INTERNATIONAL JOURNAL OF LEPROSY^ Volume 55, Number 3Printed in the U.S.A.

Certainty Levels in the Diagnosis of Leprosy'Jorg M. Ponnighaus, Paul E. M. Fine, and Lyn Bliss 2

Of fundamental importance in any studyis a clear definition of a "case" of the diseaseunder investigation. The more difficult andcontroversial the diagnosis, and the morevariation there may be in methods of as-certainment, the more important is this stepof clarifying case criteria.

Ideally, a case definition should state ex-plicitly all criteria which lead to the inclu-sion of a particular individual into a definedcategory. Although a case definition shouldbe clear and acceptable to experts in thefield, it need not necessarily be universallyapplicable. This is true insofar as the criteriaused to define a case will have to suit localcircumstances and will depend on the typesof information and staffavailable. Case def-initions may also differ according to theirpurpose. For example, a disease controlprogram must consider the implications ofits case definition for decisions concerningtreatment, whereas this may not be of majorconcern for an epidemiological researchproject. In the latter circumstance, diagnos-tic specificity is often of particular impor-tance. The inclusion of false-positives bias-es relative risk measures towards unity,thereby obscuring risk factors under study(2. [1) .

Recognition of the important implica-tions of case definitions has led recently toattempts to assign levels of certainty to thediagnosis of a number of different diseases.For example, a recent study of childhoodinfections in Kenya considered the diag-nosis of pertussis as "unlikely," "possible,""likely," or "definite" on the basis of a scor-ing system ( 13 ). And the Expanded Pro-gramme of Immunization has producedguidelines according to which a diagnosis of

' Received for publication on 22 December 1986;accepted for publication in revised form on 24 March1987.

= J. M. Ponnighaus, Dr.Med., D.T.P.H., Lepra Eval-uation Project, P.O. Box 46, Chilumba, Karonga Dis-trict, Malawi. P. E. M. Fine, V.M.D., Ph.D., and L.Bliss, Ross Institute, London School of Hygiene andTropical Medicine, Keppel Street, London WC I E 7HT,England.

Reprint requests to Dr. Fine.

measles, neonatal tetanus, polio, or child-hood tuberculosis may he graded as "sus-pect," "probable," or "certain" ( 3 ).

It is widely recognized that the diagnosisof leprosy is often difficult. Given this dif-ficulty, it is surprising that few publicationsin the leprosy field specify precisely the cri-teria upon which diagnoses are made. It haseven been commented that the absence ofclearly stated case definitions calls intoquestion much of the leprosy literature in-sofar as it renders results uncomparable andunreproducible (4 ' 8 ).

We have become acutely aware of the dif-ficulty of defining a "case" of leprosy in thecourse of the Lepra Evaluation Project(LEP), a total population epidemiologicalstudy in Karonga District, Northern Malawi( 10). The clinical evidence was often toomeager for the diagnosis of leprosy to beestablished, and yet too suggestive for it tobe precluded. Keeping unresolved "sus-pects" on observation for long periodsproved to be expensive and impractical dueto frequent changes of residence. Althoughit was hoped initially that histopathologicalinformation could resolve uncertain diag-noses, it was found that many histopatho-logical results were themselves inconclu-sive. Findings were often "compatible" withclinical leprosy but by no means conclusive,and might in fact have been due to someother skin disease ('). In addition, we be-came increasingly aware of the problem offalse-negative histopathology results in in-dividuals whose clinical findings had beenoverwhelming or in suspects whose follow-up showed clear signs of leprosy both clin-ically and histopathologically.

These experiences have led us to developa procedure for grading the degree of con-fidence with which it can be held that adiagnosis of leprosy is in fact correct. Ac-cording to this system, individuals in whomclinical leprosy is suspected are allocated toone of four groups: a) a "narrow" (certainleprosy) group in which the diagnosis of lep-rosy should be almost invariably (>99%)correct; b) a "middle" (probably leprosy)

454

55, 3^Ponnighaus, et al.: Certainty Levels in Diagnosis^455

TABLE 1. Criteria for assignment of levels of clinical certainty.

Typical skin lesion(s)Definite

anesthesiain lesion(s)

Definite nerveenlargement(no historyof trauma)

Typicalsequelae of(leprosy)

neuropathy

Clinicalcertainty

Code

YesYes (or slightly less than typi-

cal)

Yes—

YesYes

—"Yes Certain 5

YesYesYes—Yes (on face)

Yes——

—

—Yes—Yes—

——YesYes—

Extremelylikely

4

YesSlightly less than typical—

—Yes—

——Yes

———

Most likely 3

Less than typicalUntypical—

—Yes—

———

——Yes

To be con-sideredseriously

2

Untypical lesions for which noother diagnosis can be made

—

—

—

—

Slight nerveenlarge-ment

—

—Possibly 1

(—) indicates that the sign is not clearly present, or not present at all.

group in which we expect that most of theindividuals included are or were actual casesof clinical leprosy; c) a "wide" (possibly lep-rosy) group which may contain only a smallproportion of individuals with actual clin-ical leprosy; and d) an "out" (not leprosy)group, in which the initial suspicion of lep-rosy is discarded.

Our method for assignment of these over-all levels of certainty in the diagnosis of lep-rosy is presented in this paper.

SOURCES OF INFORMATIONThe basic aims and procedures of the

Lepra Evaluation Project (LEP) are de-scribed elsewhere ("). For the purposes ofthis presentation, it may be noted that theproject began in 1979 as a total populationepidemiological survey in Karonga District,Northern Malawi. In the course of the sur-vey, 112,000 individuals were examined forsigns of leprosy by paramedical LeprosyControl Assistants (LCAs). All newly foundsuspects were seen also by a Medical Officer(JMP or, on a few occasions, Dr. Gjalt Boer-rigter). Slit-skin smears were taken by theLCAs, and biopsies were taken by the Med-ical Officer. In addition, the LEP had accessto all records of cases who had been treated

by the Lepra Control Project (LCP), whichhad been in the district since 1973 ('). Thus,there were several types of information onwhich to base the overall certainty of a di-agnosis of leprosy:

Levels of clinical certainty. When exam-ining a suspect, the Medical Officer assigneda code describing the strength of the clinicalevidence in favor of leprosy. The criteria forassigning these levels of clinical certaintyare given in Table 1. For example, the clin-ical certainty was considered "extremelylikely" i f there was: a) a skin lesion of typicalappearance (for paucibacillary leprosy) anddefinite anesthesia to light touch within thelesion; or b) a skin lesion of typical appear-ance without evidence of anesthesia but witha definitely enlarged nerve (near or far fromthe lesion); or c) a skin lesion of typical ap-pearance without evidence of anesthesia ornerve enlargement, but in a person with typ-ical sequelae of leprosy neuropathy; or d) adefinitely enlarged nerve together with signsof damage to that nerve; or e) a skin lesionof typical appearance (for paucibacillaryleprosy) without evidence of anesthesia buton the face.

No clinical certainty level was assigned ifa positive (bacterial index [131] >2 from at

456^ International Journal of Leprosy^ 1987

TABLE 2. Protocol for coding of histo-pathological certainty as used in the LepraEvaluation Project."

Diag-nosis^Meaningcode

"Leprosy confirmed beyond reasonabledoubt" (or "almost certain but slightelement of doubt remains").

2^"Consistent with but not diagnostic of lep-rosy" (or "pathological and possibly dueto leprosy, but lacking specific diagnos-tic criteria").

3^"Definitely pathological but completelynonspecific," or "normal or near normaltissue."

4^"Pathological but indicative of specificdisease other than leprosy."

This is an abbreviated version of the protocol whichhas been described in full in references 5 and 7.

least one site) slit-skin smear result was al-ready known at the time of examination. Inaddition, no clinical certainty level was as-signed to patients who were currently or pre-viously under treatment by the LCP, unlessthere was a suspicion of relapse. In suchcases the assigned level of clinical certaintyreferred to the relapse, and not to the orig-inal diagnosis. Patients who had receivedtreatment elsewhere, that is, by an institu-tion or service other than the LCP, wereassigned a level of clinical certainty only ifthere were sufficient findings on which tobase it. Since such patients were not rou-tinely examined by the Medical Officer, theirclinical certainty was often assigned retro-spectively, on the basis of findings reportedby the LCAs. This was the only circum-stance in which a clinical certainty was as-signed without the patient being examinedin person by the Medical Officer.

Slit-skin smear results. Slit-skin smearswere taken by the LCAs or the Medical Of-ficer if there was any suspicion of multi-bacillary leprosy ( 1 "). Two smears were usu-ally taken from the earlobes and two fromthe lesion or lesions. The results were codedboth as average BI (i 2) and as percentagesof solids, fragments, and granules.

Histopathology results. Skin biopsies wereobtained by the Medical Officer from >90%of all leprosy suspects newly found in thecourse of the LEP after 1980. These wereroutinely taken using a 4-mm punch (Steil-

TABLE 3. Agreement code used when re-viewing Lepra Control Project records.

Code^Meaning

I^Agree with original diagnosis of leprosy.2^Original diagnosis doubtful.3^Original diagnosis of leprosy very unlikely.4^Unknown (insufficient information avail-

able).

fel Laboratories) from the most active partof a lesion ( 7 ). Occasionally, a split-nervebiopsy was obtained if the only sign of lep-rosy was an enlarged peripheral sensorynerve. If a biopsy was not taken, the reasonfor its omission was recorded. Biopsies werealso obtained from suspected relapses, butonly very occasionally from patients whilestill on treatment. In the latter cases, thebiopsies were usually taken because it wassuspected that the original diagnosis hadbeen wrong. The processing of these biop-sies is described elsewhere ( 7 ). The histo-pathologist's report was provided in stan-dard form as shown in Table 2.

Repeat biopsies were taken if an initialnegative or inconclusive histopathology re-sult contrasted markedly with the level ofclinical certainty and antileprosy treatmenthad not yet been initiated on the basis ofthe clinical findings. Repeat biopsies werealso taken if an initial negative or inconclu-sive biopsy result was consistent with theassigned level of clinical certainty but theindividual self-reported or was found laterwith additional signs suggestive of leprosy.

Lepra Control Project (LCP) records. TheLCP records contained clinical findings,classification, and smear results (smearshaving been taken only if multibacillary lep-rosy had been suspected) on all cases reg-istered by the LCP since it began in thedistrict in 1973. Many also contained re-view notes and additional comments by theLCP Medical Officer (Dr. Gjalt Boerrigter).These records were reviewed by JMP, anda retrospective "agreement code" was as-signed with reference to the original diag-nosis of each case. These codes and criteriaare given in Table 3.

History given by patient. All individualswith clinical signs of leprosy were asked forany history of antileprosy treatment. In ad-dition, a number of individuals without signs

55, 3^Ponnighaus, et al.: Certainty Levels in Diagnosis^457

volunteered such information. All reportedplaces of prior treatment were recorded, andthe most reputable such institution was cod-ed.

All of this information was recorded onspecially designed forms, coded, and ana-lyzed on computers at the London Schoolof Hygiene and Tropical Medicine. Severaldi fferent methods of amalgamating the vari-ables were explored over a period of 4 years.The procedure ultimately developed is pre-sented below.

ASSIGNMENT OFCERTAINTY LEVELS

The Figure is a flowchart illustrating thesteps used (by the computer) to assign eachsuspect—excluding relapses—to the "nar-row," "middle," "wide," or "out" group.The procedure for relapses is identical, ex-cept that it neglects all information collectedprior to the date when the relapse was firstsuspected. Although the basic logic shouldbe clear from the flowchart, it is describedbriefly below. Numbers on The Figure referto decision points as described in the text.

Any LCP or LEP slit-skin smear resultwith an average BI >1 places the individualautomatically in the narrow (N) group (point1). In the absence of slit-skin smear results,or if the average BI is 1, histopathologyresults are then considered. First, any his-topathology result with a code 1 (Table 2)places the individual in the narrow (N) group(point 2). If an individual has no code 1biopsy results, the computer searches forany histopathology result with conclusiveevidence solely of another skin disease (Ta-ble 2, code 4). If found, the individual isassigned to the out (0) group (point 3). Ifnot, but there are two or more nonspecificbiopsy results in the presence of some clin-ical activity, then the individual is still as-signed to the out (0) group (point 4).

The logic of interpreting biopsy and clin-ical results then differs according to whetheror not the individual was on treatment atthe time the biopsy was taken (points 5, 6,and 7). Any individuals on treatment withnonspecific biopsy results (Table 2, code 3)are assigned according to the review of theirrecords (point 8).

Individuals from whom no biopsy wastaken, or from whom the biopsy was un-

satisfactory, are considered separately, asshown on the left side of The Figure. If theindividual was known to the LCP, then theclinical certainty is assigned on variouscombinations of current clinical signs andrecord review (points 9, 10, and 11). It willbe noted that the reason for a biopsy nothaving been taken from inactive cases isconsidered here—if the biopsy was consid-ered unnecessary because of typical seque-lae of neuropathy then the overall certaintyis higher than if there was no residual evi-dence or only inactive lesions (point 12).

Lastly, there is a group lacking biopsiesand not previously known to the LCP, inwhich case the overall certainty is assignedon the basis of clinical signs, if present (point13), or on history of past treatment, if clin-ical signs are not present (point 14).

ILLUSTRATIONTable 4 shows a breakdown of the diag-

nostic certainties assigned to 2292 leprosypatients and suspects ascertained in the firstLEP survey, illustrating the frequency of al-locations at each point in the decision tree(The Figure). Relapses are excluded fromthis tabulation. Of the 2292 suspects, 1043(45.5%) were allocated into the narrowgroup. The main reasons for this allocationwere definite histopathological evidence ofleprosy (410, 39.3%); typical sequelae ofleprosy neuropathy (302, 28.9%); strongclinical evidence alone (167, 16.0%); andbacteriological evidence of leprosy (107,10.3%). Of the 167 who were graded onclinical grounds only, 67 were previouslytreated and thus not new suspects, while 100were newly found suspects. The majority(57) of these 100 new suspects were foundby the LEP in 1979 and 1980, before it wasattempted to take biopsies as a matter ofroutine. The majority (455 out of 744) ofindividuals allocated to the middle groupwere registered patients with no remainingevidence of leprosy and no history of a pos-itive slit-skin smear (decision point 10).These patients are discussed further below.

VALIDATIONValidation of diagnostic criteria such as

are described here presents a difficult prob-lem, insofar as there is no fully reliable stan-dard against which they can be assessed. On

55, 3^Ponnighaus, et al.: Certainty Levels in Diagnosis^459

TABLE 4. Breakdown in' certainty level and organogram decision point (The Figure) of2292 individuals in whom leprosy had been diagnosed before—or was suspected during—the first Lepra Evaluation Project survey.

Decisionpoint Description

Certainty levelTotal %

N M W 0

I Average 131 > 1 107 — — — 107 4.72 Conclusive biopsy evidence 410 — — — 410 17.93 Other skin disease — — — 14 14 0.64 Two negative biopsy results — — — 2 2 0.15 Inconclusive biopsy result 26 75 33 0 134 5.86 No biopsy evidence of leprosy 2 25 142 93 112 9.77 Inconclusive biopsy, on treat-

mem0 — — — 0

8 No biopsy evidence, on treat-ment

2 — 0 0 2_ 0.1

9 Registered cases, diagnosisdoubtful

— — 114 4 118 5.1

10 Registered cases, no signs left — 455 — — 455 19.811 On treatment, still active 27 — — — 27 1.212 Inactive lesions and/or segue-

lae of ncuropathy left302 143 — — 445 19.4

13 Clinical certainty grading only 167 46 18 17 248 10.814 History of treatment — 50 58 108 4.7

Totals 1043 744 317 188 2292(45.5%) (32.5%) (13.8%) (8.2%) (100%)

the other hand, the validity of the certaintylevels can be tested in relation to a specificrisk factor which is recognized to be asso-ciated with leprosy. Table 5 shows the pro-tective efficacy of BCG against N, M, W,and 0 cases, as derived by case control anal-yses by methods described elsewhere ( 6). Thelower the specificity, i.e., the lower the per-centage of true cases in a certainty group,the lower is the observed protection im-parted by BCG (x=, for trend = 9.24; p <0.005).

DISCUSSIONThe procedure described in this paper il-

lustrates one way of tackling the difficultproblem of diagnostic criteria for leprosy.The complexity of the method reflects thedifficulty of diagnosing leprosy and the needto take into consideration several differenttypes of information. In our case, it alsoreflects the availability of a considerableamount of relevant information on corn-

puler files where it can easily be accessedand analyzed.

The validity of the method can be sup-ported both on intuitive grounds and byrelating cases classified into different groupsto a recognized risk factor. The correlationbetween protective efficacy of BCG andoverall certainty level (Table 5) providessupportive evidence that the certainty levelis a direct reflection of diagnostic specificity( 2 " ). The lower vaccine efficacies in lowercertainty groups occur insofar as BCG's ef-fect is against leprosy per se, and not againstother conditions which arc included in in-creasing proportions in the M, W, and 0groups. These results also suggest that thedifference in specificity between successivegroups is not uniform. The greatest differ-ence in vaccine efficacy occurs between themiddle (M) and wide (W) groups. This in-dicates that the specificity of the middlegroup is quite high and that of the widegroup, quite low. More direct validation

4-

THE FIGURE. Organogram showing how individuals are allocated into "narrow" (N), "middle" (M), "wide"(W), and "not leprosy" (0) groups. Numbers indicate decision points discussed in the text.

460^ International Journal of Leprosy^ 1987

TABLE 5.^Efficacy of BCG in protectingagainst clinical leprosy, according to level ofcertainty of the diagnosis."

Cer-taintygroup

No.'casesProtection imparted by BCG

Estimate 95(ii) Conf. Mt."

N 213 37% 16% to 53%M 47 31% —36% to 65%W 58 2% —122% to 56%0 47 —4% —130% to 53%

Estimates derived by stratified (for age, sex, andschooling status) case control analysis as described inreference 6. Cases restricted to those registered duringfirst Lepra Evaluation Project survey, after 1 January1980. Controls matched for age, sex, and schoolingstatus.

'' 95% Confidence interval.

would require some alternative—and not yetavailable— test of proven and very highspecificity.

Although intuitively reasonable, thestructure and criteria of this system are ar-bitrary. There are several decision points atwhich a different allocation might be used.For example, one might argue that the:`nar-row" group includes subsets with varyingstrengths of evidence. An average BI > 1 onits own may seem to be stronger evidenceof leprosy than either a histopathology re-sult with a code 1 or an "extremely likely"clinical certainty level associated with aninconclusive biopsy result. The latter cri-teria might permit a small number of non-leprosy cases to be included in the "narrow"group. However, the procedure describedhere was decided upon so as not to sacrificetoo much sensitivity for only a slight in-crease in specificity.

It may be noted that most of the patientsknown to the LCP were allocated into the"middle" or "narrow" groups solely on thebasis of whether or not they had typical se-quelae of neuropathy at the time of exam-ination by LEP staff (The Figure, points 10-12, and Table 4). We have found it difficultto define criteria according to which treatedpatients without typical sequelae of leprosyneuropathy can be allocated reliably to eitherthe "narrow" or the "middle" group. Theassignment of an agreement code to theoriginal diagnosis was often difficult and lessreproducible than we would wish. We sus-pect that this will be a problem when re-

viewing case records of most leprosy controlprograms.

It may be noted that a finding of typicalsequelae of neuropathy on its own wasweighed differently, depending upon wheth-er the individual concerned was a newlyfound suspect or a known leprosy patient.In known patients, such sequelae were con-sidered sufficient for a narrow group allo-cation if the individual had a prior crediblehistory of leprosy (decision points 9 and 12).In newly found suspects, on the other hand,the same sequelae were grounds for a clin-ical certainty grade of 2 ("to be consideredseriously," Table 1) only. This reflects ourview that sequelae of neuropathy in the ab-sence of enlarged nerves and in the absenceof a history of antileprosy treatment canarise from other causes (e.g., trauma) andshould not, on their own, be sufficient forM or N overall certainty levels (decisionpoint 13).

It should be evident from the allocationprocedure that the overall diagnostic cer-tainty does not reflect simply current clin-ical signs of leprosy. This is important in-sofar as it means that the cases so definedcannot be translated directly into currentprevalence rates ofclinical or active leprosy,let alone of infection with Mycobacteriumleprae. We have discussed the implicationsof diagnostic certainty on the assessment ofleprosy prevalence in a separate publication(9 ).

Although scoring systems have been usedas an alternative to the flowchart approachin assigning diagnostic certainty for somediseases, we found such methods to havetwo disadvantages in this context. Oneproblem arose because of the dependencebetween variables. For example, it seemedreasonable to weigh the clinical certaintygrade differently dependent upon other, e.g.,histopathological, information (e.g., points5 and 6). Although such assumptions canbe handled numerically, it makes a scoringsystem impracticably complicated. Fur-thermore, the flowchart representationmakes the procedure's logic explicit. Giventhe complexity of the problem, we find thispreferable to the implicit logic of a numer-ical scoring method.

It is difficult to avoid terminological dif-ficulties when discussing a problem such as

55, 3^Ponnighaus, ct al.: Certainty Levels in Diagnosis^461

this. In particular, it may be pointed outthat the term "certainty" has been used inthis paper in three distinct contexts: a) withreference to the clinician's diagnosis, b) withreference to the histopathologist's diagno-sis, and c) with reference to the aggregate ofall relevant information. This implicitlyrecognizes that all the evidence and opin-ions relating to a diagnosis of leprosy neednot agree, but allows us nevertheless to ar-rive objectively at an overall decision as tothe status of each individual. It is this ag-gregate or overall decision which is then usedfor determining treatment and/or for epi-demiological analysis.

It should be emphasized that this is notpresented as a universal solution to theproblem of defining a case of leprosy. Theform and content of the data upon whichour procedure is based are probably unique.However, analogous circumstances arefound in many leprosy-endemic areas andresearch projects, and the general approachdescribed here could be modified to fit mostsituations.

SUMMARYThis paper describes a procedure for grad-

ing the degree of confidence with which itcan be held that a diagnosis of leprosy is infact correct, after considering all availableclinical, historical, bacteriological, and his-topathological information. Individual sus-pects are assigned to one of four categoriescorresponding to different levels of overallcertainty of the diagnosis. The method isillustrated using data from the Lepra Eval-uation Project in Northern Malawi, and val-idated in the context ofan analysis of BCG'sprotective efficacy against clinical leprosy.Although the procedures described in thispaper were designed for a specific epide-miological study, the method could beadapted for use in most leprosy research orcontrol programs.

RESUMENEste trabajo describe un procedimiento para valorar

el grado de confianza con el cual se puede sostener queun diagnestico de lepra es correcto. El procedimientoconsidera Ia información clinica, histológica, bacterio-legica, e histopatológica. Los individuos sospechososson asignados a una de 4 categorias que correspondena difcrcntcs niveles de certidumbre en el diagnástico.

El método es ilustrado usando datos del Proyecto deEvaluation de la Lepra en Malawi del Norte, y es valid°aplicado al andlisis de la eficiencia protectora del BCGcontra Ia lepra clinica. Aunquc el procedimiento des-crito en este trabajo se discno para un estudio epide-miolOgico especifico, el metodo se puede adaptar parausarse en la mayoria de los programas de investigaciOno control de Ia lepra.

RESUMEOn décrit ici une méthode pour renforcer le degré

de confiancc que l'on pout accorder a une diagnosticde lêpre, en considerant touts l' information disponiblesur Ics plan cliniquc, historique, bactériologique, et his-to-pathologique. Les sujets soupconnés d'être maladesont été divisés en quatre categories, qui correspondentaux dia'.rents niveaux de certitude global du diagnos-tic. La méthode est illustrée en utilisant les donnéesrecueillies dans le Projet d'Evaluation men& au Malawidu Nord; it a Le valid& dans le contexte (rune analysedu pouvoir protecteur du BCG contre la lêpre clinique.Quoique Ics procedures décrites aient été établies dansle cadre dune etude épidémiologique spécilique, la me-thode pourrait etre adaptée et utilisée dans la plupartdes programmes de recherche et de lutte contre Ia lêpre.

Acknowledgments. This work arose as part of theLepra Evaluation Project and was funded by LEPRA,the British Leprosy Relief Association. The authorswish to thank Dr. Gjalt Bocrrigtcr and Mr. MartinMathews for valuable discussions bearing upon thematerial presented in this paper, and Ms Judith Russellfor preparation of the manuscript.

REFERENCES1. BOERRIGTER, G. and PONNIGHAUS, J. M. Ten years'

leprosy control work in Malawi (Central Africa)—I. Methods and outcome after treatment. Lepr.Rev. 57 (1986) 199-219.

2. COPELAND, K. T., CHECKOWAY, H., MCMICHAEL,

A. J. and HOLBROOK, R. H. Bias due to misclas-sification in the estimation of relative risk. Am. J.Epidemiol. 105 (1977) 488-495.

3. EXPANDED PROGRAMME ON IMMUNIZATION. Pro-visional guidelines for the diagnosis and classifi-cation of EPI target disease for primary healthcare, surveillance and special studies. Geneva:World Health Organization, 1983. EP1/GEN/83/4.

4. FINE, P. E. M. Problems in the collection and anal-ysis of data in leprosy studies. Lepr. Rev. 52 Suppl.(1981) 197-206.

5. FINE, P. E. M., Jon, C. K., MCDOUGALL, A. C.,MEYERS, W. M. and PONNIGHAUS, J. M. Compara-bility among histopathologists in the diagnosis andclassification of leprosy. Int. J. Lepr. 54 (1986)614-625.

6. FINE, P. E. M., PoNNIGHAtis, J. NI., MAINE, N.,CLARKSON, J. A. and BLISS, L. The protective ef-

46 2^ International Journal of Leprosy^ I 9 8 7

ficacy of BCG against leprosy in Northern Malawi.Lancet 2 (1986) 499-502.

7. McDoticiALL, A. C., PoNNiGlixtis, J. M. and FINE,

P. E. M. Histopathological examination of skinbiopsies from an epidemiological study of leprosyin Northern Malawi. Int. J. Lepr. 55 (1987) 88-98.

8. NEWELL, K. W. An epidemiologist's view of lep-rosy. Bull. WHO 34 (1966) 827-857.

9. PoNNIGBAus, J. M., Buss, L., McDotiGALL, A. C.and FINE, P. E. NI. The influence of different casedefinitions on the observed pattern of leprosy inan endemic area in Northern Malawi. Int. J. Lepr.52 Suppl. (1984) 744.

10. PONNIGHAUS, J. M., FINE, P. E. M., BLISS, L.,

SLINEY, I. J., BRADLEY, D. J. and REES, R. J. W.The Lepra Evaluation Project (LEP); an epide-miological study of leprosy in Northern Malawi.I. Methods. Lepr. Rev. (1987) (in press).

1 I. RAIMIAKRISIINA, S., NAIR, N. G. K. and JAvABAL,P. Implications of misdiagnosis in field trials ofvaccines. Indian J. Med. Res. 80 (1984) 711-720.

12. RIDLEY, D. S. Therapeutic trials in leprosy usingserial biopsies. Lepr. Rev. 29 (1958) 45-52.

13. VOORHOEVE, A. M., MULLER, A. S., SCIIULI'EN, T.W., T'MANNETJE, W. and VAN RENS, M. MaChakOS

project studies. Agents affecting health of motherand child in a rural area of Kenya. IV: The epi-demiology of pertussis. Trop. Geogr. Med. 30(1978) 125-139.