An Examination of the Concept of Central Coherence in Women with Anorexia Nervosa

Please download to get full document.

View again

of 10
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.
Document Description
An Examination of the Concept of Central Coherence in Women with Anorexia Nervosa
Document Share
Document Tags
Document Transcript
  An Examination of the Concept of Central Coherence inWomen with Anorexia Nervosa Carolina Lopez, MSc 1 *Kate Tchanturia, PhD 1 Daniel Stahl, PhD 2 Rhonda Booth, PhD 3  Joanna Holliday, PhD 4  Janet Treasure, MD, PhD, FRCP,FRCPsych 5 ABSTRACT Objective:  To examine central coher-ence (local and global processing) inwomen with anorexia nervosa (AN). Method:  42 women with AN and 42healthy women (HC) completed neuro-psychological testing measuring visuo-spatial and verbal aspects of central co-herence: Rey-Osterrieth Complex Figure(RCFT), Embedded Figures Test (EFT),Homograph Reading Test (HRT), and Sen-tence Completion Task (SCT). Results:  People with AN displayedsuperior performance on the EFT andpoorer performance in RCFT with theexception of accuracy in the copy trial.Long hesitations in the SCT wereobserved. Verbal coherence tasks werenot sensitive enough to detect coherenceanomalies in AN. Conclusion:  Women with AN havestrengths in tasks requiring local process-ing (EFT) and weaknesses on tasks bene-fited by global processing (RCFT and SCT).These results are consistent with theweak central coherence account. Thistrait might play a role in the mainte-nance of AN and can be addressed inspecific clinical interventions.  V V C  2007 byWiley Periodicals, Inc. Keywords:  central coherence; anorexia;cognition; information processing; localprocessing; global processing; neuro-psychology (Int J Eat Disord 2008; 41:143–152) Introduction Developmental problems as well as obsessive-com-pulsive disorders (OCD) and obsessive-compulsivepersonality traits (OCPD) might be considered tobe enduring risk factors of relevance to the onsetand prognosis of eating disorders (EDs). 1 It is possi-ble that neurocognitive anomalies may underpinOCD and OCPD traits. For example, in a systematicreview of the literature, the overall conclusiondrawn from a meta-analysis of data pertaining toset-shifting (switching between strategies and stim-uli) was that there was moderate sized evidencethat difficulties in set-shifting was associated withanorexia nervosa (AN). Difficulties in set-shifting  were found to be linked to childhood perfectionismin people with AN. 2 This and other neurocognitive abnormalities areof interest as they support the argument that EDsare neurodevelopmental in srcin. 3–5 Gillberg et al. 6 raised the possibility that AN mightbe considered to be part of the autistic spectrum of disorders (ASD). He and his colleagues have col-lecteddataonneurocognitiveand socialfunctioning in a cohort of cases prospectively studied over 20 years that support this hypothesis. They found that20% of their AN group met criteria for ASD comor-bidity including social impairments and deficits inempathy. 7 They also observed normal to superiorperformance in local processing dependent tasks(e.g. the Block Design subtest of the WAIS-R) andpoor function on tasks requiring global processing strategies (e.g. the Object Assembly subscale of the WAIS-R). 6 Furthermore, the group with ASD comor-bidity performed poorly in the picture arrangementtest. 6 This pattern of performance fits with what hasbeen termed  weak central coherence  . 8–10  Weak central coherence refers to a cognitive stylein which there is a bias towards local or detailed-focus processing of information over the naturaltendency to integrate information into a context. 8,9  Weak central coherence is thought to explain some Supported by Nina Jackson Foundation. *Correspondence to:  Carolina Lopez, Department of AcademicPsychiatry, 5th Floor, Thomas Guy House, Guy’s Hospital, London,SE1 9RT, UK. E-mail: 31 July 2007 1 Division of Psychological Medicine, King’s College London,Institute of Psychiatry, IoP, United Kingdom 2 Department of Biostatistics and Computing, King’s CollegeLondon, IoP, United Kingdom 3 SGDP Centre, King’s College London, IoP, United Kingdom 4 Oxford Doctoral Course in Clinical Psychology, The IsisEducation Centre, Warneford Hospital, Oxford, United Kingdom 5 Department of Academic Psychiatry, Guy’s, King’s andSt. Thomas’s Medical School, London, United KingdomPublished online 15 October 2007 in Wiley InterScience( DOI: 10.1002/eat.20478 V V C  2007 Wiley Periodicals, Inc. International Journal of Eating Disorders 41:2 143–152 2008— DOI 10.1002/eat  143 REGULAR ARTICLE  of the cognitive abnormalities in ASD. 9,11  A recentreview of more than 50 studies measuring aspectsof central coherence in ASD concluded that there was strong evidence for a bias to local processing but mixed findings with regard to weak global co-herence in this group. 9 The concept of weak central coherence is new inthe field of EDs. 10 However, it would fit with thefindings from the neurocognitive literature where a preoccupation with detail and a narrow style of in-formation processing have been described. 10,12,13  Also it may explain some aspects of the symptoma-tology such as the preoccupation with rules anddetails captured by the Yale-Brown-Cornell Eating Disorder Scale (YBC-EDS). 14 Happe´ and colleagues have described, developed,and adapted various tests that measure the conceptof central coherence in the visuospatial, verbal, andauditory domains. 15 These tests have been used in a population of ASD and some also in first degree rela-tives. 9,16 It is still unclear whether the concept of central coherence is the result of an integratedmechanism that affects information processing fromacross different systems/domains or if it should beconsidered a variable feature within each system/do-main. Hence, there is a need to research into centralcoherence using tasks addressing different domains(e.g. verbal, visuospatial, auditory, etc.). 9  An important finding in ASD literature is that theclassical symptom groups of autism (social impair-ments, communication impairments, and re-stricted repetitive behaviors and interests) are etio-logically independent. Thus, the verbal communi-cation deficits in autism and the compulsivesymptoms have distinct causal pathways. 17 EDsmight therefore only have some of the symptomclusters in common with ASD. For example, they may be more akin to high functioning Aspergersand have fewer verbal difficulties.The aim of this study was to measure central co-herence in people with AN using some of the testsfor central coherence that have been used previ-ously in the ASD population. Our hypothesis wasthat people with AN would display a local process-ing bias and/or weak global processing. A secondary hypothesis was that this would be most pronouncedin visuospatial tasks rather than verbal tasks. Method Participants  Women meeting the DSM-IV criteria for AN (  N   5  42)and healthy controls (HC,  N   5  42) matched for sex andequivalent in age and intellectual ability participated inthis study.Participants with AN were ascertained from the SouthLondon and Maudsley NHS Trust and local community:29 (69%) had AN restrictive subtype and 13 (31%) purging subtype of AN. Also, 17 (40.5%) were taking psychoactivemedication (58.8% SSRI antidepressants) and 15 (35.7%) were undergoing inpatient treatment at the time of thestudy.HC were recruited from the staff and student popula-tion of the Institute of Psychiatry and King’s College of London, from advertisements on the Eating DisordersUnit website, and local community. Inclusion criteria included a body mass index (BMI) between 19 and 26(kg/m 2 ), no psychoactive medicines, and no personal orfamily history of psychiatric illness or EDs.The exclusion criteria for all participants were as fol-lows: a history of head injury, neurological diseases, psy-chosis, and learning disabilities. All participants wereEnglish speakers.Following a complete description of the study, writteninformed consent was obtained from all participants.Ethical approval was obtained from the South Londonand Maudsley Trust (SLAM) Research Ethics Committee. Measures ED Diagnosis.  Participants in the AN group wereassessed with the semi-structured diagnostic interview EATATE for ED (part I). 18 HC completed the Eating Disor-der Examination-Questionnaire (EDE-Q 19 ) modified toprovide a lifetime screening for ED symptoms. Addition-ally, parts II and III of EATATE were administered to allparticipants to examine for a history of other psychiatricproblems. Information Processing.  The neuropsychological bat-tery consisted of four instruments, selected to demon-strate either predominant global or local informationprocessing at both verbal and visuospatial levels. 9 The Rey-Osterrieth Complex Figure Test - recall  form.  Recall form (RCFT) is an open-ended measureof visual perceptual organization and nonverbal mem-ory. 20 In this study, participants were provided with a blank sheet of paper, colored pencils, and a sheet withthe figure. They were asked to copy and then recall thefigure after a 20-min interval without previous warning.Pencils were proffered when the participant completedany element of the figure. The drawing process was video-recordedandscoredusingproceduresdescribedbyTaylor(adapted from Osterrieth, 1944) for  accuracy  . 20–22 The  Accuracy   score (range 0–36) comprises a total of 18 ele-ments awarded a 0 to 2 quality score (see Fig. 1). Theintraclass correlation coefficient for inter-rater agree-ment in accuracy score was .89 with an average Kappa  LOPEZ ET AL. 144  International Journal of Eating Disorders 41:2 143–152 2008  —DOI 10.1002/eat  coefficient ( j ) of .85, ranging from .72 to .95 for the indi-vidual elements, in this study. The percentage of recall(recall/copy*100) was calculated as a measure of recallaccuracy.Drawing style was also assessed using both Savage’s 23 and Booth’s scoring systems (2006).Savage’s scoring system addressing organizationalstrategy (OS) scores the five main elements of the figure(0–6 points). Higher scores are awarded to better OS.Details of this system have been described elsewhere. 23 In contrast, Booth’s scoring was specifically designed tomeasure central coherence and it addresses not only thecontinuity of drawing but also whether the drawing pro-cess beings with the detail or the global elements.Following Booth (2006) the degree of coherence in thedrawing process was measured in two independent ele-ments: (i) Order of Construction Index (mean range 0–3.3): the first six elements drawn are awarded 0–4 pointsaccording to the following hierarchical categories: (a)global external structure (elements 2 and 13) 5 4 points;(b) global internal structure (elements 3, 4, 5, and 16) 5 3points; (c) local perimeter elements (elements 1, 9, 14,17, and 18)  5  1 point; and (d) local internal elements(elements 6, 7, 8, 10, 11, 12, and 15) 5 0 points. The sumof these scores is averaged leading to the order index. A higher score is given if the first stage of the drawing involves the global features rather than the detail. Theintraclass correlation coefficient for the order index was.93. (ii) Style Index (mean range 0–2): six specific ele-ments are scored according the ‘‘degree of continuity’’ of the drawing process: 2 points given where lines aredrawn in a continuous stroke or drawn consecutively, 1point where the element was partially fragmented ordrawn separately, and 0 points where the line was clearly disjointed or drawn in a fragmented manner. The sum of these scores is averaged, leading to the style index. A higher score indicates more continuity in the drawing style. The intraclass correlation coefficient for the styleindex was .94 in this study with an average Kappa coeffi-cient of .89 (ranging from .84 to .96).(3) Coherence Index (CI  5  0–2) was obtained by add-ing the proportion of the total possible scores in  order  (score/3.3) and  style indices   (score/2). A higher score inthe  coherence index   means a more coherent drawing style, e.g. drawing of global elements initially and a con-tinuous (versus fragmented) drawing style for the mainelements of the figure. The intraclass correlation coeffi-cient for  central coherence index   was .97. The Embedded Figures Test—form B (EFT).  24 Thisperceptual test measures the time taken to find 12 simpleshapes embedded in complex, colored designs. In thisstudy, the memory element of the task was omitted by displaying the simple shape and the complex designsimultaneously. Participants are given a period of 60 s tofind each simple shape. Timing is stopped while the par-ticipant demonstrates the position of the figure. If theparticipant is wrong a ‘‘false claim’’ is noted, the partici-pant is asked to keep looking and timing is resumed upto 60 s. After 60 s the item is reported as an ‘‘error’’ andgiven the maximum score. The score for each item rangesfrom 0 to 60. The total score is the median of the 12 itemscores. A shorter time and fewer errors suggest strongerlocal processing as participants need to ignore the back-ground (context) and focus just on the detail of theshape. Reliability in populations over 16 years old rangesfrom .61 and .90. 24 The Sentence Completion Task (SCT) (modified fromHappe et al., 2001 16   ).  This verbal test consists of 25sentences, 18 constructed to provoke conflict betweenlocal and global processing and 7 without that tension.This version was modified in consultation with Happe´and co-workers. Four of the 15 srcinal items that werefound to be insensitive in trials with ASD were removedand 14 new items were added. New items were selectedafter a pilot study that demonstrated they were valid forthe purpose of the task.In this task, the participant is asked to ‘‘finish off’’ theending of a pre-recorded sentence, presented on a tape-recorder. 18 of the sentences end with a word which is of-ten paired with another, such as ‘‘up and . . . (down)’’. Thiscommon pairing is a ‘‘local’’ association and in the con-text of the whole sentence it is incorrect. For example,‘‘In the morning the sun went up and . . . ’’the local associ-ation ‘‘down’’ is nonsensical whereas ‘‘shone all day’’ isan ending that is correct in the overall global context.The verbal response and the time taken to produce it arerecorded. The scoring is as follows: local completions 5 2points; answers given after a delay of 10 s, ‘‘odd’’ answersbut not clear local completions, or no answer 5 1 point;meaningful completions given within 10 s  5  0 points.Higher scores reflect a preference for local processing ordifficulties in global processing. 15 The maximum time FIGURE 1. Rey-Osterrrieth Complex Figure. CENTRAL COHERENCE IN ANOREXIA International Journal of Eating Disorders 41:2 143–152 2008  —DOI 10.1002/eat  145  allowed for each completion is 20 s and scoring is basedon participants’ first response. The Homograph Reading Task (HRT).  25,26  Thisverbal task tests the ability to process a sentence as a  whole rather than on a local level. The participant isasked to read out a set of 16 sentences where the contextof the sentence determines how a homograph within itshould be pronounced. The position of the homographin the sentences varies (after or before the context) andthe level of usage of the word (rare or frequent) varies.Thus there are four different conditions i.e. after orbefore frequent usage and after or before rare usage. Allresponses were recorded. One point was awarded foreach sentence read correctly based on participants’ ini-tial pronunciation. The total score ranges from 0 to 16, where higher scores indicate strong coherence. IQ Levels and Comorbidity.  The National Adult Reading Test(NART) 27  was used as anestimate of intellectualability.The level of comorbidity was assessed in terms of traitmeasures of the common comorbid symptoms (anxiety,depression, obsessive-compulsive symptoms) by theHospital Anxiety and Depression Inventory (HADS 28 ) andthe Obsessive Compulsive Inventory-Revised (OCI-R 29 ). Procedure   Weight and height were measured on the day of test-ing. All the neuropsychological measures were adminis-tered in the same session in this order: RFCT (copy), SCT,EFT, HRT, and RCFT (recall). Questionnaires were com-pleted a week before or after the testing session. TheNART test was not obtained from the first 20 participants.However, as they were recruited from the same sourcesas the rest of the sample we do not expect IQ differences.Three women from the AN group and one HC did notreturn all or part of the questionnaires. Their data wereincluded in the analyses of the main outcomes measures. Data analysis Sample size power calculation was conducted using NQuery 4.0 Advisor program 30 based on a pilot study forthis project. It was found that a sample size of 42 partici-pants in each group would have an 80% power to detectsignificant differences between groups with a .05 two-tailed significance level.Independent sample  t  -tests were used for group com-parisons and means (M) and standard deviations (SD) arereported. For those measures without a normal distribu-tion of the data (SCT, HRT, EFT, depression (HADS), andOCI-R), nonparametric Mann-Whitney   U   tests were usedand medians (Mdn) and 25 and 75% quartiles (Q) reported. Additionally, Kruskal-Wallis tests were performed on themean score for the four conditions of the HRT.To explore the relationship between neuropsychologi-cal, demographic, and clinical variables, Pearson’s corre-lation coefficient ( r  ) was used for normally distributeddata and Spearman’s rank ( r  s ) correlation was used other- wise. All correlation analyses were conducted separately in each group and Cohen’s  d   was calculated to obtain theeffect size for each variable. For those variables with non-normally distributed data,  d   was calculated by transform-ing the effect-size correlation  r   5  Z   ffiffiffi   N  p   (from Ref. 31) in  d  using an effect size calculator. We were also interested to measure the process in-volved to RCFT. Sherman and colleagues (2006) arguedthat the poor visual memory performance in AN couldbe partly explained by an anomaly in OS. We first repli-cated Sherman et al.’s study and second, tested anadapted hypothesis which was to examine whether cen-tral coherence indices (CI) could also explain poor vis-ual memory. A series of regression analyses were used toexamine mediating/moderating effects of organiza-tional strategy, order, style and coherence indices onrecall in the RCFT. 32,33 Group was entered as an inde-pendent dummy-coded variable, percentage of recall asthe dependent variable and CI as well as OS score as thepotential mediators. 32  We present standardized beta (st. b ) for a comparison of the effects of independent var-iables in regressions with a different measurementrange of observations. The assumptions of the linearregression models were assessed by a visual inspectionof the residuals. 34  As this was a novel study, two-tailed tests were usedthroughout. The nominal significance level was chosento be 5%. Hochberg’s improved Bonferroni correction formultiple testing was applied on the main outcome meas-ures. 35  All analyses were carried out using SPSS 13.0. Results Demographic and Clinical Characteristics The results of demographic and clinical charac-teristics are shown in  Table 1 . People with AN havefewer years of education than HC group ( U   5 558.5,  p  5  .01) despite having equivalent levels intheir estimated IQ ( t   5  .4,  p  5 .65). The AN grouphad had a prolonged illness (M 5 13.08 years, SD 5 11.2) with profound emaciation and high levels of depressive, anxiety and obsessive-compulsivesymptoms. The mean age of onset for the AN group was 15.7 years (SD 5 4.0). Neuropsychological Function The results of the neuropsychological tests areshown in  Table 2 . Rey-Osterrieth Complex Figure Test.  The AN grouphad lower scores on the RCFT drawing process var- LOPEZ ET AL. 146  International Journal of Eating Disorders 41:2 143–152 2008  —DOI 10.1002/eat  iables (CI and OS scores) and on recall but not oncopy accuracy. Overall, these results indicate thatpeople with AN display a more detailed-focus andfragmented drawing style than HC. Instead, healthy participants tended to draw the figure using a moreglobal approach. The results indicated a markeddegree of correlation ( r  s 5 .77,  p \ .001) between CIand OS in this study. Embedded Figures Test.  Participants with AN wereable to identify the hidden figures faster and com-pleted more items within the time limit than theHC group. They also made fewer false claims dur-ing the task, demonstrating better accuracy on thistask. Sentence Completion Task.  The AN group had a higher local processing score due to them hesitat-ing longer before producing an appropriate com-pletion of the sentences. However, the frequency of local completions made by either group was rare(M 5 .21, SD 5 .52 in AN and M 5 .17, SD 5 .54 inHC). Following the procedure described by Happe´and collaborators (2001), participants were dividedinto groups showing good and poor performance. An individual met the criteria for a poor perform-ance if they had either one or more local comple-tions, or two or more long hesitations. Good per-formance was defined as: no local completions andno more than one long hesitation. 35.7% of the ANgroup fell in the poor performance group whereasonly 16.7% of the HC did (Fisher’s Exact Test  5 .081). Homograph Reading Task.  No differences were foundboth in the total score and across the four homo-graph conditions (all  p  [  .44). Most resultsremained significant after correction for multipletesting. The Role of the Coherence Index (CI) and Organizational Strategies (OS) on Visual Memory in the RCFT  The hypothesis that some of the group differencein visual memory (RCFT accuracy recall) could be TABLE 1. Demographic and clinical characteristics of the sample N AN Group HC GroupTest Statistic p -Value Effect Size ( d  )AN HC  t  -Test a MW b Age (years) a 42 42 28.4 (9.6) 26.3 (6.4) 0.5 na .58 0.26Number of years of education b 39 42 15 (12–18) 17 (14.8–19) na 558.5 .01  2 0.56Estimated IQ a 37 27 112.8 (6.8) 112.2 (5.4) 0.4 na .65 0.09BMI (kg/m 2 ) a 42 42 15.8 (1.7) 21.9 (2.7) 7.8 na  \ .001  2 2.70HADS anxiety a 39 41 14.2 (4.6) 6.0 (3.2) 9.4 na  \ .001 2.07HADS depression b 39 41 10.0 (6–14) 2.0 (1–3) na 115.0  \ .001 2.20OCI-R b 39 42 28.0 (16–38) 6.5 (3–10) na 165.0  \ .001 1.90 na, not applicable. a Test statistics for  t  -test pairwise comparisons for data normally distributed, mean values displayed with standard deviations. b Test statistics for Mann-Whitney  U  for data not normally distributed, median values displayed with upper and lower quartiles.Notes: All significant results displayed remained significant after Hochberg’s improved Bonferroni correction for multiple testing 35 at  p \ .05. Anxiety,depression, and OCD symptoms are measured in terms of level of severity and they do not indicate a clinical diagnosis. TABLE 2. Neuropsychological performance on central coherence tasks AN Group ( N 5 42) HC Group ( N 5 42)Test Statistic p -Value Effect Size ( d  ) t  -Test a MW b RCFT COPY a Accuracy 30.9 (3.0) 30.9 (3.6) 0.33 na .97 0Order 2.1 (0.6) 2.5 (0.4) 3.9 na  \ .001 c 2 0.89Style 1.3 (0.5) 1.7 (0.3) 3.6 na .001 c 2 0.98RCFT RECALL accuracy a 16.7 (5.6) 19.6 (5.7) 2.3 na .02 d 2 0.52RCFT CC index copy a 1.3 (0.4) 1.6 (0.3) 4.1 na  \ .001 c 2 0.93RCFT percentage recall a 53.7 (16.1) 62.8 (15.8) 2.6 na .01 d 2 0.58Organizational strategy b 4 (2–5) 5 (3–6) na 552.5 .003 c 2 0.70SCT local processing score b 1 (0–2) 0 (0–1) na 591.5 .004 c 0.65HRT total score b 15 (14–16) 15 (14–16) na 877.5 .97 0EFT total time taken (medians) b 8.6 (5.8–13.9) 12.2 (7.9–15.3) na 639.0 .03 d 2 0.50EFT number time out failures b 0 (0–1) 1 (0–2) na 525.5 .001 c 2 0.82 na, not applicable. a t   test statistics for  t  -test pairwise comparisons for data normally distributed, mean values displayed with standard deviations. b U  test statistics for Mann-Whitney  U  for data not normally distributed, median values displayed with 25 and 75 percentiles. c After Hochberg correction for multiple testing 35 significant results displayed remained significant ( p \ .05) d After Hochberg correction for multiple testing 35 significant results displayed increased slightly to trend ( p \ .1). CENTRAL COHERENCE IN ANOREXIA International Journal of Eating Disorders 41:2 143–152 2008  —DOI 10.1002/eat  147
Similar documents
View more...
Search Related
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