Evaluation of a Screening Questionnaire forGenetic Studies of Parkinson’s Disease

Brad A. Racette,1* Melissa Rundle,1 Abbas Parsian,1,3 and Joel S. Perlmutter1,2

1Department of Neurology and Neurologic Surgery (Neurology), Washington University School of Medicine,St. Louis, Missouri

2Mallinckrodt Institute of Radiology, Washington University School of Medicine, St. Louis, Missouri3Department of Psychiatry, Washington University School of Medicine, St. Louis, Missouri

A screening questionnaire with high sensi-tivity for detection of Parkinson’s diseasewould make it easier to identify undiag-nosed, yet affected, family members for ge-netic research. We assessed the validityof a screening questionnaire developed byDuarte et al. [1995: Mov Disord 10:643–649]with reported high specificity and sensitiv-ity for Parkinson’s disease (PD). We appliedthe questionnaire to 78 asymptomatic mem-bers of families that had at least two peoplediagnosed with PD. These families were par-ticipating in a linkage study of Parkinson’sdisease. Examination of these 78 revealedthat 53 were normal (normal controls) and25 were classified (“undiagnosed” PD de-fined) as possible, probable, or clinicallydefinite PD based on standardized criteria.We compared these results with 123 patientswith clinically definite PD (“diagnosed”PD). There were significant differencesamong the mean scores on the question-naire for normal controls (4.4), subjects withundiagnosed PD (9.8), and patients with di-agnosed PD (42.1; p<0.000001) and a signifi-cant difference between undiagnosed PDand normals (p<0.01). The questionnairehad only 4% sensitivity for detection of par-kinsonism in undiagnosed PD using theoriginal criteria [Duarte et al., 1995]. Revis-ing the criteria increased the sensitivityfrom 4 to 48% in the undiagnosed group. Thepositive predictive value was 39% and the

negative predictive value was 72%. Prospec-tive application of these revised criteria isnecessary to confirm the improved sensitiv-ity. However, we conclude that this screen-ing questionnaire has inadequate sensitiv-ity for detection of mild parkinsonism anddirect examination is still critical for accu-rate classification for genetic studies. Am. J.Med. Genet. (Neuropsychiatr. Genet.) 88:539–543, 1999. © 1999 Wiley-Liss, Inc.

KEY WORDS: screening questionnaire;Parkinson’s disease; genetics

INTRODUCTION

With the recent report of a mutation in the a-synuclein gene sufficient to cause Parkinson’s disease(PD) in several, large multi-incidence families [Polyme-ropoulos et al., 1997], there has been renewed interestin genetic susceptibility factors that may exist for themajority of PD patients. Family studies have been im-paired by selection bias or lack of direct observation ofaffected individuals. A telephone screening question-naire with high sensitivity would be a useful and cost-effective method of screening family members for ge-netic studies of PD. Several studies have investigatedthe specificity and sensitivity of screening question-naires for diagnosis of PD [Li et al., 1985; Mutch et al.,1991; Schoenberg et al., 1985]. Meneghini et al. [1992]found that two questions directed at symptoms of bra-dykinesia, rigidity, and tremor yielded 100% sensitiv-ity in a group of 21 hospitalized patients with PD. Theyfound that a brief neurologic examination assessingtandem walk and rigidity at the elbow added no addi-tional sensitivity. Duarte et al. [1995] developed a nine-item self questionnaire that addressed common parkin-sonian symptoms and reported a sensitivity and speci-ficity of 100%. Their questionnaire distinguishedessential tremor from PD in all 32 subjects. When thisquestionnaire was administered to 100 unselected,asymptomatic patients in an ophthalmology practice,the authors identified three new people with PD sub-sequently verified by examination. This report sug-

Contract grant sponsor: National Institutes of Health; Contractgrant numbers: NS31001, NS32318, NS07205; Contract grantsponsor: Greater St. Louis Chapter of the American Parkinson’sDisease Association; Contract grant sponsor: Ruth KopolowFund; Contract grant sponsor: Clinical Hypotheses Research Sec-tion, Charles A. Dana Foundation.

*Correspondence to: Brad A. Racette, M.D., Washington Uni-versity School of Medicine, 660 South Euclid Ave, Box 8111, St.Louis, MO. E-mail: racetteb@neuro.wustl.edu

Received 28 September 1998; Accepted 26 February 1999

American Journal of Medical Genetics (Neuropsychiatric Genetics) 88:539–543 (1999)

© 1999 Wiley-Liss, Inc.

gested that the screening instrument could provide animportant cost-savings for family studies of PD. It stillremained to be determined whether the sensitivity andspecificity was sufficient for ascertainment of asymp-tomatic affected individuals for family studies of PDsince these people may have different biases in answer-ing the screening questions. To this end, we deter-mined the sensitivity and specificity of this question-naire in 78 asymptomatic family members in multi-incidence PD pedigrees participating in a genetic studyfor Parkinson’s disease and compared these resultswith questionnaire scores obtained from 123 symptom-atic subjects with PD.

MATERIALS AND METHODS

The questionnaire used in this study was initiallydeveloped by Tanner et al. [1990] and modified, vali-dated, and weighted by Duarte et al. [1995]. A coordi-nator trained in history taking for PD and blinded tothe examination findings read the questionnaire to thepatients. Table I lists the questions and the publishedsuggested weighting criteria. The scores were weightedaccording to the published criteria [Duarte et al., 1995]and a total score of 42 was considered consistent withparkinsonism. Our revised criteria required only asingle positive or “uncertain” answer to increase thesensitivity. An unblinded movement disorders expertdid all neurological examinations.

We screened a total of 201 individuals with the Par-kinson’s disease screening questionnaire (PDSQ). Eachsubject provided a detailed family history for all firstdegree relatives including: name, age (or age at death),medical problems, cause of death (when relevant), andwhether any of these individuals were known to havePD. A positive family history was defined as one ormore affected first degree relatives. We performed aUnified Parkinson’s Disease Rating Scale-Motor Sub-section [UPDRS; Fahn et al., 1987] for each subject.Subjects were also given the Folstein Mini MentalState Examination [Folstein et al., 1975]. Twenty-fiveof 123 subjects with known idiopathic PD also weredemented based on a history of progressive cognitivedecline, a Mini Mental State exam score less than 25,or both. Four “diagnosed” (clinically definite PD) pa-tients could not answer questions for themselves sec-

ondary to dementia or severe hypophonia and the legalguardian provided the answers. Although some diag-nosed patients may have alleviation of symptoms fromtherapy, we did not instruct them how to answer ques-tions with respect to continuous or intermittent pres-ence of symptoms. All sporadic patients and most of theprobands of familial cases are followed regularly in theMovement Disorders Center at Washington UniversitySchool of Medicine.

Based on neurological examination, we classified in-dividuals as clinically possible, probable, or definite PDusing a modification of criteria proposed by Calne et al.[1992]. All clinically definite PD patients exhibitedthree of the following: rest tremor, rigidity, bradykine-sia, or postural instability; or two of these features withone of the first three displaying asymmetry. Supportivecriteria and exclusionary criteria were taken from theUnited Kingdom Parkinson’s Disease Society BrainBank clinical diagnostic criteria [Hughes et al., 1992].Three or more of the following supportive criteria alsowere required for a diagnosis of PD: unilateral onset;persistent asymmetry; rest tremor; progression; excel-lent subjective response to levodopa (“70–100%” im-provement); severe levodopa-induced chorea; levodoparesponse for 5 years or more, or clinical course of 10years or more. The exclusionary criteria included a his-tory of repeated strokes with stepwise progression ofparkinsonian features; history of repeated head injury;history of definite encephalitis, oculogyric crises; neu-roleptic treatment at onset of symptoms; sustained re-mission; strictly unilateral features after 3 years; su-pranuclear gaze palsy; cerebellar signs; early severeautonomic involvement; early severe dementia (withinfirst year of onset) with disturbances of memory, lan-guage, and praxis; extensor plantar reflex (not ex-plained by known central nervous system lesion); cere-bral tumor or communicating hydrocephalus on brainimaging; negative response to large doses of levodopa(>2 grams per day) if malabsorption is excluded; or1-methyl-4-phenyl-1,2,36-tetrahydropyridine (MPTP)exposure. We did not include the exclusionary criteriaof “more than one affected relative.” Dementia did notexclude subjects if the onset of cognitive changes oc-curred after the first year of onset of parkinsonism andin the setting of otherwise typical idiopathic PD meet-ing criteria for clinically definite PD. If a subject meetsinclusion criteria for PD but does not have the required

TABLE I. Percent of Individuals From Each of the Diagnostic Groups Giving a Positive Answer to Questions*

QuestionWeighted

score DPDa UPDb Normal

Do you have trouble arising from a chair? 6 62 17 9Is your handwriting smaller than it once was? 7 72 14 8Do people tell you that your voice is softer than it once was? 8 66 12.5 8Is your balance, when walking, poor? 9 69 11 13Do your feet suddenly seem to freeze in doorways? 6 37 17 0Does your face seem less expensive than it used to? 6 62 11 2Do your arms and legs shake? 9 91 11 2Do you have trouble buttoning buttons? 8 78 12.5 9Do you shuffle your feet and take tiny steps when you walk? 8 68 12.5 2

*All numbers represent percentage answered affirmatively in each group.aDPD 4 diagnosed PD.bUPD 4 undiagnosed PD.

540 Racette et al.

three supportive criteria, the level of certainty is re-duced to clinically probable.

Clinically probable PD is defined as a combination ofany two of rest tremor, rigidity, bradykinesia, or im-paired postural reflexes. Alternatively, asymmetricrest tremor, bradykinesia, or rigidity makes a diagno-sis of clinically probable PD. Clinically possible PD isdefined as any one of bradykinesia, rigidity, rest/postural tremor, or postural instability. Impairment ofpostural reflexes is included unless explained by otherneurologic or orthopedic abnormalities. For the clini-cally possible category, we further specify that asym-metric postural tremor must have begun within thelast 5 years.

RESULTSWe examined 78 family members of probands in

multi-incidence PD pedigrees who considered them-selves normal. Examination of these 78 revealed that53 were normal and 25 were classified as possible,probable, or clinically definite PD. We defined these 25subjects as “undiagnosed” PD and the 53 as normalcontrols. None of the 78 subjects had ever been diag-nosed with PD or other neurologic disorder. Mean agesof the undiagnosed PD subjects and normals were 56(range 31–87) and 61 (range 27–89), respectively. Wealso diagnosed 123 symptomatic patients with clini-cally definite PD with a mean age of 68 (range 27–91).Fifty-four had a positive family history of PD, whereas69 were sporadic with no family history. The mean agesin the three groups were statistically significantly dif-ferent using an analysis of variance (F[1,2]47.09;p40.001) but had substantial overlap (Fig. 1).

The responses to the questionnaire from our undiag-nosed subjects are presented in Table I. Ten of 27 ofthese individuals had bradykinesia and rigidity astheir parkinsonian features. Most of the remainder hadasymmetric postural tremor as well as at least one par-kinsonian feature. One young subject had postural in-stability and retropulsion with no other explanation forthese symptoms. Three undiagnosed subjects had clini-cally definite PD, 15 subjects had clinically probablePD, and six had clinically possible PD. Most of the clini-cally probable PD subjects had asymmetric featuressuggestive of idiopathic PD but lacked the requiredsupportive criteria to make the diagnosis of clinically

definite PD. The question regarding tremor yielded thehighest sensitivity followed by difficulty buttoning andchange in size of handwriting. The presence of freezingpredicted parkinsonism with the greatest specificity.The most common false positive answer was difficultywith balance. Undiagnosed PD subjects most com-monly answered affirmatively to the questions regard-ing difficulty arising from a chair and freezing. Theclinical features of these undiagnosed PD subjects arepresented in Table II.

Only one of the 25 undiagnosed subjects found tohave clinically possible, probable, or definite PD waspositively identified by the questionnaire, using pub-lished criteria, yielding a sensitivity of 4% (33% forclinically definite alone). Defining any positive answeron the questionnaire as indicative of possible parkin-sonism improved the sensitivity to 48% (66% for clini-cally definite alone). The positive predictive value was39% and the negative predictive value was 72% usingthe revised criteria. Using the published criteria for thePDSQ with our patient sample, we find a sensitivity of67% and a specificity of 100% for diagnosed PD com-pared with our revised criteria, which produced a sen-sitivity of 100% and specificity of 63%.

The mean PDSQ weighted scores were 9.8 for undi-agnosed PD subjects, 4.4 for normals, and 46.1 for di-agnosed PD subjects. An analysis of variance demon-strated that there were statistically significant differ-ences among the scores of the undiagnosed subjects,normals, and diagnosed PD subjects (F[1,2]4191.7;p<0.000001). Pairwise, post hoc t-tests revealed thatthe undiagnosed PD (p<.01) and diagnosed PD patients(p<0.000001) had significantly higher scores than nor-mals. Diagnosed PD patients also had higher scoresthan the undiagnosed subjects (p<0.000001). Therewas no correlation between motor scores on the motorsubscale of the UPDRS and the PDSQ score in eitherdiagnosed PD (r4.14) or undiagnosed PD (r4.14).

DISCUSSION

We found that the published criteria for a PD screen-ing questionnaire [Duarte et al., 1995] has limited sen-sitivity for detection of parkinsonism as determined byclinical examination. Although modification of thequestions and revised criteria retrospectively increasedthe sensitivity of the questionnaire, it reduced specific-ity from 100 to 63%. Low specificity, however, is not amajor limitation for a screening questionnaire sincethe questionnaire could reduce substantially the num-ber of subjects that require further evaluation. Basedon our revised criteria we would have eliminated 61%of the total subjects screened for further evaluation,with 35 of 48 properly excluded, based on subsequentclinical exam.

There are two potential sources of bias in this study.First, we elected to read the PDSQ to subjects in thestudy instead of using it as a self questionnaire toverify that they had understood each question. Thisapproach may bias toward increased sensitivity andreduced specificity. Our subjects might be more likelyto give positive answers since they were aware of thepositive family history of parkinsonism and its associ-

Fig. 1. Age distributions for undiagnosed PD, diagnosed PD subjects,and normal controls.

Screening for Parkinson’s Disease 541

ated symptoms. Alternatively, it is possible that thesepeople may tend to deny symptoms or were unaware oftheir symptoms similar to the lack of awareness ofHuntington’s disease patients of their involuntarymovements [Snowden et al., 1998]. Second, the move-ment disorders specialist performing the neurologic ex-aminations was not blinded and may have a bias toidentify subtle signs of possible PD to avoid missing anaffected individual. This would tend to reduce the sen-sitivity of the questionnaire. However, without patho-logical, neuroimaging, or longitudinal information theclinical examination remains the only means to iden-tify potentially affected individuals, justifying this biasfor genetic studies. Subjects thought to have possible orprobable PD may need further evaluation to providegreater diagnostic confidence. The majority of undiag-nosed PD subjects had clinically probable PD sincethey lacked the requisite supportive criteria that relyon longitudinal history or examination. Additionally,we included isolated postural tremor as a criterion forpossible PD as suggested by the Calne criteria [Calneet al., 1992] since it is common in PD and may be theonly type of tremor experienced [Lance et al., 1963].

The poor performance of the PDSQ in detectingsymptomatic (diagnosed) PD subjects was surprising.Published criteria require several positive answers tomake a diagnosis of PD [Duarte et al., 1995]. We mayhave had more mild cases of PD in our sample, whichcould result in a lower total score. Our finding thatexamination improves the sensitivity of the screeningquestionnaire is in contrast to epidemiologic studies,which have failed to improve the sensitivity of symp-tom questionnaires for PD with the addition of a brief

neurologic examination [Meneghini et al., 1992]. Pro-spective validation of video tape review and correlationwith physical examination may improve the sensitivityof screening family members for parkinsonism.

The sensitivity and specificity of a diagnostic orscreening tool is critically important for genetic linkagemapping of PD. A simple screening questionnaire withhigh sensitivity could provide an important savings intime and resources. In addition, the questionnairecould permit much more accurate epidemiologic sur-veys to assess pathogenic factors contributing to PD. Atthis point, we chose to compare the results of the ques-tionnaire with clinical examination, but it is importantto appreciate that the clinical examination also haslimitations. Some subjects may be affected but have notyet developed clinical manifestations at the time of theexamination. This is particularly relevant for PD asdisease risk increases with age. Premorbid identifica-tion of these individuals would improve accuracy ofphenotypic classification for genetic linkage studies.

Theoretically, a tool such as [18F]-dopa positronemission tomography (PET) scanning could assist inthe diagnosis of presymptomatic cases of PD to improveaccuracy of diagnosis for linkage studies [Black &Perlmutter 1997; Burns et al., 1992]. For example, anearly twin study done before the availability of PETfailed to demonstrate a difference in the concordancerate between monozygotic and dizygotic twins [Ward etal., 1983]. A follow-up study by Zimmerman et al.[1991] using [18F]- dopa PET to detect preclinical PDfound an additional affected twin raising the concor-dance rate. One of the twin pairs was discordant for 26years, stressing the importance of longitudinal evalu-

TABLE II. Examination and PDSQ Scores for Undiagnosed Subjects WithAbnormal Examinations*

PatientPosturaltremor

Resttremor Rigidity Bradykinesia

Posturalinstability

PDQtotal

1 S 02 A A 03 A A A 134 A X 05 A 06 S 07 A A 08 A 09 A X 0

10 A A X 4411 A A A 012 A A 913 A A A 1714 A A 3015 A A 016 S S X 617 A S 2418 S A 019 S S X 3920 A 2321 X 922 A A A X 023 S A A X 024 A A 1725 A A X 15

*A 4 asymmetric; S 4 symmetric, X 4 present. The side of greatest involvement was the same for all asym-metric signs in each subject.

542 Racette et al.

ation and premorbid identification. Furthermore, Pic-cini et al. [1997] demonstrated reduced putamen [18F]-dopa uptake in eight asymptomatic members of a kin-dred with familial PD. Their findings support thenotion that PET may be more sensitive than clinicalexamination to detect affected individuals. Thus, a sen-sitive method of performing genetic linkage studies inPD may consist of a combination of screening question-naires and examinations for those with positive an-swers and PET studies in unaffected relatives.

REFERENCESBlack, K and Perlmutter JS. 1997. Can 18F-Fluoro-dopa positron emission

tomography help determine disease status in familial Parkinson’s dis-ease? Neurol Network Commentary 1:308–314.

Burns DJ, Mark MH, Playford ED, Maranganore DM, Zimmerman TR,Duvoisin RC, Harding AE, Marsden CD, Brooks DJ. 1992. Parkinson’sdisease in twins studied with F-dopa and positron emission tomog-rophy. Neurology 42:1984–1900.

Calne DB, Snow BJ, Lee C. 1992. Criteria for Diagnosing Parkinson’sdisease. Ann Neurol 32:S125–S127.

Duarte J, Claveria E, De Pedro-Cuesta J, Sempere P, Coria F, Calne DB.1995. Screening Parkinson’s disease: a validated questionnaire of highspecificity and sensitivity. Mov Disord 10:643–649.

Fahn S, Elton RL, and Members of the UPDRS Development Committee.1987. Unified Parkinson’s disease rating scale. In: Fahn S, MarsdenCD, Goldstein M, Calne DB, editors. Recent developments in Parkin-son’s disease. New York: Macmillan. p 153–163.

Folstein MF, Folstein S, McHugh P. 1975. Mini-Mental State. A practicalmethod for grading the cognitive state of patients for the clinician. JPsychiatr Res 112:189–198.

Hughes AJ, Daniel SE, Kilford L, Lees AJ. 1992. Accuracy of clinical di-agnosis of idiopathic Parkinson’s disease: a clinico-pathological study of100 cases. J Neurol Neurosurg Psychiatry 55:181–184.

Lance JW, Schwab RS, Peterson EA. 1963. Action tremor and the cogwheelphenomenon in Parkinson’s disease. Brain 86:95–110.

Li S, Schoenberg BS, Wang C, Cheng X, Rui D, Bolis CL, Schoenberg DG.1985. A prevalence survey of Parkinson’s disease and other movementdisorders in the People’s Republic of China. Arch Neurol 42:655–57.

Meneghini F, Rocca WA, Anderson DW, Grigoletto F, Morgante L, ReggioA, Savettieri G, Di Perri R. 1992. Validating screening instruments forneuroepidemiologic surveys: experience in Sicily. J Clin Epidemiol 45:319–331.

Mutch WJ, Smith WC, Scott RF. 1991. A screening and alerting question-naire for parkinsonism. Neuroepidemiology 10:150–156.

Piccini R, Morrish PK, Turjanski N, Sawle GV, Burn DJ, Weeks RA, MarkMH, Maranganore DM, Lees AJ, and Brooks DJ. 1997. Dopaminergicfunction in familial Parkinson’s disease: a clinical and 18F-dopa posi-tron emission tomography study. Ann Neurol 41:222–229.

Polymeropoulos MH, Lavedan C, Leroy E, et al. 1997. Mutation in thea-synuclein gene identified in families with Parkinson’s disease. Sci-ence 276:2045–2047.

Schoenberg BS, Anderson DW Haerer AF. 1985. Prevalence of Parkinson’sdisease in the biracial population of Copiah County, Mississippi. Neu-rology 35:841–845.

Snowden JS, Craufurd D, Griffiths HL, Neary D. 1998. Awareness of in-voluntary movements in Huntington’s disease. Arch Neurol 55:801–805.

Tanner CM, Gilley DW, Goetz CG. 1990. A brief screening questionnairefor parkinsonism [Abstract]. Ann Neurology 28:267–268.

Ward CD, Duvoisin RC, Ince SE, Nutt J. Eldridge R, Calne DB. 1983.Parkinson’s disease in twins. Neurology 33:815–824.

Zimmerman TR, Bhatt M, Calne DB. 1991. Parkinson’s disease in mono-zygotic twins: a follow-up. Neurology: 41 (Suppl 1):255.

Screening for Parkinson’s Disease 543