Characterizing the Normative Voice Tremor Frequency in Essential Vocal Tremor

Key Points

Question

What is the normative frequency range in essential vocal tremor?

Findings

In a cross-sectional observational study, the normative frequency of essential vocal tremor in 160 participants was 3.8 to 5.5 Hz. Small within-patient differences (range, 0.1-0.5 Hz) were not clinically meaningful.

Meaning

The frequency of essential vocal tremor is narrow. The results of this study may better characterize this disease and its pathophysiology and assist in differential diagnosis.

This cross-sectional observational study characterizes the normative voice tremor frequency in patients with essential vocal tremor.

Abstract

Importance

Essential vocal tremor (EVT) is a neurologic voice disorder characterized by periodic fluctuations in pitch and loudness that can hinder intelligibility. Defining the normative range of vocal tremor frequency may assist in diagnosis and provide insight into disease mechanisms.

Objective

To characterize the normative voice tremor frequency in EVT (in hertz).

Design, Setting, and Participants

Cross-sectional observational study of 160 patients with EVT. The setting was a tertiary voice center. Participants were identified from a database of consecutive patients diagnosed as having laryngeal movement disorders between January 1, 1990, and April 1, 2017.

Main Outcomes and Measures

The following 3 methods measured the frequency of tremor experienced by patients with EVT: perceptual method, computerized peak detection method, and laryngeal electromyography method. Within-person and population-level tremor frequencies were compared across modalities to assess measurement reliability and consistency and to characterize the normal distribution of tremor frequencies in this population.

Results

Among 160 participants (median age, 70 years; interquartile range [IQR], 64-77 years; 90.6% female [n = 145]), the median frequency of EVT was consistently between 4 and 5 Hz across all 3 methods (perceptual, 4.8 Hz [IQR, 4.4-5.5 Hz]; computerized peak detection, 4.6 Hz [IQR, 4.2-5.0 Hz]; and laryngeal electromyography, 4.3 Hz [IQR, 3.8-5.0 Hz]). The mean in-person differences between each measurement method were not clinically meaningful (range, 0.1-0.5 Hz). Including all interquartile ranges across measurement modalities, the normative tremor frequency range for EVT was 3.8 to 5.5 Hz.

Conclusions and Relevance

To our knowledge, this is the largest study to date to characterize the normal frequency of tremor in patients with EVT. The normative frequency of EVT (range, 3.8-5.5 Hz) falls within a much narrower range than previously reported. Those whose frequency is outside this range may still have EVT but should be carefully evaluated for potential other causes of vocal tremor. Defining characteristics of EVT may aid appropriate diagnosis and improve understanding of this disease.

Introduction

Essential vocal tremor (EVT) is a neurogenic voice disorder of unknown prevalence that is characterized by periodic fluctuations in pitch and loudness as the result of involuntary rhythmic oscillations within the phonatory apparatus.^1,2,3,4,5 Respiratory, laryngeal, and articulatory muscles are variably involved.^4,6,7 It can be isolated or associated with essential tremor (10%-25%)¹ or spasmodic dysphonia (SD) (54.5%).⁸ Patients with EVT describe increased vocal effort, pitch instability, and voice breaking,¹ as well as decreased intelligibility and sounding “nervous.” These features can disorder communication and cause social isolation.^3,5,9 In fact, patients with neurologic voice disorders have more severe quality-of-life sequelae than those with inflammatory or phonotraumatic etiologies.¹⁰

Clinical diagnosis of EVT relies on auditory recognition of rhythmic vocal oscillations, which is confirmed with laryngeal visualization. The reported frequency of tremor in EVT varies between 3 and 12.6 cycles per second (in hertz).^3,9,11,12 However, our team manages a large patient population with EVT; as a result, we hypothesized that the range of tremor frequency in EVT is narrower than previously cited.

Defining the tremor frequency of EVT is important. As in other nonlaryngeal movement disorders, characterizing the normative frequency of tremor may provide insight into disease mechanisms and help differentiate EVT from other etiologies of vocal tremor.^13,14 An example is the pathognomonic 4- to 6-Hz “pill-rolling” resting hand tremor of Parkinson disease (PD)^15,16 that can be used to differentiate PD from essential tremor, drug-induced parkinsonism,^17,18 and enhanced physiologic tremor (eg, medically induced or alcohol withdrawal).¹⁹

This study aimed to characterize the normative frequency of EVT using a large cohort of affected patients. The frequency of tremor among patients with EVT was measured using the following 3 methods: (1) perceptual method, (2) computerized peak detection (CPD) method, and (3) laryngeal electromyography (EMG) method. Reliability of frequency measurements across modalities is tested using Bland-Altman plots.

Methods

This cross-sectional observational study was performed at a tertiary voice center in accord with the Declaration of Helsinki,²⁰ and Good Clinical Practice applicable regulatory requirements and was approved as a minimal list study by Vanderbilt University Medical Center’s institutional review board. The requirement for informed consent of participants was waived.

Patient Population

Participants were identified from a database of consecutive patients diagnosed as having laryngeal movement disorders between January 1, 1990, and April 1, 2017, at the Vanderbilt Voice Center, Nashville, Tennessee.²¹ To be included, patients had to (1) be 18 years or older, (2) have a vocal tremor diagnosed perceptually and confirmed with flexible laryngoscopy, (3) have no concomitant SD, (4) have an evaluable voice sample with sustained vocal phonation, and (5) have had no prior onabotulinum toxin A injection. Sustained vowel sounds (ie, /i/, /u/, and /a/) were recorded at initial voice evaluation. Patients were asked to produce the vowel sound at a comfortable pitch and loudness level. The audio samples were recorded at 44 100-Hz sampling frequency.

Perceptual and acoustic characterization of EVT is best achieved during a sustained phonation task.^3,4 Voice samples of all patients clinically diagnosed as having EVT were reassessed by both a laryngologist (B.L.H.) and voice-specialty speech-language pathologist (C.P.) to ensure that EVT occurred without the presence of concomitant SD or other laryngeal movement disorder. Evaluable audio samples had minimal background noise, clear signals, and audible tremor during sustained phonation.

Frequency Measurement

Perceptual Method

Each extracted audio sample was independently reviewed and analyzed at least 3 times by 2 of us (C.P. and B.L.H.). Discrete oscillations during the sustained vowel sample were manually counted and divided by the duration of the audio sample. For example, if investigators counted 17 oscillations in 3.8 seconds, this translated into a frequency of 4.5 Hz. Discrepancies were resolved by consensus; if no consensus was achieved, the participant was excluded.

CPD Method

Data processing of extracted, sustained vowel samples was conducted using a software program (MATLAB; MathWorks). Speech amplitude envelopes were extracted using a Hilbert transform–based method and downsampled to 50 Hz. Local maxima of the envelopes were detected using the peak-finding function (https://terpconnect.umd.edu/~toh/spectrum/). Peaks were assessed by downward zero-crossings in the smoothed first derivative using a pseudo-gaussian smoothing algorithm. A peak was classified as a group of points with amplitude that exceeded the amplitude of neighboring points on either side. The rate of tremor was calculated by the total number of peaks divided by the duration of the audio sample (Figure 1).

Figure 1. CPD Recording.

m(B) is the amplitude of oscillation; 1 through 6 indicate each detected peak in this recorded sample. CPD indicates computerized peak detection.

Laryngeal EMG Method

Laryngeal EMG (VikingQuest EMG System, version 21.1.0.195; Nicolet) was used to localize the thyroarytenoid/lateral cricoarytenoid muscle complex in patients during onabotulinum toxin A injection. Muscle activation in the thyroarytenoid/lateral cricoarytenoid was recorded and measured when the participant sustained a vowel sound /i/. Tremor activity was recorded by 1 pair of surface electrodes and 1 needle electrode. All available EMG tracings derived from localization before injection. Two of us (C.P. and B.L.H.) independently reviewed each captured EMG tracing. Tracings without discrete peaks were excluded. Discrete peaks were totaled and divided by the duration of each sample (Figure 2). Discrepancies were resolved by consensus.

Figure 2. Laryngeal EMG Recording.

m(B) is the amplitude of oscillation; the distance between the vertical plot points in this scale bar is measured in 200 milliseconds (ms); 1 through 6 indicate each discrete peak counted and then divided by the duration of the sample; and T1 and T2 are simple boundaries of this recorded EMG sample and do not affect calculation. EMG indicates electromyography.

Statistical Analysis

Patient characteristics and method-specific tremor frequency (perceptual, CPD, and EMG) were summarized using descriptive statistics. Frequency data were analyzed using both parametric (mean and 95% CI) and nonparametric (median and interquartile range [IQR]) measures of central tendency. Comparison between measurement methods was done using Bland-Altman plots, which quantify agreement between 2 quantitative measurements using the mean difference and constructing limits of agreement.²² In other words, Bland-Altman plots assess if there is a systematic difference between 2 different measurement methods that are measuring the same construct. In so doing, the mean difference in frequency (in hertz) was calculated between each method pair (ie, perceptual-CPD, perceptual-EMG, and CPD-EMG). The mean difference was plotted against the combined mean of comparators (ie, perceptual plus CPD divided by 2). The result is a scatterplot xy, in which the y-axis shows the mean difference in hertz between paired measurements and the x-axis shows the combined mean hertz. Statistical analyses were performed using Stata 12MP (StataCorp LP) and the R package (R Foundation for Statistical Computing).

Results

Of 832 patients treated for laryngeal movement disorders, 160 patients (median age, 70 years [interquartile range, 64-77 years]; 90.6% female) met inclusion criteria (Table 1). In all, 69 patients (43.1%) had a diagnosis of essential tremor (eg, head or hand) at initial presentation, and 49 patients (30.6%) were taking tremor-modulating medications (eg, β-blockers, benzodiazepines, or baclofen). Among those meeting inclusion criteria, 56 (35.0%) had an evaluable EMG tracing. Reasons for study exclusions are shown in Figure 3.

Table 1. Characteristics at Presentation Among 160 Patients.

Variable	Value
Age, y
Median	70
IQR	64-77
Female, No. (%)	145 (90.6)
Race/ethnicity, No. (%)
White	137 (85.6)
Black	7 (4.4)
Other/unknown	16 (10.0)
Essential tremor, head or hand, No. (%)	69 (43.1)
Patients taking tremor-modulating medications, No. (%)a	49 (30.6)

Abbreviation: IQR, interquartile range.

^aβ-Blockers, benzodiazepines, baclofen, and others.

Figure 3. Study Flow Diagram.

CPD indicates computerized peak detection; EMG, electromyography.

Tremor Frequency by Measurement Method

All methods measured tremor frequency modalities during sustained vowel phonation (Table 2). The same voice samples were used for perceptual and CPD and were a median of 2.3 (IQR, 2.0-3.0; range, 0.8-7.0) seconds in duration, while tracing used for EMG evaluation had a median duration of 1.4 (IQR, 1.2-1.6; range, 0.6-3.2) seconds. eFigure 1 in the Supplement compares tremor frequency between the 3 measurement methods. Overall, the median frequency was highest for perceptual (4.8; IQR, 4.4-5.5 Hz), intermediate for CPD (4.6; IQR, 4.2-5.0 Hz), and lowest for EMG (4.3; IQR, 3.8-5.0 Hz). Patients taking tremor-modulating medications had slightly lower median perceptual tremor frequency (4.5; IQR, 4.2-5.0 vs 4.8; IQR, 4.4-5.5 Hz), but frequencies did not differ by medication use using other measurement methods.

Table 2. Tremor Frequencies by Measurement Method.

Variable	Median (IQR), Hz	Mean (95% CI), Hz
Perceptual method	4.8 (4.4-5.5)	5.0 (4.8-5.1)
CPD method	4.6 (4.2-5.0)	4.6 (4.5-4.7)
EMG methoda	4.3 (3.8-5.0)	4.3 (4.1-4.5)

Abbreviations: CPD, computerized peak detection; EMG, electromyography; IQR, interquartile range.

^aAmong 56 patients.

Bland-Altman Plots

eFigure 2 in the Supplement shows Bland-Altman plots comparing within-person differences between each measurement method. Perceptual and CPD tremor frequency measurement differences showed a mean difference of 0.3 Hz (95% limits of agreement, −1.1 to 1.8 Hz) (eFigure 2A in the Supplement). Within-person mean differences in measurement were 0.5 Hz (95% limits of agreement, −1.4 to 2.3 Hz) between perceptual and EMG measurements (eFigure 2B in the Supplement) and 0.1 Hz (95% limits of agreement, −1.7 to 2.0 Hz) between CPD and EMG measurements (eFigure 2C in the Supplement).

Discussion

This is the largest study to our knowledge to objectively measure the frequency of EVT. Three distinct and readily accessible methods were used to confirm measurement consistency. The results indicate that the range of tremor frequency in EVT is narrow and consistent across measurement methods. The median frequencies were all between 4 and 5 Hz. The variance around these central tendencies was consistently narrow. In fact, accounting for the IQRs for all modalities, the normative frequency range for EVT is 3.8 to 5.5 Hz. Bland-Altman plots confirmed clinically insignificant within-person differences in tremor frequency. Perceptual evaluation, CPD, and EMG all provided comparable and consistent ratings within and across patients with EVT. This finding is important because it helps to better define a pathognomonic disease characteristic of EVT that should aid clinicians in differentiating it from other causes of voice tremor or other movement disorders.

The results herein refine those previously reported. Our cohort was predominantly older, female, and of white race/ethnicity, which is consistent with prior literature.^1,23 Previous small case series (ie, 4-23 participants) described the EVT frequency as falling between 3 and 12.6 Hz.^3,9,11,12 Each used a variety of instrumental approaches to objectively measure tremor frequency. Methods used included an oscilloscope,^3,11 computer analysis,¹¹ strip chart recordings,⁹ and the vocal demodulator.¹² One study² analyzed perturbation measures in 28 patients with EVT but did not capture frequency because of technological limitations.

An important confounder in most prior studies relates to heterogeneity of patients. Studies^11,12 included patients with PD, essential tremor, and SD. These conditions can be associated with voice breaks. For example, in a cohort of 8 patients with PD, vocal tremor was observed in half of them.¹³ It is not known if coexistent SD with EVT affects tremor frequency.^11,12 In the present study, we tried to achieve homogeneity by excluding patients with PD or concomitant SD.

In our subanalysis of patients taking tremor-modulating medications, the measured tremor frequency was not significantly different. Further work is needed to assess if these medications change other tremor variables (eg, amplitude) or if a subjective change correlates with changes in objectively measured patients with tremor.

Limitations

While this study represents the largest series to our knowledge to measure EVT frequency using EMG, approximately two-thirds of patients (n = 104) did not have evaluable EMG tracings. The within-patient frequency of tremor identified with EMG was clinically indistinguishable from the other methods used (ie, perceptual and CPD). Within-patient differences between measurement methods existed (range, 0.1-0.5 Hz); however, these were small and not deemed clinically meaningful. Therefore, we are confident that the normal frequency reported is accurate. The studied EVT population was ultimately treated with onabotulinum toxin A. Selection bias is possible because patients not considered botulinum treatment candidates may systematically differ in tremor frequency range.

Conclusions

This study is the first to our knowledge to reliably characterize the normal frequency of tremor in patients with EVT. The normative tremor frequency 95% CI is narrow (range, 3.8-5.5 Hz). Those whose frequency is outside the 95% CI may still have EVT but should be carefully evaluated for other potential causes of vocal tremor. This study establishes the expected frequency of EVT, which defines part of the disease process, as has been done historically with PD. Understanding the normal tremor frequency may help clinicians to better identify EVT and differentiate it from other causes of vocal tremor.

Supplement.

eFigure 1. Box Plots (Median, Interquartile Range) Comparing Frequency of Tremor Measured Using Perceptual, Computerized Peak Detection (CPD), and Electromyographic Methods

eFigure 2. Bland-Altman Plots Comparing Within-Person Mean Differences Across Measurement Methods: A) Perceptual vs CPD, B) Perceptual vs EMG, C) CPD vs EMG