Absolute or relative voice rest after phonosurgery: a blind randomized prospective clinical trial

Abstract

Objectives: The first aim, was to compare participant compliance with postoperative voice rest advice in two groups. The second aim was to compare vocal function and recovery in the short-term, seven days post-surgery and in the long-term, 3–6 months post-surgery.

Design: Preliminary randomized prospective blind clinical trial.

Methods: Twenty patients scheduled for surgery for benign vocal fold lesions were randomized into seven days of absolute or relative voice rest. Compliance with voice rest advice was monitored with a voice accumulator for seven days following surgery. Vocal recovery was tracked through (a) self-perceived vocal function, (b) perceptual assessments of voice recordings and (c) visual assessment of high resolution and high speed digital imaging (d) vocal stamina and reaction to vocal loading, explored with a vocal loading task.

Results: The absolute voice rest group phonated significantly less than the relative voice rest group during seven days post-surgery, but they were not silent. The absolute voice rest group self-reported more difficulty with compliance than the relative voice rest group. The relative voice rest group coped with significantly more vocal loading at long-term check-up. In the short-term the absolute voice rest group improved morphological recovery to a significant degree, however relative voice rest renders superior long-term recovery.

Conclusions: Absolute voice rest is difficult to comply with. Neither short-term, nor long-term vocal recovery differed significantly between groups. Within-group comparisons showed significant improvements in vocal stamina, immediate recovery from vocal loading and self-assessments of voice problems only for the group with recommended relative voice rest.

Keywords:

• Voice rest
• phonosurgery
• vocal recovery
• voice accumulation
• vocal loading task

Introduction

Voice rest is one predictor of treatment outcome in patients undergoing vocal surgery [1]. There is, however, a lack of studies examining how microscopic changes to vocal fold tissue affect general vocal function [1] and the patients’ general wellbeing [2] and there is still no standard protocol for providing voice rest advice [3]. A limited number of recent studies have provided clinical evidence for brief absolute voice rest (3–5 days) following surgery of benign lesions in the vocal folds [4,5], but few quantify differences between absolute and RVR [6]. Absolute voice rest is a traditional clinical routine following surgery for benign lesions of the vocal folds, based on the hypothesis that re-epithelization occurs within 7 days [1,7], which in turn is based on a model of recovery in canine vocal folds [8]. No studies present results which clarify the rationale for absolute voice rest, compared to RVR, which is also administered clinically [1,3,7]. Indeed in 2012, Roy noted that so far we do not know anything about the dose–response relationship in recovery of the lamina propria and underlying structures [9].

Absolute voice rest, i.e. complete silence, not speaking or using the voice for any other purpose, is the most common advice for vocal rehabilitation following phonosurgery [3]. Thus far, no study has rationalized how or why patients are expected to comply with absolute voice rest following phonomicrosurgery. In fact recent studies, comparing longer silence periods (7–10 days) to shorter periods (3–5 days), concur that shorter rest is more purposeful [4,5], although the longer rest prevails as clinical routine. When looking at functional recovery of knee ligament injury in canines, it is suggested that early mechanical stimulation improves function through stimulation of collagen synthesis [10]. A problem with this comparison, pointed out by Ishikawa and Thibeault [1], is the fact that vocal fold tissue needs to be pliable and viscous to enable mucosal vibration, whilst orthopedics is dependent on a robust, stiff scarring for stress–strain stability. However, they also point out that increased collagen synthesis induced by movement could encourage vocal fold fiber to heal in the direction of movement, preventing scar tissue which could hamper mucosal vibration. This hypothesis is backed by Titze et al. [11], who have shown biological changes in fibroblasts in healing vocal folds, when put in movement. Furthermore, when following the course of wound healing in the vocal folds, Kaneko et al. [5] saw better improvement in vocal function in participants who had early onset voice therapy 3 days following surgery, i.e. during the end of inflammation, before re-epithelialization [10]. This notion is supported by Li and Verdolini et al. [12,13], who suggest controlled vocal exercises may be better than absolute voice rest and spontaneous speech for recovery.

This preliminary study has two aims, comparing two randomized groups following surgery of benign lesions in the vocal folds:

1. Track participant compliancy with absolute and RVR advice during 7 days.

2. Compare short and long-term vocal recovery of voice function, including vocal stamina.

The hypotheses are that (1) absolute voice rest is more difficult to comply with than RVR, and that vocal recovery will be better in the group following RVR and (2) the voice function of patients following RVR will be better than the absolute voice rest group at long-term check-up. These hypotheses are owed to clinical experience, as no such studies have previously been reported in the literature.

Materials and methods

Twenty (n = 20) participants took part in this study. Inclusion criteria were patients of 18 years or older, planned for surgical treatment of benign organic lesions in the vocal fold mucosa. Exclusions criteria were malignancies, lesions affecting muscle fiber, laryngeal papilloma, and neurological voice pathologies. Patients were orally requested for participation by different surgeons in the phoniatric clinics in Lund, Malmö, and Helsingborg in southern Sweden. All accepting participants gave written consent for participation. Participants were informed of their right to terminate the study at any given time, without stating the reason for termination. Surgery was routinely performed by three phoniatricians/phonosurgeons through cold excision ad modum Bouchayer and Cornut [14], without use of fibrin glue in Reinke’s edema. According to local clinical routine, participants were not given any voice therapy before or after phonosurgery (apart from one participant, a singer, who wanted further assistance finding her new voice). Surgery and data collection were carried out in these clinics during a 2.5-year period from September 2013 to April 2016. Of all patients meeting inclusion criteria (86), 23% (n = 20) participated, see Table 1. Also according to clinical routine, participants were given sick leave for the duration of recovery (7 days) if their work prevented them for participating in the study. Participants were not told absolute rationale for the research project, only that recovery was monitored following surgery in different groups. This study was approved by the Regional Ethical Review Board in Lund, Sweden on April 25, 2013 (#2013/174).

Table 1. Participant characteristics.

	Absolute voice rest (AVR)	Relative voice rest (RVR)
Participants	10	10
Female	7	7
Male	3	3
Age, mean (range, SD)	53 (35–68, 12.2)	45 (25–73, 16.1)
Vocal pathology and operating technique	Reinke’s edema (2 unilateral, 2 bilateral) Cyst granuloma (1 unilateral with excised granuloma) Polyp (4 unilateral, whereof 1 hemorrhagic) Hyperplastic squamous epithelium (1)	Reinke’s edema (2 unilateral, 1 bilateral) Cyst (3 unilateral, whereof 1 intraplical) Polyp (2 unilateral, whereof 1 hemorrhagic) Sulcus (2 unilateral, blunt dissection)
Occupation	Retired (3), school teacher (2), chef (1), consultant (1), office worker (1) pharmacist (1), retail clerk (1)	Retired (2), singer (2), school teacher (1), singing teacher (1), chef (1), consultant (1), office worker (1), deli clerk (1)
Medical status	Smoker (4), depressive (3), high blood pressure (2), chronic obstructive pulmonary disease (1), hyperthyroidism (1), reflux (2), ulcerative colitis (1), hearing impairment (0)	Smoker (3), high blood pressure (2), depressive (1), multiple sclerosis (1), reflux (1), hearing impairment (0)
Voice therapy	Preoperative (0) postoperative (0)	Preoperative (0) postoperative therapy (1 singer)
7 day voice rest advice	Let your voice heal by being completely silent. Do not use your voice AT ALL. Do not whisper, instead use gestures or writing for communicating. Please note that the voice is often used for more than speaking, e.g. when coughing, throat clearing and during physical activity. Do not use your voice at all	Let your voice heal by speaking with a gentle, comfortable voice for the duration of 7 days. Do not whisper, do not use loud voice for speaking or shouting and do not sing

Block randomization [15] was made by preparing 20 sealed envelopes containing one of two sets of voice rest advice: n = 10 absolute voice rest (AVR) and n = 10 RVR, see Table 1 for the exact phrasing. The envelopes were mixed into random order by a person external to the research project, and handed out from the top of the batch by each operating surgeon for each participant. This study was blinded. Voice rest advice envelopes were opened by each participant after surgery and no surgeons or test leaders knew which participants had received which voice rest advice until after short-term check-up measurements were taken after 7 days. Figure 1 shows a flow chart describing the process of data collection over time. One participant from each randomized group did not participate in the long-term check-up.

Figure 1. Process flow chart of data collection.

Compliance to voice rest advice was quantified though long-time voice measurement, using a portable voice dosimeter as suggested by Misono et al. [16] (VoxLog by SonVox AB, Umeå, Sweden, as described by Van Stan et al. [17]). The outcome measure for compliance was phonation time (percentage relative to total measurement) during the 7 days following surgery. Compliance to AVR should thus render a 0% phonation time. The accelerometer and microphone were placed in a neck collar. The microphone does not correct to a standard distance, thus sound pressure level is reported ad modum Södersten et al, ie, ca 7 dB higher compared with measurements taken at a 30-cm distance, however phonatory intensity was not a primary outcome in this study. Participants’ written self-assessment of compliance was qualitatively evaluated in voice activity questionnaires accompanying the long-time voice measurement. The questionnaire evaluates self-assessed general voice problems on a 100-mm visual analogue scale (VAS: 0 = none, 100 = maximal) and specific voice problems through eight underlying voice health questions and 10 voice health questions on shifting/fluctuating symptoms, based on VHI-T [18], as adapted by Whitling et al. [19]. The eight underlying voice health questions were: (1) I run out of breath when I speak, (2) My voice problems affect my private economy, (3) My voice problems restrict my private and social life, (4) My voice makes it hard for others to hear what I am saying, (5) Others ask me what is wrong with my voice, (6) I feel handicapped because of my voice, (7) I feel left out of conversations because of my voice, (8) I am worried by my voice problems. The 10 shifting voice health questions were: (1) This is my current stress level, (2) My voice feels fatigued, (3) I need to clear my throat, (4) I need to cough, (5) My throat/neck (same word in Swedish: “hals”) feels tense, (6) I am hoarse, (7) I am having a hard time making myself heard (like at a party), (8) My voice can suddenly change when I speak. (9) It is effortful to get my voice working, (10) I have a feeling of discomfort in my throat/neck.

In order to follow recovery from vocal surgery and from heavy vocal loading, participants filled out VAS and shifting questions four times per day throughout 7 days’ voice accumulation. Underlying questions were not expected to shift throughout the day, thus they were filled out every evening. For analyses, mean scores for each self-assessment parameter were pooled over 7-day periods. Immediate reaction to heavy vocal loading (see below) and recovery therefrom was tracked by general VAS self-assessments, filled out at seven time points around each vocal loading task (VLT): just before and after a VLT, every 15 min for an hour and again 2 h after the VLT. VAS scores were standardized and calculated ad modum Atkinson et al. [20], as applied by Vogel and Maruff [21] and Whitling et al. [22], by neutralizing baseline to 0 and dividing the difference between true baseline and subsequent scores with the within-group standard deviation at each time point. Long-time voice accumulation was repeated for 7 days at long-term check-up. In order to prevent bias, the group receiving AVR were instructed to desist recording VAS scores and on shifting voice questions desisted after the VLT, as they were not supposed to phonate at all during the first week of recovery.

Vocal stamina was assessed by means of a VLT, ad modum Whitling et al. [22,23]. The method examines changes to vocal function caused by loud phonation through loud reading, seated in a double-walled soundproof booth that complies with the maximum permissible ambient SPL as specified in ISO 8252–1. Ambient multitalker speech babble aired in free field at 85 dBA. The time (maximum 30 min, unknown to participants) participants managed to stay in the VLT relates to mechanical and psychological coping with loud speech and the test situation. Participants set the time limit for the VLT and were told to terminate the task when and if they felt distinct discomfort from the throat, which lead to differing participation times. VLTs were carried out twice: once a couple of days prior to surgery and again at long-term check-up.

Perceptual voice assessments of voice recordings and assessments of digital imaging, both collected pre and post VLT, were carried out ad modum Bless et al. [24], as adapted by Lyberg Åhlander et al. [25], by one panel of two experienced speech and language pathologists and another panel of three experienced phoniatricians. Both panels were blind to any underlying project information, assessing in consensus. The setup for perceptual assessments of voice recordings and visual assessment of digital imaging as Whitling et al. [22] was applied. This entailed an expert panel of three experienced phoniatricians who blindly assessed the following by consensus: (1) digital imaging (high-resolution mode for overall morphology, symmetry, and adduction and abduction of the vocal folds); and (2) high-speed mode (4000 frames/s) for mode and symmetry of vibratory patterns at the glottal level on imaging performed by the last author pre- and post VLT for each participant. The three experts assessed the following on a four-point scale (0 = unaffected, 1 = slightly affected, 2 = moderately deviant, and 3 = highly deviant): glottal shape, glottal regularity, morphological alteration, adduction of the right vocal fold, adduction of the left vocal fold, abduction of the right vocal fold, abduction of the left vocal fold, wave amplitude of the right vocal fold, wave amplitude of the left vocal fold, wave propagation of right vocal fold, wave propagation of the left vocal fold, phase difference (i.e. difference in wave amplitude and propagation of the vocal folds), right vestibular fold, left vestibular fold, corniculate tubercle symmetry during phonation, and corniculate tubercle symmetry during rest. Another expert panel of two experienced speech and language pathologists by consensus assessed perceptual changes in the signal of the passage The North wind and the sun (ca 45–60 s) ad modum Stockholm Voice Evaluation Approach, using a 100-mm VAS, manually with pen and paper (0 = unaffected, 100 = deviant in the greatest possible manner). Voice recordings were made in the aforementioned sound booth, with subjects sitting down, wearing a head-mounted microphone (MKE 2, no 09_1, Sennheiser GmbH & Co, Hanover, Germany), calibrated at 94 dB SPL, at a distance of 15 cm from the mouth (converted to 30 cm). The voice signal was digitized at 16 kHz with 16 bit resolution. The signal was automatically retrieved and examined for fundamental frequency and speaking SPL, using Soundswell Core 4.0 and Soundswell Voice 4.0 (Hitech development AB, Täby, Sweden).

The two experts blindly assessed the following vocal parameters: hyperfunction or press, breathiness, instability, roughness, glottal fry, sonority (reversed outcome scale), and grade of the overall voice pathology. To minimize the instrumental variation effect, each trial was completed using the same technical equipment, calibration protocol, test leader (first author), laryngologist performing the laryngeal digital imaging (last author), and expert panels of voice professionals. New recordings and digital imaging were performed once at short-term check-up and again twice, pre and post reiterated VLT procedure at long-term check-up.

Statistical analyses and group comparisons

Statistical analyses were performed with IBM SPSS Statistics 23 (SPSS Inc., Chicago, IL). Samples with interval output were explored for normal distribution and statistical analyses were chosen accordingly. Ordinal data were explored using nonparametric analyses only. A statistical power analysis based on self-assessments in Whitling et al. [23], showed that in order for a Cohen d effect size difference of self-assessment of voice problems of .5 to be detected at a 80% chance as significant at the alpha level .05 [26], a sample of at least nine participants in each compared group was needed. In order to reduce risk of alpha inflation, multiple measures were taken into account using strict Bonferroni adjustment. All a priori analyses were carried out to fit the data, i.e. ordinal data were compared using non parametric test for comparisons. Within- and between-group comparisons were made a priori, based on affiliation to randomized voice care advice: AVR versus RVR. Post hoc tests were run based instead in a division of all participants, based on how much they in fact had phonated during voice rest week: low phonation time versus high phonation time. The cut-off was made at the median phonation time among all participants.

Results

The RVR group phonated significantly more (range 0–28%) than the AVR group (range 0–14%) during voice rest week. Despite AVR advice the AVR were not completely silent. Phonatory sound pressure levels measured with VoxLog were compared between groups during voice rest week. Results showed no significant group differences (AVR mean 82 dB SPL, RVR mean 86 dB SPL). Table 2 shows results from both sets of long-time voice accumulation (relative phonation time and mean self-assessment scores, pooled over 7 days).

Table 2. Results in two groups (AVR = absolute voice rest, RVR = relative voice rest) from long-time voice accumulation during seven days following surgery (“voice advice week”) and the procedure re-iterated at long-term check-up.

		AVR			RVR			Group comparison Mann–Whitney U test
		Voice rest week	Long-term follow-up	Wilcoxon signed rank test	Voice rest week	Long-term follow-up	Wilcoxon signed rank test	Voice advice week		Long-term follow-up
Relative phonation time (%)	Mean	3	11	Z = −2,701 p = .007*	7	9	Z = −1.886 p = .059 (NS)	U = 20.000 p = .023*		U = 30.000 p = .354 (NS)
	Median	1	12		5	8
	SD	4	5		8	3
	95% CI	[−.4003, 5.4483]	[7.3144, 15.1478]		[1.5709, 12.8631]	[6.6604, 11.9041]
Pooled self-assessment VAS (100 mm)	Mean	–	21	–	54	15	Z = −2.668 p = .008*	–		U = 32.000 p = .700 (NS)
	Median	–	13		55	10
	SD	–	21		19	16
	95% CI	–	[2.78, 38.22]		[39.17, 69.05]	[2.62, 27.83]
Self-assessment SVQ (40 p)	Mean	–	11	–	17	8	Z = −2.530 p = .011*	–	U = 26.500 p = .359 (NS)
	Median	–	10		19	5
	SD	–	5		8	7
	95% CI	–	[6.53, 14.72]		[12.67, 23.93]	[2.75, 13.92]
Pooled self-assessment UVQ (32 p)	Mean	7	5	Z = −1,268 p = .205 (NS)	13	5	Z = −2.383 p = .017*	U = 20.00 p = .043 (NS)		U = 34.500 p = .883 (NS)
	Median	8	2		11	3
	SD	5	6		8	6
	95% CI	[3.16, 10.40]	[−.06, 10.31]		[7.68, 19.12]	[0.65, 9.35]

*Significant at alpha level p ≤ .025, following strict Bonferroni correction. Unable to compare, as AVR group was silent.

Nine (n = 9/10) patients in the AVR group self-reported evident difficulty with complying with voice rest advice. In the RVR group, n = 3/10 patients had some difficulty with compliance. There were reports of practical non-compliance, i.e. speaking or shouting in both groups. Table 3 gives a comprehensible picture of how voice rehabilitation affects the wellbeing of patients undergoing phonosurgery. Participants F3, F11, and M2 in the AVR group reported distinct trouble with compliance as AVR affected their general wellbeing. Participant F2 in the RVR group reported trouble with the voice rest advice being too vague.

Table 3. Patients’ written self-reports on how they experienced complying with their voice rest advice combined with relative phonation time (%) during 7 days’ voice accumulation following surgery in two groups: AVR = absolute voice rest, RVR = relative voice rest.

AVR			RVR
Participant	Written self-report	Relative phonation time (%)	Participant	Written self-report	Relative phonation time (%)
F1	Difficult, because you always have to write. I wanted to speak.	1	F2	I have followed them. However, I wish they would have been more specific. It is rather scary to undergo surgery, to have a really sore throat and be told to speak without knowing whether that entails 1 minute per day or as I usually speak. I have spoken softly and tried to find my pain threshold and tried not to exert my voice. As it has started feeling better I have increased my speaking	4
F3	At first I was pleased to receive the advice, because I thought it would be good to do differently than after previous surgery. It was VERY difficult during the weekend, especially with my 4 year old son who wants his mother. I felt as though I would have become really depressed if the period had been longer. The induced silence made it clear to me how much I normally use my voice	0	F5	Because I have a 3 year old who is very active and mischievous at any chance I sometimes forgot not to shout. During the first couple of days I whispered, because the soft, comfortable voice was difficult to find	4
F4	Difficult to be quiet	0	F6	I think it went well. I have been very careful	3
F7	Difficult	14	F8	It was OK	9
F9	Been mostly silent. Been using text messaging	2	F10	No problem with the voice rest advice	28
F11	It’s been unusual! Not to be able to take part in conversations, respond, make myself understood quickly has sometimes been frustrating. My family yearns for me to speak again! Been out in public even though I was not allowed to speak. An exciting experience and I am thankful to have my voice back! I look forward to regaining a perfectly tuned instrument. The pleasure of spreading joy with my singing is just around the corner! Finally!! Worried to use my voice “incorrectly”!	0	F12	It went well. But every now and then I forgot not to speak with a loud voice	6
F14	I’ve been talking	3	F13	No problem with advice, my problem was my cold	1
M2	It is tough not being able to speak to my dogs – they don’t understand this!	2	M1	Went well	8
M5	Difficult! I have not been able to follow it completely!	1	M3	No problem	0
M6	Almost impossible to be completely silent. Became increasingly more difficult towards the end of the period	2	M4	No problem, but somewhat strange, because my voice felt OK straight away	11

There were no significant group differences for any self-assessments of voice problems carried out before surgery, at short-term check-up or at long-term check-up. Nor were there any significant group differences at these time points for overall perceptual assessments, overall visual assessments, or for participation time in the VLT. Visual overall assessments in the AVR group were explored post-hoc with Wilcoxon’s signed rank test, showing significant improvement at short-term check-up (Z = −2.246, p = .025). No post hoc analyses for RVR were required. Overall visual assessments from digital imaging improved for both groups. Individual parameters in visual and perceptual assessments from before surgery were compared with Wilcoxon’s signed rank test to results at long-term check-up, which showed overall improvement in glottal shape (Z = −3.245, p = .001), regularity (Z = −2.98, p = .003) and over all morphology (Z = −3.448, p = .001), press (Z = −2.329, p = .02), breathiness (Z = −3.007, p = .003), and sonority (Z = −3.071, p = .002). There were no significant group differences, nor were there any significant changes in individual assessment parameters caused by the VLT’s themselves. For results before surgery compared to short-term check-up and at long-term check-up, see Table 4.

Table 4. Results from self-assessments of general voice problems (100 mm VAS), 10 questions on shifting voice problems (SVQ), 8 questions on underlying voice problems (UVQ), overall points from visual assessments (max 48), overall perceptual assessments (100 mm VAS) and participation time in a vocal loading task in Absolute (AVR) and Relative (RVR) voice rest groups before surgery, at short-term and at long-term follow-up.

		AVR					RVR
		Before surgery	Short-term follow-up	Long-term follow-up	Within group^a/b	Between measure-ments, post-hoc (i), (ii) and (iii)^b	Before surgery (i)	Short-term follow-up (ii)	Long-term follow-up (iii)	Within group^a	Between measure-ments, post-hoc (i), (ii) and (iii)^b
Self-assessment (100 mm VAS)	Mean	56	–	21	Z = −1.859, p = .063^bNS		53	54	15	Χ^2 = 8.222, p = .16^a*	i to ii: Z = −.561, p = .575^bNSi to iii: Z = −2.312, p = .021^b** ii to iii: Z = −2.668 p = .008^b**
	Median	47	–	13			45	55	10
	SD	26	–	22			23	19	16
	95% CI	[36.68, 72.21]	–	[2.78, 38.22]			[35.48, 70.52]	[39.17, 69.05]	[2.62, 27.83]
Self-assessment SVQ (40 p)	Mean	21	–	15	Z = −2.524, p = .012^b**		20	17	8	Χ^2 = 10.800, p = .005^a*	i to ii: Z = −1.531, p = .126^bNSi to iii: Z = −2.433, p = .015^b** ii to iii: Z = −2.530, p = .011^b**
	Median	19	–	15			19	19	5
	SD	8	–	6			5	8	7
	95% CI	[14.52, 27.26]	–				[16.14, 24.30]	[12.67, 23.93]	[2.75, 13.92]
Self-assessment UVQ (32 p)	Mean	10	–	3	Z = −2.392, p = .017^b’		15	13	5	Χ^2 = 10.000, p = .007^a*	i to ii: Z = −1.225, p = .221^bNSi to iii: Z = −2.677, p = .007^b** ii to iii: Z = −2.383 p = .017^b**
	Median	8	–	3			16	11	3
	SD	7	–	3			7	8	6
	95% CI	[4.74, 15.93]	–				[9.06, 20.05]	[7.68, 19.12]	[0.65, 9.35]
Perceptual assessment (overall 100 mm VAS)	Mean	55	40	18	Χ^2 = 4.667, p = .097^aNS	–	28	42	18	Χ^2 = 4.065, p = .131^aNS	–
	Median	53	29	5			16	39	4
	SD	31	36	24			32	38	24
	95% CI	[30.83, 78.95]					[3.63, 52.81]	[13.35, 71.32]
Visual assessments (overall 48 p)	Mean	15	8	7	Χ^2 = 10.800, p = .005^a*	i to ii: Z = −2.246, p = .025^b** i to iii: Z = −2.666, p = .008^b** ii to iii: Z = −1.944 p = .052^bNS	11	10	7	Χ^2 = 3.353, p = .187^aNS	–
	Median	17	7.5	5			11	8	6
	SD	6	5.4	5			5	6	3
	95% CI	10.466, 19.312]					[7.728, 14.939]	[5.654, 14.124]
Time in VLT (min)	Mean	16	–	17	Z = −.105, p = .916^bNS		18	–	26	Z = −2.371, p = .018^b**
	Median	12	–	13			15	–	30
	SD	10	–	10			9	–	6
	95% CI	[8.58, 23.22]	–				[11.22, 25.00]	–	[20.83, 30.73]

NS: No significant within-group differences. *Significance at alpha level p ≤ .017, using Bonferroni correction for a priori test using ^aFriedman’s test of repeated measures. **Significance at alpha level p ≤ .025, using Bonferroni correction for post hoc test ^bWilcoxon signed rank test. – Unable to compare, as the AVR group was silent.

Vocal stamina, partly assessed through participation time in a VLT before surgery and at long-term check-up is presented in Table 4. Stamina was also assessed through reaction to vocal loading and immediate recovery course following each VLT, measured with standardized scores from 100 mm VAS. Figure 2 shows these reactions and immediate recovery courses, with the RVR group reacting more than others in the VLT at long-term check-up and joining up at 30 min after the VLT. As these results show standardized data, presenting relative change in self-assessment, statistical analyses are not applicable. Visual and perceptual assessments, as well as acoustic analyses were carried out on digital imaging and voice recordings (LTAS, speech range profiles and phonatory sound pressure level) collected pre and post VLT at both occasions. There were no significant differences between the groups for any of these parameters at any occasion, nor were there any significant group differences for participants’ self-assessments of general (VAS) and specific (SVQ, UVQ) voice problems pre to post VLTs.

Figure 2. Immediate recovery with standardized self-assessment scores from 100 mm visual analogue scale (VAS) in the absolute voice rest (AVR) group and the Relative voice rest (RVR) group, following two vocal loading tasks: on performed before surgery, the other at long-term check-up.

Post hoc analyses based on real phonation time

When reanalyzing the data, dividing participants into two groups, based on whether or not they had actually phonated during the week following surgery or not, results were very similar. Putting a cut-off between groups based on real phonatory behavior, instead of on received voice care advice gave two groups based on high or low phonation time during the voice advice week. The cut off was set at the median phonation time for both groups put together =3%. ≥3 = high phonation time, <3= low phonation time. These new groups held the same gender distribution as the original grouping. N = 4 participants changed groups in this new constellation. New group configuration is described in Table 5.

Table 5. Group configuration when distinction was made according to actual phonation time during recovery week. Low phonation: n = 10, high phonation n = 10, cut-off at between group median (2.79% phonation time).

	Low phonation time (<3%)	High phonation time (≥3%)
Participants	10	10
Female	7	7
Male	3	3
Age, mean (range, SD)	48 (25–65, 14.6)	49 (25–73, 15.3)
Vocal pathology	Reinke’s edema 4 (2 unilateral, 2 bilateral) Cyst 2 (unilateral, whereof 1 intraplical) Polyp 3 (unilateral, whereof 1 hemorrhagic) Hyperplastic squamous epithelium (1)	Reinke’s edema 3 (2 unilateral, 1 bilateral) Cyst 1 (unilateral) Cyst with excised granuloma 1 Polyp 3 (unilateral, whereof 1 hemorrhagic) Sulcus 2 (unilateral, blunt dissection)
Occupation	chef (1), consultant (1), pharmacist (1), retail clerk (1) retired (2), singer (1), singing teacher (1), school teacher (2)	chef (1), consultant (1), deli clerk (1), office worker (2), retired (3), singer (1), school teacher (1)
Medical status	smoker (4), depressive (3), high blood pressure (2), reflux (2) hyperthyroidism (1), ulcerative colitis (1), chronic obstructive pulmonary disease (1), hearing impairment (0)	smoker (4), enlarged heart (1), high blood pressure (1), multiple sclerosis, depressive (1), hearing impairment (0)
Voice therapy	preoperative (0) postoperative (0)	preoperative (0) postoperative therapy (1 singer)
Preoperative vocal status (median)
Perceptual assessment (100 mm)	48	38
Visual assessment (48 p)	14	13

The low phonation time group phonated significantly less than the high phonation time group during voice advice week (U = 2.000, p < .001). The low phonation time group also phonated significantly less during voice advice week than at voice accumulation at long-term follow-up (Z = −2.666, p = .008). There was a significant improvement in the high phonation time week when self-assessments from voice advice week were compared to those from voice accumulation at long-term follow-up (Z = −2.366, p = .018). The same pattern was shown in results from SVQ (Z = −2.375, p = .018) and UVQ (Z = −2.213, p = .027).

Table 6 shows (1) self-assessment (VAS, SVQ, and UVQ), (2) perceptual assessment of overall voice quality and (3) visual inspection of digital imaging from laryngoscopy were compared from (1) before surgery, before VLT, (2) at 7-day check-up, and (3) 3–6 months following surgery, before second VLT. Within-groups comparisons were made with Friedman’s test for repeated measures at all three time points. The first and last time points were also compared within groups using Wilcoxon’s signed rank test. The comparisons showed significant decrease in self-assessment of general voice problems (100 mm VAS) within both groups. High phonation time: df 2, χ² = 10.174, p = .006, Z = −2.201, p = .028, low phonation time: (χ² = 2.545, p = .280). Z = −1.991, p = .046 The same pattern was shown when comparing SVQ, high phonation time: df 2, χ² = 15.154, p < .001, Z = −2.542, p = .012, low phonation time: (χ² = 6.588, p = .037) Z = −1.992, p = .046 and for UVQ; high phonation time: df 2, χ² = 8.968, p = .011, Z = −2.533, p = .011, low phonation time: (χ² = 4.667, p = .097) Z = −2.384, p = .017.

Table 6. Results from self-assessments of overall voice problems (100 mm VAS), 10 questions on shifting voice problems (SVQ), 8 questions on underlying voice problems (UVQ), overall points from visual assessments (max 48), perceptual assessments of overall affected voice quality (100 mm VAS) and participation time in a vocal loading task in two groups: low phonation time and high phonation time before surgery, at short-term and at long-term check-up. NS: No significant within-group differences.

		Low phonation time				High phonation time
		Before surgery	Short-term check-up	Long-term check-up	Within group^a	Before surgery (i)	Short-term check-up (ii)	Long-term check-up (iii)	Within group^a	Between measure-ments, post-hoc (i), (ii) and (iii)^b
Self-assessment (100 mm VAS)	Mean	54.8	53	21.88	χ^2 = 2.545, p = .280^aNS	65.5	57.5	16.67	χ^2 = 10.174, p = .006^a*	i to ii: Z = −2.380, p = .017^b** i to iii: Z = −2.201, p = .028^bNSii to iii: Z = −2.366 p = .018^b**
	Median	52	53	14.5		73	61	14.5
	SD	28.5	5.7	20.5		23.5	22.4	14.5	–	–
	95% CI	[32.87, 76.68]	[2.18, 103.82]	[4.71, 39.04]		[48.68, 82.32]	[38.77, 76.23]	[1.42, 31.91]	–	–
Self-assessment SVQ (40 p)	Mean	20	18	13	χ^2 = 6.588, p = .037^aNS	26.9	12.3	9.8	χ^2 = 15.1548, p < .001^a*	i to ii: Z = −2.668, p = .008^b** i to iii: Z = −2.542, p = .012^b** ii to iii: Z = −2.375 p = .018^b**
	Median	19	18	12		27.5	11	8.5	–	–
	SD	5.6	1.4	5.7		4.6	8.6	7.2	–	–
	95% CI	[15.69, 24.31]	[5.29, 30.71]	[8.32, 17.93]		[23.59, 30.21]	[5.03, 19.47]	[2.28, 17.38]	–	–
Self-assessment UVQ (32 p)	Mean	12.9	23.5	3.75	χ^2 = 4.667, p = .097^aNS	17.6	17	6.67	χ^2 = 8.968, p = .011^a*	i to ii: Z = −1.606, p = .108^bNSi to iii: Z = −2.533, p = .011^b** ii to iii: Z = −2.213, p = .027^bNS
	Median	11	23.5	3		16.5	18	6
	SD	5.2	3.5	3		6.9	8.3	6.3
	95% CI	[8.89, 16.89]	[−8.27, 55.27]	[1.23, 6.27]		[12.64, 22.56]	[10.09, 23.91]	[−.03, 13.36]
Perceptual assessment (overall 100 mm VAS)	Mean	43.4	39.2	17	χ^2 = 4.222, p = .121^aNS	42	48.7	18.9	χ^2 = 2.000, p = .368^aNS	–
	Median	48	32	5		38	67	5
	SD	37.3	33.9	22.6		35.1	40.8	23.4
	95% CI	[42.94, 72.11]	[13.13, 65.31]	[−.41, 34.41]		[15.00, 69.00]	[17.29, 80.04]	[.92, 36.86]
Visual assessments (overall 48 p)	Mean	14	9.2	6.8	χ^2 = 7.824, p = .020^aNS	12	10.3	7	χ^2 = 2.774, p = .250^aNS	–
	Median	16	8	6		13	8	6
	SD	6.1	6.4	4.4		4.5	5.7	3.6
	95% CI	[9.63, 18.37]	[4.63, 13.77]	[3.68, 9.92]		[8.21, 15.79]	[5.46, 15.04]	[3.97, 10.04]
Time in VLT (min)	Mean	15	–	19	Z = −.679, p = .497^bNS	19	–	25	Z = −2.207, p = .027^bNS	–
	Median	12	–	17		17	–	30
	SD	10	–	10		9.0	–	8
	95% CI	[7.66, 22.74]	–	[11.27, 26.13]		[12.12, 26.63]	–	[19.67, 30.83]

*Significance at alpha level p ≤ .017, using Bonferroni correction for a priori test using ^aFriedman’s test of repeated measures. **Significance at alpha level p ≤ .025, using Bonferroni correction for post-hoc test ^bWilcoxon signed rank test. The latter was not applicable for Low phonation time.

There were no statistically significant results found when perceptual assessments were compared over time although the assessments of severity or grade of the overall voice pathology, decreased considerably for both groups over time. Low phonation time: median 32 to 5, χ² = 4.222, p = .121. High phonation time: median 67 to 5, χ² = 2.000, p = .368. Visual assessment results show substantial decrease in morphological deviance before surgery to short time check-up. This decrease was stable for both groups in the long term check-up. The decrease was statistically insignificant in both groups, see Table 6.

We also wanted to assess the new groups’ (1) ability to identify heavy vocal loading, and (2) short-term recovery from heavy vocal loading. VAS scores were standardized and calculated ad modum Atkinson et al. [20], as applied by Vogel and Maruff [21] and Whitling et al. [22], by neutralizing baseline to 0 and dividing the difference between true baseline and subsequent scores with the within-group standard deviation at each time point. Results show that the group with high phonation time reacted more distinctly to heavy vocal loading in the second VLT. They also reacted more than the low phonation time group in both VLTs. In accordance with Whitling et al. [22], this strong reaction is a sign of adequate vocal health, as patients with muscle tension voice disorders tend to react little or not at all to a heavy VLT. On the other hand, the high vocal loading group did not return to baseline recordings of voice problems within 2 h following heavy vocal loading, reflecting the fact that they were able to spend significantly longer (double time) in their second VLT than their first (median 30 versus 15 min: Z = −2.384, p = .017). This significant difference was not found in the group who had low phonation time during voice advice week, who however also increased participation time (median 12 versus 20 min). The other three recovery courses were similar to each other, as is seen in Figure 3.

Figure 3. Short-term recovery with standardized self-assessment scores from 100 mm visual analogue scale (VAS) in the low phonation time group (LPT) group and the high phonation time group (HPT), following two vocal loading tasks: one performed before surgery, the other at long-term check-up.

Discussion

This study investigated differences in compliancy with absolute versus RVR advice during 7 days following phonomicrosurgery. Participants compliancy to voice rest advice varied in both groups. The AVR groups were not absolutely silent, but they phonated significantly less (3%) than the RVR group (7%) during voice rest week. Phonation time significantly increased for the AVR group when the voice rest week was compared to iterated voice accumulation at the long-term check-up (11%), while no such difference was noted in the RVR group (9%). Phonation times during voice rest week indicate return to typical vocal behavior in both groups. They can be compared to phonation times of a control voice healthy control group in a study by Whitling et al. [19], who’s phonation time during one week’s voice accumulation was 8% during work and 9% during leisure. Given that participants in the current study were aware of being monitored they probably would have phonated even more during voice rest week without a dosimeter present. In line with Rousseau et al. [2], participants in the AVR group had trouble complying with their voice care advice. Instead of letting participants estimate how much they had phonated, compliance was successfully monitored through voice dosimetry, as suggested by Misono et al. [16].

This study also examined vocal recovery from surgery through assessments from participants, speech, and language pathologists (perceptual assessments of voice quality) and phoniatricians (visual assessments of digital imaging), both in the short-term (7 days after surgery) and in the long term (3–6 months after surgery). These assessments showed only one significant difference in the favor of AVR: the AVR group showed significant improvement in phoniatricians’ visual assessments at short-term check-up. Their mean score decrease from 14 to 8 points. Such an improvement was not evident in the RVR group, whose mean score decreased from 11 to 10 points. However, the maximum score in the protocol used for visual assessments [20,21] is 48 points, which means neither group showed any great morphological deviations to begin with; both groups had the same score of seven at long-term check-up and there were no significant differences in group comparisons, nor were there any for individually assessed parameters within the protocol. This group difference may also indicate short-term, or acute, recovery of mucosal tissue, is more effective when the tissue is not put in vibration during the first 24 h following phonosurgery. However, recovery of vocal function is more complex than inflammatory reactions within the vocal folds, and must be examined as such.

Self-assessments at long-term check-up showed significant improvement of all voice problems (VAS, SVQ, and UVQ) for the RVR group. There was evidence for improvement also for the AVR group, who showed significant improvement for UVQ (specific voice problems which are stable over time). Given the dignity of self-assessment within voice pathology [27], and the importance of long-term improvement, these results speak in favor of RVR.

In the second VLT, reactions to vocal loading, i.e. self-assessed changes in general voice problems on a VAS, from participants in the RVR group are equivalent to those of women who have high levels of vocal loading in their everyday life and who do not suffer from any voice problems, i.e. voice healthy individuals, who show resilience to vocal impairment [22]. Results from VLTs showed significantly improved vocal stamina for the RVR group, who also reacted in a more voice healthy way to vocal loading compared to the AVR group. Functional recovery was made evident by a doubling in median participation time and also in adequate self-assessments of vocal loading.

Results in the current study indicate that AVR is not necessary for successful recovery of vocal function, and that RVR is easier for patients to comply with. AVR may be fruitful for short-term histological recovery, but the long-term benefits are not shown in this study. These findings may however suggest short-term sick leave from work that strains the voice. Results are based on a novel procedure for clinical evaluation of vocal recovery following phonomicrosurgery, combining (1) comparison of absolute and RVR following vocal surgery with (2) voice accumulation to estimate compliance through relative phonation time and (3) short-term and long-term check-up of recovery, and (4) re-iterated VLTs to measure vocal stamina and reaction to vocal loading.

The strength of randomization in this study is limited by its small sample size. Randomized distribution of diagnoses in the small groups may have made them problematic to compare. A larger sample would enable detection of smaller group differences. Enrolment in the current study involved two additional, time consuming clinical visits, without any extra compensation. About 77% of eligible patients who did not participate in this study stated the trouble of extra visits in the voice clinic as the primary reason. We have striven to focus on thus far lacking, although time-consuming, data collection of clinical factors, which enabled us to evaluate effects of voice rest and give us a holistic view of participants’ vocal recovery. More research comparing randomized absolute and RVR with long-term recovery following phonomicrosurgery is warranted.

Post hoc group compilations, dividing participants into groups of high or low phonation time, showed no other significant differences than the original voice care advice groups. This indicates overall good compliance to voice care advice and that phonation time in general was low following phonosurgery. The high phonation time groups were better adapted at the long-term check up for a VLT. They stayed in it for longer, showing increased stamina. These post hoc analyses may reinforce RVR following phonosurgery.

Conclusions

This study presents evidence to support RVR over AVR following surgery of benign lesions in the vocal fold mucosa on the following rationale:

1. Patients showed difficulty in complying with AVR to a higher degree than RVR, both regarding self-assessed compliance and objective measurements of relative phonation time. Patients phonated to low extent when recommended AVR, but they were not completely silent.

2. The AVR group reached significant morhpological recovery in the short-term. This effect may speak in favour of short-term AVR from a histological point of view, however this did not hold true for other parameters, such as perceptual or self-assessment evaluations. Combined with the strain reported in this study, of having to comply with AVR, weighs heavier on favour of RVR for positive long-term outcome.

3. Patients recommended for RVR showed significantly better vocal stamina and immediate recovery from vocal loading, as well as significantly improved within-group self-assessment of voice problems at long-term check-up.

Acknowledgements

The authors express their gratitude to all participants’ diligent participation in this study and clinical colleagues who have been helpful with recruitment of subjects and perceptual and visual assessments of data. Special thanks to Lena Asker-Árnason for impartial assistance with randomization, to Anders Löfqvist for valuable feedback on the manuscript and to David Whitling for proofreading. We also want to thank two independent reviewers, who’s remarks have greatly improved the quality of this study.

Disclosure statement

The second author is an associate editor of Logopedics, Phoniatrics and Vocology and as such has taken no part in the review or editing process of the current paper, but acted as co-author only. There are no affiliations with or involvement in any organization or entity with any financial interest in the subject matter or materials discussed in this manuscript.

Additional information

Funding

This work was supported by AFA Försäkring, a work-related insurance company in Sweden, under grant number 110230.

Notes on contributors

Susanna Whitling

Susanna Whitling, SLP, PhD is a registered Speech and Language Pathologist and a post doc researcher and lecturer of Speech and Language Pathology at the Department of Logopedics, Phoniatrics and Vocology, Faculty of Medicine, Lund University, Lund, Sweden.

Viveka Lyberg-Åhlander

Viveka Lyberg-Åhlander, SLP, PhD is a registered Speech and Language Pathologist and an associate professor (docent) of Speech and Language Pathology at the Department of Logopedics, Phoniatrics and Vocology, Faculty of Medicine, Lund University, Lund, Sweden.

Roland Rydell

Roland Rydell, MD, PhD is a chief physician, phoniatrician and laryngologist at the Division of Ear, Nose and Throat Diseases, Head and Neck Surgery Faculty of Medicine, Lund University and an associate professor (docent) of Phoniatrics at the Department of Logopedics, Phoniatrics and Vocology, Faculty of Medicine, Lund University, Lund, Sweden.