verbal fluency in multiple sclerosis

Please download to get full document.

View again

of 9
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Information Report
Category:

Documents

Published:

Views: 4 | Pages: 9

Extension: PDF | Download: 0

Share
Related documents
Description
hjhg
Transcript
  Neuropsychologia 44 (2006) 1166–1174 Verbal fluency deficits in multiple sclerosis Julie D. Henry a , ∗ , William W. Beatty b a School of Psychology, University of New South Wales, Sydney, NSW, Australia b University of Oklahoma Health Sciences Center, Department of Psychiatry and Behavioral Sciences, OK, USA Received 25 August 2005; received in revised form 5 October 2005; accepted 5 October 2005Available online 15 November 2005 Abstract A quantitative review of 35 studies with 3673 participants was conducted to estimate and compare the magnitude of deficits upon tests of phonemic and semantic fluency for participants with multiple sclerosis (MS) relative to healthy controls. Participants with MS were substantiallybut similarly impaired on tests of phonemic and semantic fluency. These deficits were larger than deficits on measures of verbal intelligence,confrontation naming and another widely used measure of executive functioning, the Wisconsin Card Sorting Test, but were of a comparableor smaller magnitude relative to deficits on the oral version of the Symbol Digit Modalities Test (SDMT). This is consistent with other researchsuggestingthatmeasuresofverbalfluencyandtheSDMTmaybeamongstthemostsensitiveneuropsychologicalmeasurestocognitiveimpairmentin MS. Increased neurological disability and a chronic progressive (as opposed to a relapsing remitting) disease course were associated with largerdeficits on tests of phonemic and semantic fluency. However, it is suggested that this latter finding is attributable to the distinct clinical features of chronic progressive and relapsing remitting sub-types. Thus, patients who follow a chronic progressive course tend to be older, have an increasedduration of illness and experience greater neurological disability. Once these variables were controlled for, differences between the two sub-typeswere substantially attenuated.© 2005 Elsevier Ltd. All rights reserved. Keywords:  Meta-analysis; Fluency; Disease course 1. Introduction In addition to motor abnormalities, cognitive impairment isa common feature of multiple sclerosis (MS). Neuropatholog-ically, MS is associated with multiple focal areas of axonaldemyelination,anddiffusewhitematterpathologyisbelievedtobe particularly associated with executive dysfunction (Stuss &Gow,1992).However,whilstsomestudieshavefoundthatexec-utivedeficitsoccurwithconsiderablefrequencyinMS(Benedictet al., 2002; Marie & Defer, 2001), others have found little evi- dence that executive dysfunction particularly characterises thedisorder (Bryant, Chiaravalloti, & DeLuca, 2004; Chiaravalloti & De Luca, 2002).Tests of verbal fluency have consistently been found to bemore sensitive to impairment in MS relative to other measuresofexecutivefunctioning.RosserandHodges(1994)haveargued that identical executive processes are involved in the initiation ∗ Corresponding author. Tel.: +61 2 9385 3936; fax: +61 2 9385 3641.  E-mail address:  julie.henry@unsw.edu.au (J.D. Henry). and monitoring of both of these tasks, but that semantic flu-ency is relatively more dependent on the integrity of semanticmemory(seealsoHenry&Crawford,2004).Assessingtherela- tiveprominenceofdeficitsonphonemicandsemanticfluencyisthereforeimportant,andbearsonwhetherMSisassociatedwithexecutive function impairments and/or semantic memory dys-function. Whilst there might be more than one reason for equalimpairmentinphonemicandsemanticfluencytasks,apatternof comparableimpairmentwouldbeconsistentwiththepossibilitythatthedeficitsreflectexecutivedysfunction.Incontrast,greaterimpairment on measures of semantic fluency may be indicativeof semantic memory dysfunction. However, whilst some stud-ies have reported comparable deficits on measures of phonemicand semantic fluency (Beatty, 2002; Parry, Scott, Palace, Smith, & Matthews, 2003), others have found phonemic fluency to bemore affected by the disorder (Fischer, unpublished; Nocentiniet al., 2001). Greater impairment on semantic fluency has alsobeen reported (Foong et al., 1997; Roig et al., unpublished).ItalsoremainsunclearwhetherfluencydeficitsinMSqualifyas differential deficits. In particular, MS patients are typicallyimpaired on the oral version of the Symbol Digit Modalities 0028-3932/$ – see front matter © 2005 Elsevier Ltd. All rights reserved.doi:10.1016/j.neuropsychologia.2005.10.006   J.D. Henry, W.W. Beatty / Neuropsychologia 44 (2006) 1166–1174  1167 Test (SDMT), and this has been interpreted as reflecting areduction in information processing speed (Beatty, Goodkin,Beatty, & Monson, 1989; Beatty, Goodkin, Monson, & Beatty,1989; Huijbregts et al., 2004). Since tests of verbal fluency also imposesubstantialdemandsuponspeedofinformationprocess-ing (Salthouse, Atkinson, & Berish, 2003), fluency deficits may simplyreflectamoregeneralizedreductioninprocessingspeed,and not specific executive decline.Two previous meta-analyses have quantified executiveimpairmentinrelationtoMS(Wishart&Sharpe1997;Zakzanis, 2000). Both suggest that fluency measures are more sensitive tothe presence of MS relative to other executive measures. Theyalso raise the possibility of a semantic memory deficit upontasks that additionally impose substantial demands upon cogni-tive speed. Zakzanis (2000) f ound semantic fluency to be more impaired than phonemic fluency, whilst in both reviews, phone-mic fluency was more impaired than confrontation naming, anon-speeded measure that is very sensitive to the integrity of semanticmemory.However,sinceneitherstudyquantifiedmeaneffects specifically for the  oral  version of the SDMT, it remainsunclear whether fluency deficits are disproportionate relative tomeasures of information processing that do not impose substan-tialdemandsonmotorabilities.Itisalsoproblematicthatinbothmeta-analyses different studies assessed each of the constructsof interest. This raises a potentially important problem, as theremayhavebeensubstantivedifferencesbetweenthepatientswithMS contributing to each statistic. 1.1. Aims of the current meta-analysis Following on from these meta-analyses, the relative promi-nence of deficits on tests of phonemic and semantic fluency willbeinvestigatedusingamethodologythatrestrictsstudiesineachcomparison to  only  those that assess both measures to be com-pared.The first aim is to derive effect size estimates for phonemicand semantic fluency for patients with MS relative to healthycontrols, and assess the relative prominence of deficits on each.The second aim is to assess whether either of these deficitsqualify as differential deficits relative to the  oral  version of the SDMT (Smith, 1982). The oral SDMT is considered to tap information processing speed, but minimises confounds aris-ing from motor disturbances. It is also important to addressthe possibility that phonemic and semantic fluency deficits sim-ply reflect a current general impairment in verbal abilities (seeCrawford & Henry, 2005). Thus, the pattern of deficits acrossfluency versus verbal intelligence (VIQ) as measured by theWAIS (Wechsler, 1955, 1981) Verbal and Vocabulary scales (VIQ) will be compared. Performance on tests of verbal flu-ency will also be compared with scores on the Boston NamingTest (BNT; Kaplan, Goodglass, & Weintraub, 1983), a mea- sure of confrontation naming. The Wisconsin Card Sorting Test(WCST) will also be included for comparison standards, as likeverbal fluency, the WCST is considered to impose substantialdemandsuponexecutiveprocesses.Thethirdaimistoassesstherelationshipbetweenfluencyperformancewithpatients’levelof neurological disability, age and disease course. 2. Method 2.1. Sample of studies A search involving the  Web of Science ,  Psych Lit CD-ROM   and  Science Direct   databases was undertaken, using the terms: letter fluency, FAS, semanticfluency,categoryfluency,controlledoralwordassociation,COWA(T),wordflu-ency, verbal fluency, oral fluency, phonemic fluency, executive test, frontal testand multiple sclerosis. A manual search of issues of   Neuropsychologia ,  Mul-tiple Sclerosis ,  The Journal of the International Neuropsychological Society ,  Neuropsychology , TheClinicalNeuropsychologist  ,  Neuropsychiatry ,  Neuropsy-chology and Behavioural Neurology ,  Journal of Neuropsychiatry and Clinical Neurosciences  and the  Journal of Clinical and Experimental Neuropsychology was also conducted. The search was completed in December 2004.The inclusion criteria were that the study had to include: (1) a patient groupconsisting entirely of adults with MS, (2) a healthy control group free fromneurological or psychiatric disease and (3) a measure of phonemic or seman-tic fluency. Effect size estimates for WCST CC, WCST PE, BNT, oral SDMTand VIQ were derived from studies that also reported fluency results. The studymust also have (4) presented  precise  statistics convertible to effect size. Somestudies could not be included because only a total score collapsed across phone-mic and semantic fluency was reported (Tsolaki et al., 1994), others because despite inclusion of an MS group and a healthy control group, precise statis-tics convertible to effect size for the fluency measure(s) of interest were notreported (Cohen & Fisher, 1989; Grossman et al., 1995; Oliveri et al., 1999). One study was excluded as it was not entirely clear what ‘word generation’referred to (van Dijk, Jennekens-Schinkel, Caekebeke, & Zwinderman, 1992).Finally, some studies were excluded because they reported data for MS patientsthat were already included in another eligible study (Beatty & Monson, 1989;Foong et al., 1999; Ryan, Clark, Klonoff, & Paty, 1993). 2.2. Statistical analysis The basis of meta-analysis is the effect size, a standardized statistic thatquantifies the magnitude of an effect. In the present study, the effect size  r   wasused. For each construct, effects were pooled using a random effects model toderive an estimate of the mean, with each effect weighted for sample size tocorrect for sampling error. To estimate the degree of heterogeneity of the effectscontributing to each mean, the homogeneity statistic  Q  was estimated, as wellas the S.D. of random effects, and the 95% confidence intervals (CI) withinwhich random effects can be expected to fall. To test whether differences in themagnitude of mean effects were significant paired  t  -tests were computed usingthe number of studies ( K  ) as the d.f. For a fuller outline of these statistics, seeHenry, Crawford, and Phillips (2004). 3. Results 3.1. Participant characteristics Thirty-five research articles published or conducted between1985and2004contributedtothepresentanalyses,inwhichtherewere a total of 2339 patients and 1334 controls. Demographicinformation for patients and controls is presented in Table 1; it can be seen that they are closely matched for age, education andgender. Clinical characteristics of patients are also presentedin Table 1. Patients’ mean score on the Expanded Disability Status Scale (EDSS; Kurtzke, 1983), an index of disease sever- ity that categorises level of neurological disability, is 3.8; thismeans that the average patient was fully ambulatory without aiddespite relatively severe disability. In Appendix A, effect sizes forphonemicandsemanticfluency,aswellasinformationrelat-ingtodemographicandclinicalvariables,arepresentedforeachindividual study that contributed to the meta-analysis.  1168  J.D. Henry, W.W. Beatty / Neuropsychologia 44 (2006) 1166–1174 Table 1Descriptive statistics for MS patients and controlsMS sample Control sample K M   S.D.  K M   S.D.Age 35 41.2 4.38 33 40.8 5.69Education 27 13.8 1.18 25 14.1 1.17Gender (male, %) 28 33.5 12.50 27 33.7 13.40Age at diagnosis 26 32.1 3.45 – – –Disease duration(since diagnosis)26 9.1 3.36 – – –Disease duration(since symptomonset)8 12.6 2.49 – – –Expanded DisabilityStatus Scale28 3.8 1.59 – – –Table 2Performance on phonemic fluency (PF), semantic fluency (SF) for MS patientsvs. healthy controlsFluency measure  M K N Q  S.D. 95% CILower UpperPhonemic .42 ** 16 1060 43.4 ** .14 .15 .69Semantic .42 ** 16 1060 56.1 ** .16 .10 .73 **  p <.01. 3.2. Effect sizes for patients with MS relative to healthycontrols Cohen (1977) defines correlations of .1 as small, .3 as mod-erate and .5 as large. It can be seen in Table 2 that the deficits for phonemic and semantic fluency are of a moderate to largemagnitude(both r  =.42),andarebothassociatedwithsignificantheterogeneity.In Table 3, mean effects for each of the cognitive measures are presented, calculated using only those studies that includethe particular measure of interest  in addition  to phonemic flu-ency (upper half of  Table 3) or semantic fluency (lower half  of  Table 3). Thus, for comparisons with WCST CC, WCST PE, BNT,SDMTandVIQthemeaneffectsforphonemicandseman-tic fluency have been re-calculated (these effect sizes appear inthelastcolumnof Table3).Allmeaneffectsarestatisticallysig- nificant(all  p <.05).RelativetotheWCSTandtheBNT,phone-mic and semantic fluency are more impaired, although only thecomparisonswiththeBNTattainsignificance(forphonemicflu-ency:  t  =3.28, d.f.=10,  p =.008; for semantic fluency:  t  =3.89,d.f.=7,  p =.006). Phonemic fluency is also significantly moresensitive to the presence of MS than VIQ ( t  =4.06, d.f.=10,  p =.002). However, neither phonemic nor semantic fluency ismore impaired than the SDMT. 3.3. Moderators of verbal fluency deficits in MS  M  s and S.D.s are presented in Table 4 for phonemic andsemantic fluency, sub-divided according to disease course.Moststudiesdifferentiatebetweenrelapsingremitting(RR)andchronicprogressive(CP) formsof thedisease; whilst theformerrefers to patients who experience periods of relapses and remis-sions, the latter refers to patients whose course is chronicallyprogressive. The studies that contributed to the following analy-sesconsistedentirelyofpatientswith either  aCPoraRRcourse,andincludedtestsofbothphonemicandsemanticfluency.Giventhatonlythreestudiescontributedtoeachofthesetwogroups,itwould be inappropriate to conduct inferential statistics compar-ingthesetwogroups(statisticalpowerwouldbelow).However,there is a trend for patients presenting with a CP course to bemore impaired on both phonemic and semantic fluency rela-tive to patients presenting with an RR course. Patients with a Table 3Performance on phonemic fluency (PF), semantic fluency (SF) and other cognitive measures for MS patients vs. healthy controlsVariable  M K N Q  S.D. 95% CILower UpperStudies with PFWCST CC .25 ** 8 675 15.9 * .12 .02 .48 .34 ( K  =8) a WCST PE .25 ** 9 911 7.2  − − −  .31 ( K  =9) a BNT .22 ** 11 876 11.2 .04 .15 .29 .36 ( K  =11) a SDMT .48 ** 6 350 9.3 .10 .29 .67 .41 ( K  =6) a VIQ .18 ** 11 1692 24.9 * .10  − .01 .36 .29 ( K  =11) a Studies with SFWCST CC .29 ** 5 366 8.1 .11 .08 .51 .33 ( K  =5) b WCST PE .28 ** 6 400 3.6 – – – .34 ( K  =6) b BNT .27 ** 8 461 10.3 .08 .11 .44 .44 ( K  =8) b SDMT .41 ** 6 635 15.2 * .12 .18 .65 .42 ( K  =6) b (–) The random effects variance has been estimated to be zero.  Note : For comparisons, the mean effects for PF and SF were recalculated for each comparison of interest. For example, six studies included both PF and SDMT. In addition to calculating the mean effect for SDMT from these six studies ( r  =.48), the mean effectfor PF was also recalculated based  only  on these six studies (i.e.  r  =.41). Thus, in each comparison exactly the same participants have been tested upon each of themeasures of interest, ‘controlling’ for any substantive differences between studies, such as in level of dementia severity. a PF,  M  . b SF,  M  . *  p <.05. **  p <.01.   J.D. Henry, W.W. Beatty / Neuropsychologia 44 (2006) 1166–1174  1169Table 4Phonemic and semantic fluency mean effects stratified according to disease sub-typePhonemic fluency Semantic fluency K   Patient,  N   Duration EDSS  M   S.E.  K   Patient,  N   Duration EDSS  M   S.E.Chronic progressive 3 112 15.3 6.4 .54 .05 3 112 15.3 6.4 .47 .10Relapsing remitting 3 110 6.7 3.5 .30 .07 3 110 6.7 3.5 .38 .06 CP course also have a longer duration of illness (over twice aslong), and a substantially higher level of neurological disability.To investigate the relationship between age, duration of ill-ness and neurological disability with fluency performance, cor-relations were calculated. It can be seen in Table 5 that although not significant, phonemic and semantic fluency deficits wereweakly correlated with age, and moderately correlated withduration of illness. Correlations of a large magnitude wereobserved in relation to level of neurological disability;  r  =.47and .49 for phonemic and semantic fluency, respectively; whilstthe former correlation was significant (  p =.015), the latter justfailed to attain significance (  p =.067).Raw and partial correlations are also presented in Table 5between each of the fluency measures with different types of disease course. Using the same methodology as Thornton andRaz (1997), the percentage of patients who were RR or CP ineach study was calculated. (Studies that did not indicate diseasecourse did not contribute to these analyses.) Consistent with thefindings reported in Table 4, raw correlations indicate that the magnitudeofimpairmentontestsofphonemicandsemanticflu-encyissubstantiallynegativelyassociatedwiththepercentageof patients presenting with a RR disease course, and substantiallypositively associated with the percentage of patients presentingwith a CP disease course (although the correlations with seman-tic fluency did not attain significance). However, when partialcorrelations were calculated, controlling for duration of illness,neurologicaldisabilityandage,thesecorrelationsweresubstan-tially attenuated (partial  r   ranged from − .05 to .18). Table 5Raw and partial correlations (controlling for duration of illness, neurologicaldisability and age) between fluency and patient characteristicsVariable Raw correlations Partial correlationsd.f.  r p  d.f.  r p Phonemic fluencyDuration of illness 21 .36 .103 – – –EDSS 25 .47 * .015 – – –Patients age 31 .24 .188 – – –Relapse remitting (%) 24  − .50 * .013 11 .03 .924Chronic progressive (%) 24 .47 * .019 11 .08 .797Semantic fluencyDuration of illness 15 .29 .282 – – –EDSS 14 .49 .067 – – –Patients age 19 .17 .484 – – –Relapse remitting (%) 16  − .38 .146 7 .18 .678Chronic progressive (%) 16 .44 .091 7  − .05 .915 *  p <.05. 3.4. Assessing the possibility of publication bias Funnelplotdiagramswereconstructedforeachofthefluencyand non-fluency measures of interest. In these diagrams, sam-ple size is plotted against the corresponding study-level effect;if statistically non-significant results have been discriminatedagainst, there should be a relative absence of studies with smallsample sizes that report weak effects. For none of the variableswas there evidence of this bias operating. 4. Discussion 4.1. Fluency deficits in MS  The only previous meta-analytic review that has quantifiedmean effects for both phonemic and semantic fluency found thelatter to be substantially more impaired (Zakzanis, 2000), con- sistent with the possibility that MS is associated with deficitsin semantic memory. However, as noted earlier, it is importantthatthepatientscontributingtothemeaneffectforsemanticflu-ency do not differ from the patients contributing to the mean forphonemic fluency, if comparisons between these two measuresare to be fair. When mean effects were calculated in the presentstudy using this methodology, there was no difference in thesensitivity of the two measures to the presence of MS ( r  =.42for both measures).The finding of equivalent phonemic and semantic fluencydeficits has important clinical implications. Verbal fluency testsprovide brief but sensitive measures of cognitive functioning inMS. These tests have the additional advantages of being rel-atively low in stress for patients and minimally affected bythe motor and visual impairments that are commonly associ-ated with the disorder. A phonemic fluency test is one com-ponent of the Brief Repeatable Battery (BRB; Rao, 1990) that is widely used in the English-speaking parts of North Amer-ica, Europe and Australasia. However, the frequency of wordsthat began with a particular letter varies from one language toanother and this factor may produce biases in the sensitivityof the test for patients who speak different languages. Using asemantic fluency test with a large number of exemplars (e.g.,animals and supermarket items) in every language is one solu-tion that has been adopted in the non-English speaking versionof the BRB which is used in Europe (Boringa et al., 2001). The present findings that phonemic and semantic test are equallysensitive to MS encourages the view that the results on flu-ency tests for patients who speak different languages will becomparable.
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks