An analysis of Crohn ' s disease activity indices from a registry of patients in eastern Pennsylvania

The objectives of this report were to compare Crohn’s disease patients’ assessment of disease activity and its effect on their well-being to their physician’s assessment of the disease activity and the patient’s well-being; and to use existing Crohn’s disease indices in comparing the severity of Crohn’s disease across various indices. The indices included in this study were the National Cooperative Crohn’s Disease Study index, the Harvey and Bradshaw Index, the Oxford Index, the Modified Organisation Mondiale de Gastroenterologie Index, the Cape Town Index, the Bristol Index, and the St Marks Index. The comparison of Crohn’s disease patient self-evaluation and physician evaluation of the patient indicates a significant agreement between patient and physician assessment of disease activity (Χ2 41.68, P=0.00001). Cohen’s weighted kappa, a chance-corrected coefficient of agreement for nominal scales, was also used to evaluate the quality of agreement between physician and patient assessment of well-being. The weighted kappa (0.2322, P=0.0002) indicates a significant level of agreement.


T l IE DEVELOPMENT OF A RELIARLE
and objective means of measuring disease 'acti vity' in Crohn's disease in terms of patienL well-being has spanned several decades a nd has resul ted in several indices (Tables 1-7) (1 -9) .These indices have been used to assess the Hctivity ()f Crnhn 's disease and the response of th e disease to a treatment regimen .This rep()rt is nei t her cl comprehensive review of these indices, nor is it a critique of I hose indices.For a review or critique of the d1se;1,c activi t y indices, the authors defer to the papers hy Bartho lomc u sz and S he a rman ( I 0), DeDom hal ( 11 ), or Gar rett and Drossman (I 2).
A major objecti ve of the preli minary study Lo th e Nat io nal Cooperative Crohn \ D isease Study (I) was to develop an index of Lhe overall activity of C rohn's Jisease that wou!J: incorporate factors considered to be impo rtan t indicators of disease activity; use ohservatio ns readil y avai lable during a patient visit; weigh the component fac-t<m in an intuitively appropriate directi ll n; require only relmi ve ly simple computations; be consiste nt with the physician's overall appra isal of changes in the patient's condition; correlate well with the physic ia n's overall appraisa l of disease activity; and weigh the component factors to optimize accurncy with the predicted physicians' apprnisa l (I).
Eighteen in dependent variables were included.Each va riable was coded in such a way th at the normal vn lue was  One day scoring 0, and pr(')grcss ively la rger values we re expectcJ co reflect greater activity of Crohn'~ disease.T he dependent va riable for each patiem wa~ th e physic ian 's overa ll evaluat ion of 'how the patient was J o ing'.Fou r categories were established : 'very wel l', 'fa ir to good', 'poor', and 'very poor'.A stepwise mu lt ip le regression model iden t ifi ed e ight signi fica n t vari ahles as concrihutors to th e Croh n'~ d bease activity index (CDAJ) ( J ). CDAI values of J 50 or bclmv arc associated wit h q ui escent d isease (phys ic ian evaluation of 'very well' or 'fair to good').CDAI va lues above 150 a re assoc iated wit h act ive di sease ( phys ician evaluat l\lll of 'poor' or 'very poor' ).The C DA I ha!> served as the ha llmark in the deve lopment of subsequent Crohn's d bease indi ce~ (2-9).T he authors abo use this index as the basis of t he present report.
T he objectives of th is report were: to compan: Crohn's d isease pati ents' assess me nts of d isease ac ti vity, wit h respect to active and qui escen t d isease, t0 their physic ians' asse~smem s Li{ disease activ ity; and to use the exi~t ing    physicians on September 7, l 990.The purposes of the letter were to inform the ph ysic ian of ini t iation of the project and of an intentio n to contact thei r pa tients, and to ask for t he ir support in completing the physician assessment of their pat ie nt's disease.Each reg istry patient wa~ t hen sent a le tter t hat expla ined the study and incl uded a C OA i d iary, instructions for compl et ing the diary.and a pre-addressed, postage-paid envelope.T he patient diary was used to collect data over a one week period for computat ion of the C DAI (Table 8).

MATERIALS AND METHODS
A review of the Crohn \ disease index components c nahlcd the authors to design and test a composi re intervie w/ data collection instru me nt.T h i~ fo rm was used for patient inte rv iews and facili tated computation of the various ind ices (Table 9).Patient teleph one interviews bega n on September I 0, 1990 a nd we re completed on Ocroher 9, 1990.
After the patie nt had been interviewed, a three r an questionnaire was mailed to the patie nt's phys ic i,111 (Table I 0).Part I of chi~ questionna ire was rhe phys ic ian's as essmcnt of the patient's status related to the d isease.T h e COAi assessment categories of'very well', 'fair to poor', 'poor' and 'very pnor' were used.Part 2 was the physic ia n Cllnfi rmat io n of t he in flmn m,1 t,)ry bowel disease d iagnosis ,md the procedure used in maki ng t he d iagno~i~ (path o log ical, e ndoscop ic, rad iological, m cl inical).Part, w;:is darn relat ingrot hc patient's     Cohen's weighted kappa (Kw), a chance-cnrrecied coefficiem of agreemcnr for n,)minal scales, W'1S used ro evaluate the quality of agreement hetween the physician and pm icnt assessments of well-being.The Kw coefficient adjusb for the differential seriousness of misclassificat 10n errors.
Pearson prnduu moment correlation analysb was used ltl measure the relationship between the physician assessment of well-betng and the Crohn\ disease mJex of dbeasc severity.Because of rhc large numher of srmistical tests being performed, the usual alpha (0.05), was adjusted to 0.025 (Bonfcr-ron1 adjustment) tn reduce the probability of detecting spurious con-elauotb.
In many situatinm it is useful to have     Analysis was li mited to Crohn's disease patie n ts 111 t h e Infl ammatory Bowel Disca;,e Rcgbtry.

RESULTS
Three hundred and twenty patient;, of the 359 in rh c Inflammatory Bowel Disease Registry were contacted and interviewed hy one of the authors (LAF).Th ree hundred and seventeen completed the interview, and 2 19 returned a seven day diary (CDAI).Seventy-two physicians were repre;.cnrcdhy the 317 patients.The physicinm then ret urned 276 three part questionnaires.The physic .;npopulation consisted nf 37 internists, 28 gastroen terolngbts, and 12 colorectal surgeons.
Of the 2 l8 Crohn 's Jiscase patient~ in the registry, 171 (78.4%) completed the patient interview, 120 returned the seven day diary, and 171 corresponding physician assessments were ret urned to rhe authors' office.
Of the I 71 patient assessments of well-being, I 04 rcporred themselves as heing 'very well', 55 'fair tn gnod', eigh t 'poor', and fou r 'very poor'.Of l 7 1 physician assessments, I 08 reported their patients as heing 'very well', 56 'fair to good', seven 'poor', and none 'very poor'.;md 'fair tn good' categories for hoth physician and patient evaluations fell on or below the 7.8 cutoff point.ln the 'poor' and 'very poor' categories, the median plots are ahove the 7.8 cutoff point.
T h e OXI plots are given in Figure 3.An OX! score of 2. 7 corresponds to the COAi cutoff prnnr of 150.Physician evaluanon of disease state for patients in the 'very well ' and 'fair to good' categories are below the 2. 7 cutoff point.Patient evaluation in the 'fair to good' category is slightly above rhe cutoff point.
The OMGE and CTI (Figures 4,5) performed as expected, with the 'very well ' and 'fair to gooJ' box plot categories at or below the OMGE 4.94 cutoff point and the CTI 6.06 cutoff point.
The SM! and BR! indices are nearly identical ( Figures 6, 7).The cutoff points corresponding to that of the CDAI ( 150) ,ll'C 7.87 and 7.00, respectively.The hnx plots were as expecceJ.
In Figure~ 8 and 9, the cnnsistency of OMC,E (0. 771 ), clinical opininn and OXI (0.557), anJ OMOE and OX! (0.756), than were found in the present study.The differences het ween the corn.:latiunsarc likely due 10 the large number of physicians involvcJ in the pre~ent Sllldy and the known interobserver vmi:1hil1ty within physician assessment of pat ient well-hemg, abdominal pain ancl nhJrnninal mass.Another factor which coukl contri bute to the lower correlat iom 1:, the v.iriabil it yin health statu;, nf patient~ with inflammatory howel dbcasc.
One conclusion from the prc;,cnt study is clear.With any of the indices or when patient status i~ evaluated hy patient or physician, a consistent pattern between active and quiescent disease b eviJent.
The question of which Crohn's disease inJex is more re liable has not been answered.l lowever, given a poim in time either from the patient's or physi-cian\ per!>pect1\'e, consistent measures o( dis~•ase activity nre ava ilable. c: ::::, ::~;:.• :: ::y ~r;ea•• ~er: i r: da [ ta5form ~all 1~90) Complete one column of this form each day before going o bed.Your answer should describe the preceding 24 hours.Record cores in Q!'Oper column for each consecutive d~ (IBD) Registry survey -Form B: Crohn's disease activity indices physician survey (Fall 1990) Patient's name IBD Registry# _ _ _ _ _ _ Part l.Phys1c1an evaluation of patientGeneral well-being (l =Very well; 2=Folr to good: 3=Poor; 4=Very poor) the data (distribution).One simple method is the box plot.In the box p lot, rhe upper and lower quartiles of the darn arc portrayed hy the tnp and hottnm of a rccrangle, and the median is portni yed hy a
lnfnrmation letters were ma iled to

TABLE 7 St Marks Index of Crohn's disease activity (6,7) Clinical'
'Scoring: O=Absent, I =MIid with no limitation of activity; 2=Moderate with some limitation of activity: 3=Severe with considerable /Imitation or activities. 1 Scorlng: o to 2 (total possible laboratory score Is 8) C

Table 11
2 d isplay, the box r iots forthe I !Bl.An l lBI score ofle;.sthan l)r equal w 7.8 wa~ identified as the cutoff point corresponding ro the CDAl cutoff point nf 150 between active and quiescent disease.The med ian for the '\'cry well' (x2 4 l .68,P=0.0000I ) .The strength of rhe agreement ( Kw 0.2 322, P=0.0002) was also significant.Table12derails a complete correla-uon matrix between the Crohn's disea;.emdcx and physician•~ c l inical Judgement.The correlarinns range from Figure