. SANS

    The SANS is recognised as a valuable tool in assessment of neg-1. Introduction

    There has been much research intthe last few decades, with new ndinand management of patients with2010). Large sample studies take yehuge nancial input with heavy rCollaboration between research projobtain a large study sample, with0022-3956/$ e see front matter 2011 Elsevier Ltd.doi:10.1016/j.jpsychires.2011.10.002toms of psychosis overing to better treatmentl illness (Tandon et al.,complete, and requireon research funding.

    a more efcient way toe need for new data

    scales for collaborative research purposes, and in the discussionwewill explore possible methods for achieving this.

    2. The rating scales

    2.1. SAPS

    The SAPS is a widely used tool for measurement of positivesymptoms. The scale has been shown to have good interrater andPsychosis recently developed standardised remission criteria for schizophrenia. 2011 Elsevier Ltd. All rights reserved.Remission criteria discusses potential methods for combining the scales in collaborative research, including use of theReview

    Can we combine symptom scales for co

    John P. Lyne a,b,*, Anthony Kinsella a, Brian ODonogaDETECT Services, Avila House, Block 5 Blackrock Business Park, Co. Dublin, IrelandbCollege of Life Sciences, University College Dublin, Beleld, Dublin 4, Ireland

    a r t i c l e i n f o

    Article history:Received 21 July 2011Received in revised form4 September 2011Accepted 10 October 2011


    a b s t r a c t

    Collaborative research prootherwise require huge respsychotic symptoms in scPositive Symptoms (SAPS)Symptom Scale (PANSS), aused. High quality researcmerge study efforts, somecorrelations in published s

    journal homepage: www.eAll rights reserved.fforts have used these three scales in different projects, and in order toans of combining data from these scales may be necessary. We reviewedes for these three scales, nding them to be highly correlated, however one a,b

    s have the potential to answer important research questions, which mayces, funding, and time to complete. There are several scales for measuringphrenia and other psychotic disorders, with the Scale for Assessment ofale for Assessment of Negative Symptoms (SANS), Positive and Negativeborative research projects?

    SciVerse ScienceDirect

    iatric Research

  • on global ratings is probably a more sensitive index than thecomposite score for both the SAPS and SANS (Andreasen, 1982).

    2.3. BPRS

    Overall andGorham introduced the Brief Psychiatric Rating Scale(BPRS) in 1962, and it is still a commonly used measure of symp-tomatology today (Overall & Gorham,1962). The initial 16 item scalewas subsequently expanded to 18 items (Overall & Klett,1972), withmore recent versions expanded further to 24 items (Lukoff et al.,1986). BPRS items are rated on a seven point scale (1e7), with theBPRS total score calculated by summing all items. The BPRS has beenshown to be a sensitive and specic measure, with good interraterreliability (Ventura et al., 1993; Roncone et al., 1999).

    The BPRS can be divided further into subscales, including theBPRS-PS, and the BPRS-WR. The BPRS-PS is used to measure posi-tive psychotic symptoms, comprising the items hallucinatorybehaviour, unusual thought content, suspiciousness, and concep-tual disorganisation (Ventura et al., 2000; Harris et al., 2005). TheBPRS-WR comprises a withdrawal-retardation factor, including theblunted affect, emotional withdrawal and motor retardation items,which is used as a measure of negative symptomatology (Welhamet al., 1999; Eckert et al., 1996). The sensitivity of the BPRS-WR tonegative symptom change has been questioned by Eckert et al. whoshowed that an additional measure of negative symptoms, such asthe SANS, may increase the ability to detect changes in negativesymptoms with substantially fewer subjects than the BPRS-WRalone (Eckert et al., 1996).

    2.4. PANSS

    PANSS consists of 30 items rated on a seven point scale (1e7).Seven of the items contribute to an overall positive syndromescore, the PANSS Positive Scale (PANSS-PS), while another sevenitems contribute to a negative syndrome score, the PANSS Nega-tive Scale (PANSS-NS); 16 items contribute to a measure of generalpsychopathology. The PANSS has been recommended as appro-priate for use in positive and negative symptom evaluation (Alphs,2006), and has been validated by van der Gaag et al. (2006a;2006b).

    3. Comparison of items present in the three scales

    3.1. Positive symptom item comparisons

    Table 1 illustrates the positive symptom items for the SAPS,PANSS-PS, and BPRS. For ease of comparison the subscale headingsof the SAPS have been used to classify items for each of the scales.There are items which are similar across the three scales for thedelusions, hallucinations, and positive formal thought disorderfactors. The bizarre behaviour subscale of the SAPS has no clearlycomparable items with the PANSS-PS, however Andreasen et al.have positioned the mannerisms/posturing item, present in thegeneral PANSS scale, alongside the SAPS bizarre behaviour subscalein their remission criteria paper (mannerisms/posturing item hasbeen excluded from the PANSS-PS section of Table 1 as it is part ofthe general PANSS scale) (Andreasen et al., 2005). The BPRS alsocontains the mannerisms/posturing item, and the extended versionBPRS includes a bizarre behaviour item. Some items contained inthe PANSS-PS, including grandiosity, hostility, and excitement, are


    Mannerisms and posturing

    J.P. Lyne et al. / Journal of Psychiatric Research 46 (2012) 233e238234The PANSS was constructed by combining items from the BPRS,and the Psychopathology Rating Scale (Singh & Kay, 1975). The

    Table 1Comparison of symptoms in the SAPS, PANSS-Pos, and the BPRS-PS.


    Hallucinations Auditory hallucinationsVoices commentingVoices conversingSomatic or tactile hallucinationsOlfactory hallucinationsVisual hallucinations

    Delusions Persecutory delusionsDelusions of jealousyDelusions of sin or guiltGrandiose delusionsReligious delusionsSomatic delusionsIdeas and delusions of referenceDelusions of being controlledDelusions of mind readingThought broadcastingThought insertionThought withdrawal

    Bizarre Behaviour Clothing and appearanceSocial and sexual behaviourAggressive and agitated behaviourRepetitive or stereotyped behaviou

    Positive Formal Thought Disorder Derailment (loose associations)TangentialityIncoherence (word salad)IllogicalityCircumstantialityPressure of speechDistractible speechClanging

    Other Inappropriate affect

    a BPRS-PS items.b Also contained in the PANSS general subscale.Conceptual disorganisation Conceptual Disorganisationa


    ExcitementHostilityalso present in the BPRS, however these items are not included inthe BPRS-PS. Likewise, a BPRS-PS item, unusual thought content, ispresent as a general subscale item on the PANSS.


    Hallucinatory behaviour Hallucinatory behavioura


    Unusual thought contenta,b



    Bizarre behavioura

  • 3.2. Negative symptom item comparisons

    Table 2 illustrates the negative symptom items for the SANS,PANSS-NS, and BPRS. Again for ease of comparison the subscaleheadings of the SANS have been used to classify items for each ofthe scales. There are fewer items on the PANSS-NS and BPRS-WR,however the content of some of these items overlap with morethan one SANS item. Some of the general PANSS items, alsooverlap with other negative symptom items on the SANS andBPRS. For example, the general PANSS item, disturbance of voli-tion could be viewed as overlapping with the grooming andhygiene item and the recreational interests and activities item ofthe SANS. The motor retardation item is considered part of theBPRS-WR subscale, but is included in the general subscale in thePANSS. The motor retardation item is not directly comparablewith any single item of the SANS, but could be considered tooverlap with a number of the SANS items such as decreasedspontaneous movements, paucity of expressive gestures,increased latency of response, and physical anergia. The generalPANSS item poor attention may overlap with social inattentive-ness in the SANS, however there have been doubts expressed as towhether attention items should be considered under the negative

    4. Review of studies comparing the intercorrelation of thethree scales

    4.1. Correlation coefcients

    Correlation coefcients (r) are a way of summarising the inter-relationship between instruments by calculating the correlationbetween two scale total scores. Redundancy analysis is also used forstudying the relationship between scales, by maximising theexplained variance (or redundancy) between them (Welham et al.,1999; Czobor et al., 1991). We performed an electronic search for allpublished studies including at least two of the terms SAPS, SANS,PANSS, and BPRS in the title for the MEDLINE, Embase, CINAHL,and PsycINFO databases. A similar search was performed for theunabbreviated names of these scales. All studies which reporteda correlation gure between two or more of these scales is reportedhere (Table 4). The references of each article selected were alsoreviewed in search of further studies.

    4.2. SAPS/SANS and PANSS

    Norman et al. examined the inter-correlations between total

    J.P. Lyne et al. / Journal of Psychiatric Research 46 (2012) 233e238 235symptom construct (Toomey et al., 1998; Palacios-Araus et al.,1995; Miller et al., 1993).

    3.3. Comparison of other items on the PANSS and BPRS

    The general scale of the PANSS and other items of the BPRS arepresented in Table 3. Some of these items may represent othersymptoms/factors in schizophrenia, which are not adequatelyrepresented on the positive or negative scales. As the PANSS wasderived from the BPRS many of these items overlap, however itshould be noted that the item denitions were changed when thePANSS was created, and Bell et al. showed that several of theseitems, which are identically named on both scales, were modiedsufciently to make them not necessarily interchangeable (Bellet al., 1992). When comparison between shared items wasmeasured with weighted kappas, the kappa coefcient was in theexcellent range for only three items (blunted affect, hallucinations,and grandiosity), none of which are contained in the general scaleof the PANSS.

    Table 2Comparison of symptoms in the SANS, PANSS-Neg, and BPRS-WR.


    Affective Flattening or Blunting Unchanging facial expressionDecreased spontaneous movementsPaucity of expressive gesturesPoor eye contactAffective non-responsivityLack of vocal inections

    Alogia Poverty of speechPoverty of content of speechBlockingIncreased latency of response

    Avolition-Apathy Grooming and hygieneImpersistence at work or schoolPhysical anergia

    Anhedonia-Asociality Recreational interests and activitiesSexual interest and activityAbility to feel intimacy and closenessRelationships with friends and peers

    Attention/Other Social InattentivenessInattentiveness during mentalstate testinga BPRS-WR items.scores on SAPS/SANS and PANSS correlating: (i) Average score ofthe same rater for both scales, and (ii) Average score using differentraters for each instrument (Norman et al., 1996). Using the ratingsfor each tool by the same rater, high and signicant correlationbetween the SAPS global scores and PANSS positive scale (0.91,N 85) were found, while correlations across different raters werealso highly correlated (0.81). For negative symptoms, signicantcorrelation was found between SANS global scores and PANSSnegative scale when average scores by the same rater were used(0.88), and only moderate correlation was found when measuredby different raters (0.58). The interrater reliabilities for itemsassessing positive symptoms tended to be higher than that fornegative symptoms, and the greater number of items in the SAPS/SANS were noted to provide more detailed assessment of specicsymptoms where necessary.

    Peralta et al. also found high correlation for the SAPS and PANSSpositive scale (0.70), as well as for SANS and PANSS negative scale(0.80; N 100), however the negative symptom scale correlationwas higher than that for the positive symptom scales (Peralta et al.,


    Blunted affect Blunted affecta

    Lack of spontaneity and ow ofconverstation

    Passive/apathetic social withdrawal Self-neglect

    Emotional withdrawalDifculty in abstract thinkingStereotyped thinkingPoor rapport

    Emotional withdrawala

    Motor retardationa

  • Table 3Comparison of the PANSS general scale items and other BPRS items.

    PANSS e general scale items BPRS e other items

    Somatic concern Somatic concernAnxiety AnxietyGuilt feelings GuiltTension TensionDepression DepressionUncooperativeness UncooperativenessDisorientation Disorientati...


View more >