ORIGINAL COMMUNICATION

Handwriting as an objective tool for Parkinsons disease diagnosis

Sara Rosenblum Margalit Samuel

Sharon Zlotnik Ilana Erikh Ilana Schlesinger

Received: 5 May 2013 / Revised: 3 June 2013 / Accepted: 4 June 2013 / Published online: 16 June 2013

Springer-Verlag Berlin Heidelberg 2013

Abstract To date, clinical assessment remains the gold

standard in the diagnosis of Parkinsons disease (PD). We

sought to identify simple characteristics of handwriting

which could accurately differentiate PD patients from

healthy controls. Twenty PD patients and 20 matched

controls wrote their name and copied an address on a paper

affixed to a digitizer. Mean pressure and mean velocity was

measured for the entire task and the spatial and temporal

characteristics were measured for each stroke. Results of

the MANOVAs for the temporal, spatial, and pressure

measures (stroke length, width, and height; mean pressure;

mean time per stroke; mean velocity), for both the name

writing and address copying tasks, showed significant

group effects (F(6,32) = 6.72, p \ 0.001; F(6,31) =14.77, p \ 0.001, respectively). A discriminant analysiswas performed for the two tasks. One discriminant function

was found for the group classification of all participants

(Wilks Lambda = 0.305, p \ 0.001). Based on thisfunction, 97.5 % of participants were correctly classified

(100 % of the controls and 95 % of PD patients). A Kappa

value of 0.947 (p \ 0.001) was calculated, demonstratingthat the group classification did not occur by chance. In this

pilot study we identified two simple short and routine

writing tasks which differentiate PD patients from healthy

controls. These writing tasks have future potential as cost-

effective, fast and reliable biomarkers for PD.

Keywords Parkinsons disease Biomarker Non-motor Handwriting Diagnosis Digitizer

Introduction

To date, clinical assessment remains the gold standard in

the diagnosis of Parkinsons disease (PD). In recent years,

an effort has been made to identify an objective biomarker

by organized initiatives such as the Parkinson Progression

Marker Initiative (PPMI) [1] and the Longitudinal and

Biomarker Study-Parkinsons disease [2] using varied

approaches including examination of blood, saliva, exhaled

breath [3], cerebrospinal fluid and neuroimaging. Finding a

biomarker is important not only for the diagnosis and fol-

low-up of disease progression but also for testing of

potential neuroprotective interventions with the hope of

screening and preventing this disease [4, 5]. In striving for

a tool that will objectively diagnose PD, the US Food and

Drug Administration (FDA) recently approved the dopa-

mine transporter (DAT) ligand [(123)I] ioflupane, with

single-photon emission computed tomography (SPECT,

DaTscan) for evaluating parkinsonian syndromes and for

distinguishing Parkinsons disease from essential tremor.

However, this testing modality exposes patients to radio-

active material and does not distinguish Parkinsons dis-

ease from multiple system atrophy or corticobasal

degeneration [6].

The statistical analysis was performed by Sara Rosenblum from the

Department of Occupational Therapy, Faculty of Social Welfare and

Health Sciences University of Haifa, Israel.

S. Rosenblum (&)Department of Occupational Therapy, Faculty of Social Welfare

and Health Sciences, University of Haifa, 31905 Haifa, Israel

e-mail: rosens@research.haifa.ac.il

M. Samuel S. ZlotnikDepartment of Occupational Therapy, Rambam Health Care

Campus, Haifa, Israel

I. Erikh I. SchlesingerDepartment of Neurology, Technion Faculty of Medicine,

Rambam Health Care Campus, Haifa, Israel

123

J Neurol (2013) 260:23572361

DOI 10.1007/s00415-013-6996-x

Handwriting is one of the most common daily activities

performed by adults in a variety of leisure and professional

settings. Handwriting is a complex human activity, which

requires fine dexterity abilities and involves an intricate

blend of cognitive, sensory and perceptual-motor compo-

nents [7]. Therefore it is not surprising that abnormal

handwriting, is a well recognized manifestation of PD, with

micrographia being characteristic [8]. It may appear years

before the clinical diagnosis is made and thus may be one

of the first signs of impending disease, enabling neuro-

protective interventions when identified. Previous research

has shown that handwriting measures have the potential for

identifying various stages of PD, effects of varied inter-

ventions [9] and the effect of medication [10]. Studies

focusing on understanding the mechanism underlying mi-

crographia found significant differences between the

handwriting of PD patients and healthy subjects [1114].

In this pilot study we sought to identify simple charac-

teristics of handwriting kinematics which could accurately

differentiate PD patients from healthy controls so that they

may potentially be used as a diagnostic and screening tool

for incipient disease.

Subjects and methods

Subjects

Forty subjects, 20 PD and 20 controls (aged 3881), with at

least 12 years of education (mean 13.37 SD 2.93) were

included in the study. PD patients, diagnosed according to

the UK Brain Bank criteria [15], Hoehn and Yahr [16]

stages II-III, mini-mental state exam (MMSE) scores of

[17] 30, were randomly selected from those attending the

Movement Disorder Clinic at Rambam Health Care Cam-

pus in Haifa, Israel. The control group comprised of

healthy subjects matched for gender, age (mean age of

controls, 61.66 10.00; PD 61.18 9.11), years of edu-

cation, and hand dominance. For further details see

Table 1.

The study was approved by the institutional review

board. All subjects signed a written informed consent.

Methods

Writing tasks were performed on A4-size lined paper

affixed to the surface of an Intuos II WACOM (404 X 306

X 10 mm) X and Y digitizing tablet, using a wireless

electronic pen with a pressure-sensitive tip (Model GP-

110). Displacement, pressure, and pen-tip angle were

sampled at 100 Hz via a 1,300 MHz Pentium (R) M laptop

computer. A computerized handwriting evaluation (Com-

PET, a revised version of the Penmanship Objective

Evaluation tool) [18], was used to administer the stimuli

and to collect and analyze the data. The tool includes two

main functions: (1) data collection, which is language-

independent and easy to use for any handwriting tasks, and

(2) data analysis, which is programmed via MATLAB

software toolkits. The computerized system enables the

collection of spatial, temporal, and pressure data for the

entire task, as well as for each stroke, while the subject is

writing.

Participants were requested to write their name and to

copy an address (same address for all). Patients performed

the study in the on state.

Outcome Measures

Handwriting performance of written outputs were varied in

their length among participants (as in writing ones full

name). Thus, we compared the measures of strokes and not

those of the entire task. A stroke was defined as a continuous

line written by the subject. A stroke can be one of two types:

on-paper stroke, which is a stroke, written on the paper, or

in-air stroke, which is a stroke that the pen creates when it

does not touch the paper, i.e. between on-paper strokes.

The primary outcome measures included temporal, spatial,

and pressure measures of handwriting kinematics. Temporal

measures included the mean time taken to write each stroke,

the mean on-paper stroke time, the mean in-air stroke

time, and the mean velocity for the entire task in seconds.

The spatial measures included the mean strokes path

length, width, and height in centimeters. The pressure

measure was the mean pressure applied to the writing sur-

face in non-scaled units from 0 to 1,024 for the entire task.

All analyses were done by one of the authors (S.R.) who was

blinded to subject diagnosis (healthy vs. PD).

Statistical analysis

MANOVA analyses were used to test for group differences

(controls vs. PD) across the computerized spatial, temporal

and pressure measures for each of the two writing tasks

Table 1 Demographic and clinical characteristics of the studyparticipants

Characteristic Controls (n = 20) PD (n = 20) p value

Age, mean, years 61.66 10.00 61.18 9.11 NS

Female sex, no. (%) 11 (55) 11 (55) NS

Years of education 13.78 1.65 12.80 1.70 NS

MMSE 30 0 29.88 0.33 NS

Hoehn and Yahr 2 0.79

UPDRS 25.05 14.04

Disease duration 8.05 3.90

2358 J Neurol (2013) 260:23572361

123

(e.g., writing ones full name and address copying). In

order to examine the source of any significance, the data

from each task were subjected to univariate ANOVAs.

T-tests were used for analyzing the group differences in

two components of the mean stroke duration: stroke

duration on paper and in air. Finally, discriminant analysis

was conducted in order to determine which of the com-

puterized measures best predicted group membership (i.e.,

control vs. PD) for each of the tasks

Wilks Lambda 0:601F6;74 8:20; p\0:0001 g2 0:399:

Results

Results of the MANOVAs for both the name writing

(F(6,32) = 6.72, Wilks Lambda = 0.442 p \ 0.001g2 = 0.558) and address copying tasks (F(6,31) = 14.77,Wilks Lambda = 0.259 p \ 0.001 g2 = 0.741), showedsignificant group effects. The univariate ANOVAs indi-

cated significant differences between the groups for several

measures of the name writing task and for all measures of

the address copying task. The means and standard devia-

tions are presented in Table 2.

In order to further analyze the mean stroke duration,

t-tests were conducted for the mean stroke duration on

paper and in air for the address copying task. Mean

stroke duration on paper (controls 0.19 0.02 s vs. PD

0.27 0.10 s p = 0.001) and in air (controls

0.21 0.06 s vs. PD 0.38 0.15 s p \ 0.001) were sig-nificantly different between controls and PD patients.

For both tasks, compared with controls, PD participants

wrote significantly smaller letters, applying significantly

less pressure on the writing surface, and requiring signifi-

cantly more performance time. The gap in the stroke

duration in air was more dramatic than the gap in the

stroke on paper. For both groups, the stroke size

decreases as the task requirements increased, as shown in

Table 2.

In order to assess the relative importance of the dif-

ferent variables in differentiating between PD and control

participants, a discriminant analysis was performed for the

two tasks. One discriminant function was found for the

group classification of all participants (Wilks Lambda

(= 0.305, p \ 0.001). As shown in Table 3, the variableswhich made the greatest contribution to group member-

ship were the mean width, duration and height in both

tasks. Based on this function, 97.5 % of participants were

correctly classified (100 % of the controls and 95 % of

PD patients). A Kappa value of 0.947 (p \ 0.001) wascalculated, demonstrating that the group classification did

not occur by chance. Thus, taken together these two

handwriting tasks had 95 % sensitivity and 100 %

specificity.

Discussion

Using two simple short and routine writing tasks, PD

patients could be distinguished from healthy controls with

very high accuracy. None of the controls were categorized

as suffering from PD while one of the PD patients was

incorrectly categorized as being healthy. Thus, taken

together these two handwriting tasks had 95 % sensitivity

and 100 % specificity. Handwriting was chosen as a pos-

sible biomarker for PD based on previous reports of mi-

crographia appearing many years before the diagnosis of

PD [19]. It is a noninvasive method that is easily per-

formed. Hence, the sensitivity and specificity for the dis-

tinction of PD patients from healthy controls could allow a

quick and simple initial diagnosis in a non-specialist

Table 2 Kinematic measuresfor writing tasks

* p \ 0.05, ** p \ 0.01

Measures Writing ones full name Copying an address

Healthy mean

(SD)

PD mean (SD) Healthy mean

(SD)

PD mean (SD)

Spatial measures

Stroke length (cm) 0.96 (0.40) 0.60 (0.34)* 0.75 (0.14) 0.53 (0.16)**0

Stroke width (cm) 0.32 (0.11) 0.18 (0.07)** 0.23 (0.04) 0.17 (0.04)**0

Stroke height (cm) 0.47 (0.16) 0.36 (0.25) 0.35 (0.07) 0.27 (0.07)**0

Temporal measures

Stroke duration (sec) 0.16 (0.04) 0.21 (0.09) 0.20 (0.04) 0.32 (0.12)**

Velocity (cm/sec) 5.18 (1.26) 3.10 (1.30)** 4.86 (0.96) 2.67 (0.91)**

Pressure measures

Pressure(Non-scaled units

01,024)

783.55 (149.04) 591.77 (160.98)** 788.61 (146.59) 603.29 (164.08)**

J Neurol (2013) 260:23572361 2359

123

setting. Hence, the writing task could be applied by general

practitioners to ensure the speedy referral of new PD

patients to the specialists, who would confirm the diagnosis

and choose an adequate therapeutic approach.

Our study was designed to examine the possibility of

handwriting as a biomarker for PD, but in doing so also

shed light on the ability of PD patients to perform this task.

We found that it is importance to analyze handwriting not

only on paper but also in air, as significant differences

were observed between these two conditions, more so for

the address copying task. In fact, in-air time is a mani-

festation of planning the next movement, as required in

the sequential process of handwriting [14] and thus reflects

cognitive ability, as shown in our previous studies [20, 21],

and supplies important information about the writer [22].

Hence, the current study results indicate a motor planning

deficit among PD patients, even in a very short functional

common task such as copying an address. Our findings

further support previous observations that in comparison

with controls, patients with PD exhibit more variable peak

accelerations and stroke sizes and smaller than normal

handwriting, a phenomenon known as micrographia [8, 11

14, 23].

Less pressure was applied on the writing surface by the

PD group for both the name and address writing tasks. This

may be an indication of a deficit in force modulation as seen

in the speech of PD patients, expressed as hypophonia [24].

In this pilot study we identified two simple, short and

routine writing tasks which differentiate PD patients from

healthy controls. If our findings are corroborated in larger

studies, these writing tasks have future potential as cost-

effective, fast and reliable biomarkers for PD, and hand-

writing, an everyday non-threatening activity, could be

assessed as a possible marker of preclinical PD in at-risk

populations.

Conflicts of interest The authors declare no disclosures.

References

1. Marek K, Jennings D, Lasch S (2011) The Parkinson progression

marker Initiative (PPMI). Prog Neurobiol 95:629663

2. Ravina B, Tanner C, Dieuliis D, Eberly S, Flagg E, Galpern WR,

Fahn S, Goetz CG, Grate S, Kurlan R, Lang AE, Marek K,

Kieburtz K, Oakes D, Elliott R, Shoulson I; Parkinson Study

Group LABS-PD Investigators (2009) A longitudinal program for

biomarker development in Parkinsons disease: a feasibility

study. Mov Disord 24:20812090

3. Tisch U, Schlesinger I, Ionescu R, Nassar M, Axelrod N, Rob-

ertman D, Tessler Y, Azar F, Marmur A, Aharon-Peretz J, Haick

H (2013) Detection of Alzheimers and Parkinsons disease from

exhaled breath using nanomaterial-based sensors. Nanomedicine

(Lond) 8:4356

4. Wu Y, Le W, Jankovic J (2011) Preclinical biomarkers of Par-

kinson disease. Arch Neurol 68:2230

5. Chahine LM, Stern MB (2011) Diagnostic markers for Parkin-

sons disease. Curr Opin Neurol 24:309317

6. de la Fuente-Fernandez R (2012) Role of DaTSCAN and clinical

diagnosis in Parkinson disease. Neurology 78:696701

7. Carmeli E, Patish H, Colman R (2003) The aging hand. J Ger-

ontol A Biol Sci Med Sci 58:146152

8. Horowski R, Horowski L, Vogel S, Poewe W, Kielhorn FW

(1995) An essay on Wilhelm von Humboldt and the shaking

palsy: first comprehensive description of Parkinsons disease by a

patient. Neurology 45:565568

9. Eichhorn TE, Gasser T, Mai N, Marquardt C, Arnold G, Schwarz

J, Oertel WH (1996) Computational analysis of open loop

handwriting movements in Parkinsons disease: a rapid method to

detect dopamimetic effects. Mov Disord 11:289297

10. Poluha PC, Teulings HL, Brookshire RH (1998) Handwriting and

speech changes across the levodopa cycle in Parkinsons disease.

Acta psycho (Amst) 100:7184

11. Teulings HL, Contreras-Vidal JL, Stelmach GE, Adler CH (2002)

Adaptation of handwriting size under distorted visual feedback in

patients with Parkinsons disease and elderly and young controls.

J Neurol Neurosurg Psychiatry 72:315324

12. Tucha O, Mecklinger L, Thome J, Reiter A, Alders GL, Sartor H,

Naumann M, Lange KW (2006) Kinematic analysis of dopami-

nergic effects on skilled handwriting movements in Parkinsons

disease. J Neural Transm 113:609623

13. Oliveira RM, Gurd JM, Nixon P, Marshall JC, Passingham RE

(1997) Micrographia in Parkinsons disease: the effect of pro-

viding external cues. J Neurol Neurosurg Psychiatry 63:429433

14. Ondo WG, Satija P (2007) Withdrawal of visual feedback

improves micrographia in Parkinsons disease. Mov Disord

22:21302131

15. Gibb WR, Lees AJ (1988) The relevance of the Lewy body to the

pathogenesis of idiopathic Parkinsons disease. J Neurol Neuro-

surg Psychiatry 51:745752

16. Hoehn MM, Yahr MD (1967) Parkinsonism: onset, progression

and mortality. Neurology 17:427442

17. Folstein MF, Folstein SE, McHugh PR (1975) Mini-mental

state: a practical method for grading the cognitive state of

patients for the clinician. J Psychiatr Res 12:189198

18. Rosenblum S, Parush S, Weiss PL (2003) Computerized temporal

handwriting characteristics of proficient and non-proficient

handwriters. Am J Occup Ther 57:129138

19. Becker G, Muller A, Braune S, Buttner T, Benecke R, Greulich

W, Klein W, Mark G, Rieke J, Thumler R (2002) Early diagnosis

of Parkinsons disease. J Neurol 249(Suppl 3:III):4048

20. Rosenblum S, Dvorkin A, Weiss PL (2006) Automatic segmen-

tation as a tool for examining the handwriting process of children

Table 3 Discriminant analysis structure matrix predictors loadingvalues

Measure Loading

Writing ones full name Stroke width on paper 0.496

Copying an address Stroke duration in air -0.477

Copying an address Stroke width on paper 0.422

Copying an address Stroke height in air 0.392

Copying an address Stroke duration on paper -0.389

Writing ones full name Stroke duration in air -0.219

Writing ones full name Stoke height in air 0.183

Copying an address Stroke duration on paper -0.143

2360 J Neurol (2013) 260:23572361

123

with dysgraphic and proficient handwriting. Hum Mov Sci

25:608621

21. Werner P, Rosenblum S, Bar-On G, Heinik J, Korczyn A (2006)

Handwriting process variables discriminating mild Alzheimers

disease and mild cognitive impairment. J Gerontol B Psychol Sci

Soc Sci 61:228236

22. Sesa-Nogueras E, Faundez-Zanuy M, Mekyska J (2012) An

Information Analysis of In-Air and On-Surface Trajectoriesin

Online Handwriting. Cogn Comput 4:195205

23. McLennan JE, Nakano K, Taylor HR, Schwab RS (1972) Mi-

crographia in Parkinsons disease. J Neurol Sci 15:141152

24. Hatzitaki V, Hoshizaki TB (1998) Dynamic joint analysis as a

method to document coordination disabilities associated with

Parkinsons disease. Clin Biomech 13:182189

J Neurol (2013) 260:23572361 2361

123

Handwriting as an objective tool for Parkinsons disease diagnosisAbstractIntroductionSubjects and methodsSubjectsMethodsOutcome MeasuresStatistical analysis

ResultsDiscussionConflicts of interestReferences