Abstract
Introduction:
Every human being has a slight action tremor of the hands. This physiological tremor increases with age, undergoes daily fluctuations and is influenced by excitement, nutrients and drugs. Tremor is also a symptom of many diseases, particularly in essential tremor (ET) and Parkinson's disease (PD) patients. Tremor has both static and kinetic features, broadly described by involuntary oscillations around a joint. Symmetric postural hand and arm tremor is typical of ET. Spiral drawing on paper is generally rated clinically to assess task-directed hand tremor severity. Digital spiral drawing represents a technological advanced alternative suitable for use at large scale. To date, evidence is limited on the boundaries within which tremor may be considered ordinary in the general population.
Aims:
We aimed to derive the normative range and frequency distribution of the Bain-Findley clinical tremor rating scale (TRS), to assess tremor from drawn spirals, among an adult population sample from the Cooperative Health Research In South Tyrol (CHRIS) study, according to age, sex and dominance of the drawing hand.
Methods:
Digital spiral drawing was assessed in 10,994 participants recruited between February 2013 and December 2018 in the ongoing CHRIS study, using a Wacom Intuos 5L Digitizer tablet connected to a PC with 200Hz sampling frequency, 5080lpi resolution, 2048 pressure levels and ±0.25mm accuracy. Participants were instructed to draw 2 or 6 spirals with a digital pen within margins of a clockwise Archimedean spiral template printed on paper and attached onto the surface of the tablet, starting from the center, with each hand, in alternate sequence. Participants were blind to the digital trace on screen. Drawing sessions were digitally recorded using a semi-automated in-house developed software. The hand dominance was self-reported by each participant at the start of the drawing session. Participants were free to start the session with their hand of preference.
We sampled 3000 spirals of paired hands from 1500 selected participants among the first two available spirals drawn by each participant. For the selection, we developed an ad-hoc iterative algorithm based on 15 digital spiral analysis (DSA) automated metrics derived from each digitized trace, and the age and sex of participants. The metrics captured various aspects of the tracing including tremor amplitude, acceleration, speed, and pressure. The selection algorithm maximized the variability across metrics and the coverage of participant demographics.
Spirals were then extracted and digitally printed on a bitmap at 300 dpi when displayed at actual size, with trace mark and pattern width reflecting the applied pressure on the pen. All 3000 spirals were then randomly uploaded on an online assessment form to return their TRS score between 0 and 10, by an expert neurologist.
Finally, we derived the frequency of clinical tremor rating boundaries and normative distribution percentiles according to sex, age, and drawing hand dominance. Frequencies and percentiles were calibrated to the age and sex distribution of the target population by post-stratification weights and accounted for the clustering of spirals within participant.
Results:
The median age of the present study participants was 51 (interquartile range:34-66). Sexes were balanced within the extracted sample (50.8% females). By exclusion criteria, there were 2874 spirals available for analysis. All metric distributions were slightly flattened and had higher variability, compared to their original distribution in the overall CHRIS sample. Sampled participants’ age reflected greater proportion of the elderly population compared to the CHRIS participants, consistently with extreme DSA metric values within this group.
TRS scores on the selected spirals ranged between 0 and 9 (median=2; interquartile range:2-3). Men (M) returned higher scores than women (W), while spirals drawn with the dominant hand (D) had lower scores than those drawn with the non-dominant hand (ND). Accounting for the survey design, mean TRS scores were 2.43 (95%CI: 2.38-2.49), 2.11 (2.05-2.16), 2.23 (2.17-2.28) and 1.78 (1.73-1.84) in the 4 groupings M-ND, M-D, W-ND, W-D, respectively.
The proportion of participants with TRS scores of 0 (no evidence of tremor) was particularly high among the W-D (2.38%), while the proportion of scores >4, corresponding to a clinically relevant magnitude of tremor amplitude, within the same grouping was lowest at 0.82%. The M-ND had the highest proportion of TRS scores>4 (1.64%). All spirals within the 5-th percentile and 95-th percentile considered as normative range were rated between 1 and 4, except in the W-D grouping, whose 95-th percentile was equal 3. TRS scores increased on a quadratic function with age across all 4 subgroups (adjusted Wald test for age2 p<0.001).
Conclusions:
To our knowledge, this is the first study to investigate normative ranges of TRS scores from digital spiral drawings on a large population set of predominantly healthy adult individuals, representative of their target population. Normative scores and proportions of meaningful boundaries varied between men and women and according to both age and the drawing hand dominance, with relatively more favorable values among women, younger participants and for the dominant hand. The methods are easily exportable to other population settings using a similar template to the present digital spiral analysis, even on a larger scale. Results deserve external independent validation in available population-based studies that used comparable designs and outcome.