INTI,RN.\110N \ I JOI RNAI

• 0 DE ,T-(j`•

Volume 66. I■7'1..n11ber 4Printedirillw

INTERNATIONAL(ISSN11148-916X11

hl;^ts,

JOU ANAL OF LEPROSYAnd Other Mycobacterial Diseases

VOLUME 66, NUMBER 4^

DECEMBER 1998

Sensitivity and Specificity of Methods of Classification ofLeprosy Without Use of Skin-Smear Examination'

Richard P. Croft, W. Cairns S. Smith, Peter Nicholls, and Jan H. Richardus2

The classitication of leprosy is closelylinked to chemotherapy as seen in the pre-multidrug therapy (MDT) era, when theRidley-Jopling system (") was used to de-termine the duration of dapsone therapy.The introduction of MDT in 1982 included

revised and simplified classitication intotwo groups, multibacillary (MB) and pau-cibacillary (PI3) ('). MB vias classitied us-ing skin-smear results xvhen the bacterialindex (BI) was .2+. This was revised bythe World Health Organization (WHO) EX-pert committee in 1988 (1 ') so that the find-ing of a positive smear at any site led toMB classification. In 1993 the WHO StudyGroup suggested that clinicai methodsalone could be used for classitication wherefacilities for skin smears were unavailableor unreliable ('').

' Reeeived for publication on 22 September 1998.Accepted for publication in revised forni 011 28 Octo-ber 1998.

= R. P. Croft, M.A., R.M., B.Ch.. M.R.C.G.P.,Danish BanrIzidesh Leprosy 1sion. Nil-

plutinari 5300, Bancladesh. W. C. S. Smith, M.D.,Ph.D., Professor and'11ead: P. Nicho! Is, M.Se.,Aberdeen,. Department of Public Ilealth, University ofAberdeen, Seotland. J. II. Riehardus, M.D.. P111).,Epidemiologist, Department of Public Ilealth, Eras-mus University, Rotterdam, The Netherlands.

Reprint requests to Dr. Richard R Croft. 70 Cul verLane, Reading RG6 11)Y, U.K.e-mail: richar;Wreroft. demon.comk

The qual ity of skin smears and of mi-croscopy has been found to be une of theweakest links in most leprosy control pro-grams ("). Becx-Bleumink lias drawn at-tention to the undue emphasis placed onskin-smear examinations, and in the con-clusion of her detailed analysis of leprosyclassitication at ALERT, Ethiopia, statedthat "... patients can be classilied based onclinicai signs and, hence, in the absence ofskin-smear services for routine classifica-tion purposes." (').

Between 1993 and 1997 some clinicalmethods were developed based upon thenumbers of skin lesions, nerve lesions andbody arcas affected, or combinations ofthese ("). In Bangladesh, the national pro-gram guideline classilied as N113 those caseswhose total number of skin patches and pal-pably enlarged nerves ("nerve lesions") is10 or more and/or the skin smear is posi-tive and classified as PB those cases wheretotal number of skin patches and nerve le-sions is fess than 10 and which are smearnegative (2). This system of classiticationwas in place until the end of 1997. TheBangladesh Technical Guide states: "... ah-sence of skin sulcar examination facilityshould not be considered as contra-indica-fluo for commencement of MDT in de-tected cases." Some leprosy control pra-granis in Bangladesh do not have access toa skin-smear examination facility and so

445

446^ International Journal of Leprosy^ 1998

they rely on the purely clinicai grounds de-scribed above to classify leprosy as MB orPB for treatment purposes, while others dohave access to such facilities and so rou-tinely use the skin-smear result as an ad-junct to classilication.

The WHO is now advocating the use of asimilar clinicai system to classify leprosythat may be used in the absence of a skin-smear examination facility. This systemclassifies patients having six or more skinlesions as MB and less than six skin lesionsas PB ("). This system of classification wasintroduced isto Bangladesh on 1 January1998.

The limitation of using a purely clinicaisystem for classifying leprosy cases is that apaucilesional smear-positive case will befalsely classified as PB and will receive in-adequate chemotherapy, with the attendantrisk of relapse. On the other hand, largenumbers of smear-negative patients wouldbe classified as MB and treated accordingly.This has to be weighed against the gains ofmaking MDT available to a larger popula-tion without the expense and complicationof a skin-smear examination service.

In this paper, a large data base of leprosycases is analyzed to determine the propor-tion of paucilesional smear-positive casesthat will be missed if classification is de-pendent on a count of skin/nerve fesionsalone. This study makes no use of histo-pathology in the classification of leprosy,but regards a classification system based onclinicai and bacteriological findings as the"gold standard." It is recognized that this isnot as sensitive as a system using skin andnerve biopsy. However, at the field levei itis not practical to use histopathological ser-vices for anything other than research. Thepossible impact on leprosy control is dis-cussed.

MATERIALS AND METHODSThe Danish-Bangladesh Leprosy Mis-

sion (DBLM) is a large leprosy controlproject covering four districts of northernBangladesh with a total population of 5.5million, an area known to have a highprevalence of leprosy DBLM followsthe Bangladesh national guidelines for lep-rosy classification and treatment as de-scribed earlier. AIl newly detected leprosycases had body charts drawn and slit-skin

TARLi? I. Sinear results at diagnosistisne of study group.

Average BI° Freyueney0 2473

233 204 445 406 14Unknown 28

Total 2664

BI = Bacterial index.

smears taken by leprosy control assistants(LCAs). Skin smears were taken from uneearlobe and two active skin patches, orfrom both earlobes and the forehead if aclear skin lesion could not be identified.Confìrmation of diagnosis and prescriptionof MDT are then made by an experiencedleprosy control supervisor (LCS) based onthe clinicai findings, classifying as PB orMB according to the protocol describedearlier. The skin smears were read at one ofDBLM's two laboratories in Nilphamari orThakurgaon, and the results posted back tothe field, where the classification waschanged if a "PB" case was found to besmear positive. The result entered was theaverage BI calculated by adding the resultsfrom each cite and dividing by 3. Fractionswere rounded up to the next whole integer.

The following nerves were routinely pal-pated by field staff and the degree of en-largement recorded: facial, supra-orbital,great auricular, supraclavicular, radial, ul-nar, mediar, radial cutaneous, ulnar cuta-

TABLE 2. Ricllev-Jopling classificationof study gro//p at diagnosis time.

Ridley-Jopling classification' Frequency49

TT 263BT 2087BB 18BL 84LL 51PN 112Total 2664

1 = indeterminate leprosy; TT = tuberculoid leprosy;BT = borderline tuberculoid leprosy; BB = ntid-border-line leprosy; BL= borderline lepromatous leprosy; LL =lepromatous leprosy; PN = pure neural leprosy.

66, 4^Cro.fi, et al.: Classification Without Skin Smears^447

TA ou: 3. Diagnostic criteria ler classi-lication systems used ia the absence afskin-smear residis.

Classilication system^Delinition

Bangladesh MB^

1() skin + nerve lesionsBan.gladesh PB^<I() skin + nerve lesionsW110 MB^skin lesionsW110 PB^<6 skin lesions

neous, common peroneal (lateral popliteal),infrapatellar, musculocutaneous, sumi andposterior tibial. Enlargement in any degreeol. one of these nerves was regarded as a"'terve lesion" and counted toward the totalnumber of lesions used in classifying as PBor MB.

DBLM lias a computer data base of aliits registered leprosy cases (26,000). Thecohort of cases registered between 1 April1995 and 31 March 1996 constitutes thestudy group for the Bangladesh AcuteNerve Damage Study (BANDS), a prospec-tive cohort study aimed at investigating theepidemiology and response to treatment ofacute 'terve damage in leprosy. The com-puter data base includes tields recording thenumber of skin patches and 'terve lesions aswell as smear results. This data base, whichhas been regularly audited, lias been ana-lyzed in a simple manner to show the dif-ference in the number of cases that wouldbe classitied as MB or PB, depending onwhether or not skin-smear results are takeninto account.

The computer data base is written in Mi-crosoft FoxPro for Windows, version 2.6.Statistical analysis was carried out usingEpi Info version 6.01.

RESULTSTable 1 shows the numbers of patients

with different Bis. A total of 163 patientswere smear positive, with a majority of the

TABU': 4. Numbers of patients classifiedusing the WHO method—skin lesion countsonly.

Skin^Sulcar^Smearlesions^positive^negative

145<6^

18

Total^

163

TARE". 5. Numbers of patients classifiedusing the Bangladesh method—skin and'terve lesion counts.

Skin and^

Sulcarnerve lesions^positive

>I() 150 281 431<10 13 2192 2205

Total 163 2473 2636

positive cases (98, 60.1c/o) having an aver-age BI of 4+ or more. A smear result wasnot available for 28 cases who were allsmall children for whom smear takingproved impossible in the clinicai setting.These 28 cases have been excluded fromthe study group, leaving a total of 2636 whowere analyzed using different mathematicaltilters. Table 2 shows the Ridley-Joplingclassitication of the patients at diagnosis,with the addition of the indeterminate (1)and pure neural (PN) groups recognized bythe Indian classification

The two different classification systemsare summarized in Table 3. Tables 4 and 5show the results of classifying the cases bythe WHO and Bangladesh classificationsystems without using smear results. Thesensitivity of the WHO system comparedwith the "gold standard" of smears was89.0% and the specificity was 88.1%. Usingthe Bangladesh system, the sensitivity was92.0% and the sensitivity was 88.6%. Theaverage BIs of these "missed" cases areshown in Table 6. Table 7 shows the citar-acteristics of the smear-positive cases as-signed PB by the WHO method.

TABU: 6. Bacterial indexes ef vincar-positive cases "Inissed" clas.rtficationsystems not using sineta- residis (false-neg-atives) (i.e., cases.falsely classified as P13).

BI"Missed" by Bangladesh

system not usingsulcar result

"Missed" by WHOsystem not using

sulcar result

I+ 5 82+ 5 53+ O o4+ 3 35+ O6+ O O

13^

18

2942179

2473

Smearnegative Total

Total

4392197

2636^Total

448^ International Jonrnul Lehrosv^ 1998

Tr\uI.1', 7. .mear-positive cases assignetias PR by the WH() nmthocl.

No. skin lesions No. cases Skin-sinear 131

O 4 I+, 1+, 2+, 5+4 I +, 2+, 4+, 4+

2 O3 5 1+, 1+, I+, 2+, 2+4 2 I+,^I+5 3 i +, 4+, 5+

Total 18 Mean 131 = 2.2+

DISCUSSiONThe above analysis shows a high degree

of concordance between the two methods ofclassification when smear results are used.Both methods classify approximately thesame number of cases as MB (WHO/MB =439, Bangladesh/MB = 431 out of the studygroup of 2636). However, the groups do notquite correspond. When the Bangladeshsystem is used for classification, 34 out of431 MB cases are classified as PB by theWHO system (7.9%, 95% Cl 6.6%-9.2%).Conversely, if the WHO system is used forclassification, 42 out of 439 MB cases areclassified as PB by the Bangladesh system(9.6%, 95% CI 8.2%—I 1.0%). Thus, eachsystem of classification excludes nearly10% of the patients who would be includedby the other system. Operationally, it is un-likely that there would be nutch effectwhichever system is used.

There is, however, a small number ofsmear-positive paucilesional cases whowere missed by both classification systemswhen the smear result was net used. Thereis a hard core of 13 cases which neither sys-tem of classification picked up as MB with-out the smear result, and an additional tivecases if the WHO system is used alone.Thus, 18 out of the total of 163 smear-posi-tive cases would be missed (11.0%, 95% CI6.2%-15.8%) by the WHO classificationsystem in the absence of smear results anddiagnosed as PB. When the Bangladeshsystem is used alone without smear results,13 out of the total of 163 smear-positivecases would have been missed (8.0%, 95%Ci 3.8%-12.1 %/o). The results indicate thatthe Bangladesh system of classification,when used without smear results, is slightlymore sensitive and specific than the WHO

system at detecting smear-positive cases,but the difference is not significam (p =0.345).

Out of the 18 cases who would havebeen falsely diagnosed as PB by applicationof the WHO criteria without using smearresults, four cases were recorded as havingno skin patches; these were alI LL. Therewere a total of nine lepromatous (BL/LL)cases in chis group. The body charts of alithese cases show tnarkings of inliltration,so potentially aIl could have been diag-nosed as lepromatous and, thus, MB byalert leprosy control workers or doctors.However, signs of infiltration are oftenmissed and sometiInes not noticed until thepositive smear result prompts a more thor-ough examination. A classification systemfocusing on patch counts alone is very opento Chis potencial oversight.

Groenen, et al. in a clinical/histopatho-logical study carried out in Bangladesh, con-cluded that skin-smear reading facilities canbe perfectly satisfactory, given reasonablesupervision ('). In the two hospitais inwhich their study was carried out (one wasDBLM Nilphamari), no positive slideswere missed and no BI readings differed bymore than one unit from the control in 95%of the cases. They also found that inclusionof skin-smear results would considerablyimprove the reliability of the classificationstrategy if the results were used optinutllyby improving both the specificity and thesensitivity of MB diagnosis ('),

in the DBLM project area, the propor-tion of smear-positive cases among the totalcaseload is not high (6.1 %), but remainsmoderately high among MB cases (37.8%of Bangladesh-classified MB cases). How-ever, in other parts of Bangladesh thesmear-positive rate among the total casesremains much higher (20%-30%), as inother countries. In such situations, this studyindicates that there is likely to be a consider-able number of paucilesional smear-positivecases who will be missed by a classificationsystem that relies on counting skin patches(and nerve) lesions alone, possibly resultingin a subsequent relapse.

A prospective study on the effectivenessand safety of WHO/MDT among a cohortof patients in Thailand has recently beenpublished ('). When the authors used apurely clinical classification system (>_6 le-

66, 4^Croft, et al.: Cla.swification Without Skin Smears^449

sions = MB), they found that 12% of the"true" MB patients would be undertreated.They defined MB as cases with clinicallydiagnosed LL, BL or BB features, or anyother cases with a BI of 2+ or more at anysite. Their MB classitication criteria weremore stringent than ours, but the percentaszeof "missed- (false-negative) paucilesionalMB cases is very similar to the 11% weffiund.

The issue of relapse continues to pro-voke discussion. Although a relapse atterMDT has generally been found to be lessthan 1%, Jamet, et al. found a 20% re-lapse rate among patients with a BI of 4+ ormore atter a long period of follow up. Re-cently, a bacteriological study found the in-cidence of viable Mrcebacterium lepraeextracted from the nerves and skin of 26MB leprosy patients released from MDT tobe alarmingly high (46% from nerves, 23%from skin) ( '"). Desikan has drawn attentionto the risk of relapse atter MDT, recom-mending that attempts must be made to im-prove the standard of field laboratoriesrather than closing them down, and sug-gesting that particular attention should bepaid to the issue of MB cases with a high BIat diagnosis (-7)•

The potential problems of relapse amongMB patients with high Bls is likely to bemagnified if they only receive PB/MDT.Classitication of leprosy using clinicaimethods without the use of skin-smear ex-amination is less than ideal. Continued useof skin smears is recommended where theyare being reli ably done. In regi uns where nobacteriological services are available or ofunreliable quality, then both the Bangladeshand WHO methods can be considered. TheBangladesh is marginally more specitic andmore sensitive than the WHO approach, butthe latter is much simpler. The WHO sys-tem provides a reasonable balance of sensi-tivity and speciticity. II would be difficult toimprove on one without a deterioration inthe other. However, the results may be dif-ferent if applied to different populations.

Althouch the WHO system of classify-ing leprosy cases as MB is simple to applyand lias a reasonable balance between sen-sitivity and specificity, it must be recog-nized that the system will lead to a sinal!but significant number of MB cases beingtreated with a PI3 regimen.

SUMMARYA 12-month cohort of 2664 new leprosy

cases in BansJadesh has been analyzed toprovide information about the sensitivityand specificity of two different methods ofclassifying leprosy into paucibacillary (PB)and multibacillary (MB), if the results ofskin-smear examination are not taken intoaccount. The two methods are: 1) a proce-dure based on counting skin lesions recom-mended by the World Health Organization(WHO) (<6 skin lesions = PB, skin le-sions = MB); and 2) the "Bangladeshmethod" (<10 skin and nerve lesions = PB,

10 skin and nerve lesions = MB). In thelatter system, any degree of nerve enlarge-ment is taken to be a nerve lesion.

The WHO method was found to be 89%sensitive and 88% specitic at detectingsmear-positive MB cases from among thecohort; the Bangladesh system, 92% sensi-tive and 88.6% specitic. The WHO methoddid not detect 18 smear-positive cases asMN; the Bangladesh method lett 13 smear-positive cases unclassitied as MB. Severalof these "missed" (false-negative) cases hada hia bacterial index.

The WHO system of classifying leprosycases as MB is simple to apply and lias areasonable balance between sensitivity andspecificity. However, it must be recog-nized that the system will iead to a smallbut significant number of skin-smear-posi-tive MB cases being treated with a PBtreatment regimen.

RESUMENSe estudiaron 2664 casos nuevos de lepra eu

Bangladesh para determinar la sensibilidad y especili-cidad de dos métodos de clasilicación de la lepra comopaucibacilar (PB) o multibacilar (MB) que no tomaneu cuenta los exámenes baciloscópicos. Los dos méto-dos sou: ) nu procedimiento recomendado por la Or-ganización Mundial de ia Salud (OMS) basado eu la(menta de las lesiones eu ia piei (<6 lesiones = PB,lesiones = MB); y 2) "el método Bangladesh- (<10 le-siones eu piei y nervios = PB, O lesiones eu piei ynervios = MB). Eu el último método, cualquier gradode afección nerviosa se toma conto

Se encontro que ia sensibilidad (lei método de iaOMS fue dei 89% y su especilicidad dei 88% (ai hacerreferencia a los (latos bacteriológicos); el sistemaBangladesh tuvo una sensibilidad (lei 92% y una es-pecilicidad dei 88.6%. El método de ia OMS no de-tectó IS casos bacteriológicamente positivos; elmétodo Bangladesh dei!) lucra I 3 casos bacteriológi-

450^ International Jotanal ofLe/)ro.n_V^ 1998

camcntc positivos. Varios de estos casos "falsos nega-tivos'' tuvieron indices bacteriológicos correspondi-entes a casos MB.

El sistema de la OMS es simple de aplicar y está ra-zonablentente balanceado cri scnsibilidad y especifici-dad. Sin embargo, dchc reconocerse que la aplicación deeste método puede ocasionar que un pequeno poro sig-nificativo número dr. casos M13 scan tratados como PB.

RÉSUMÉUne cohorte de 2664 nouveaux cas de Ièprc au

Bangladesh, sur une périodc de douze mois, a étéanalyséc dans le hut d'ohtenir des renscignements surla scnsihilité et la spécificité dr. deux différentcs mé-thodes dr. classilication dr. la lèpre: paucihacillaire (P13)versus nmltuhacillarre (MB), quand 1'examen du suedermique n'cst pas pris cn corri 1_es deux méthodessom les suivantes: 1) une procédurc haséc sur Ie cle-nombrement des lésions cutanées rccon unendée par1'Organisation Mondiale de la Santé (OMS) (<6 lésionscutanécs = PB. ?6 lésions cutanécs = MB); et 2) la"méthode hangladaise" (<10 lésioncs eutanées etncrvcuscs = P13. _10 lésioncs cutanécs et nerveuses =M13). Selon cc dernier systèmc, tout dcgré d'épaississe-ment (ruo ncrf est considéré comine lésion ncntuse.

Cettc étude a montré que la méthode dr. 1'OMSprésentait une sensitivité dr. 89% et une spécificité dr.88% pour la détection des cas MB positifs à l'examendu sue dermique. Le systèmc bangladais pour sa partétait sensiblc à 92%/e et spéciliquc à 88.6(4 . La méthodedr. l'OMS n'a pas détccté 18 cas dr. Ièprc MB positifsau test du sue dermique; la méthode bangladaise n'apas permis de classer comine MB 13 cas positifs au testdu sue dermique. Plusieurs dr. ces cas "manqués" (f ,u txnégatifs) présentaient des index bactériens élevés.

La système de classilication des cas de Ièprc dr.!'OMS est simple à appliquer et présente un équilihrcraisonnable entre scnstivité et spécificité. Cependant, ilfaut raconnaítre que ce nouveau systèmc aura comineconséquence la présencc d'un nomhre réduit, mais sig-nificalif, dr. cas MB positifs au test du sue dermique quiseront traités avec un traitemcnt desliné aux paticnts PB.

REFERENCES1. Br:Cx-BLEUxuNR, M. Allocation of paticnts to PB

or MB drug regimens for the treatment of leprosy:a comparison of methods hascd mainly on skinsmears as opposed to clinicai methods; alternativeclinicai methods for classilication of paticnts. Int.J. Lcpr. 59 (1991) 292-303.

2. DASANANIAI.1, K., ScnRr.uot:R, P. A. M. and Pi-RAVAVARAt'ORN, C. A study on the effectivenessand safety of the WHO/MDT regimen in the

Northeast of Thailand; a prospeclive study, 1984-1996. Int. J. Lcpr. 65 (1997) 28-36.

3. Dr:siKAN, K. V. 'hhe risk of relapse alter multidrugthcrapy in leprosy. (Editorial). Lcpr. Rev. 68(1997) 114-116.

4. GROn:Ni:N, G., SAHA, N. G., RAsnn), M. A., HAntu),M. A. and PirnyN, S. R. Classilication of leprosycases under field conditions in Bangladesh. 1. Use-fulness of skin-smear cxaminations. Lcpr. Rev. 66(1995) 126-133.

5. Gttor:Nr:N, G., SAHA, N. G., RAsnu), M. A., HAntn),M. A. and PArrYN, S. R. Classification of leprosycases under field conditions in Bangladesh. Il. Re-liahility of clinical enterra. Lepr. Rev. 66 (1995)134-143.

6. HAs -rtNCS, R. C., r.D. L<'pIos.); Edinhurgh: ChurchillLivingstone, 1985.

7. JAmu.r, P., Jr. 13. and rHr: MARCUOUx Cnr:nun in.tt-^tw Slt n )Y GROn'. Relapse after lona terra füllowup of MB palients treated hy WHO unultidrug reg-ímen. Int. J. Lcpr. 63 (1995) 195-201.

8. Rir. it:\RDt s, 1.11. and Citou. R. P. Estimaling thesize of lhe leprosy prohlem: the Bangladesh expe-rience. Lepr. Rev. 66 (1995) 158-164.

9. Ru)t.ES, D. S. and Jotm.iNC. W. H. Classilication ofleprosy according to inununity; a tive-group sys-tem. Int. J. Lepr. 54 (1966) 255-273.

10. SHt rrv, V. R, Sucuri - os, K., Ui't.ra<Ait, M. W. andANTIA, N. H. Higher incidente of Rlvcohuclerium

leprae within the verve as compareci to skinamong MB leprosy paticnts released from mul-tidrug thcrapy. Lcpr. Rev. 68 (1997) 131-138.

1 I. SAN BR,ntit_t., W. H., De Sot.Do_NUOrr, R. and Mc-Dot x„u.t., A. C. The allocation of leprosy paticntsfinto paucihacillary and nmltihacillary groups fornwltidrug thcrapy, taking mio account the 'mulherof body arcas affected hy skin. or skin and nervelesions. Lepr. Rev. 63 (1992) 231-246.

12. TB AND Lra'Rosv CONTROL. St:Rvicts. Technicalguide and operacional manual for leprosy controlin Bangladesh. Dhaka: Government of Bangladesh,1995.

13. WHO ExPGRT COnuurrrnt. ON Li_t'ROSV. Sixth re-

port. Geneva: World Health Organization, 1988.Tech. Rep. Ser. 768.

14. WII0 Ext>rarr COntntrrrEi ON Lr[ritosv. Seventhreport. Geneva: World Health Organization, 1988.Tech. Rep. Ser. 874.

15. WHO STI:DY GROtct'. Chemotherapy of leprosy.Geneva: World 1lcalth Organization, 1994. Tcch.Rep. Ser. 847.

16. WHO Srt:DV GROUis Chenu )therapy of leprosyfor control progrannncs. Geneva: World HcalthOrganization, 1982. Tech. Rep. Ser. 675.