Comparison of Fundamental Frequency in Postmenopausal Women Who Are Treated With Hormone Replacement Therapy vs Those Who Are Not

Key Points

Question

Is hormone replacement therapy (HRT) for treatment of menopausal symptoms associated with changes in speaking fundamental frequency?

Findings

This systematic review and meta-analysis found that the speaking fundamental frequency was significantly higher in women who received HRT than in women who did not. The difference appeared to be more pronounced in women with a normal body mass index, but not in women with a high body mass index.

Meaning

Hormone replacement therapy may be associated with a higher speaking fundamental frequency in postmenopausal women with a normal body mass index.

Abstract

Importance

Hormonal changes during menopause have been associated with significant changes in voice. Although hormone replacement therapy (HRT) is used primarily to manage systemic symptoms of menopause, its association with voice in postmenopausal women has not been adequately investigated by large-scale studies.

Objective

To compare fundamental frequency between postmenopausal women who used HRT and those who did not use HRT.

Data Sources

PubMed, Ovid MEDLINE, CINAHL, Cochrane EBM Reviews, and Embase were searched from 1946 to February 19, 2020.

Study Selection

Studies included in the final review were those in English that compared voice outcomes in postmenopausal women who were or were not receiving HRT for treatment of climacteric symptoms associated with menopause.

Data Extraction and Synthesis

The study was conducted in accordance with Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline. Data extraction was performed by 2 independent investigators. Study quality was assessed using a validated quality tool. Whenever possible, data were pooled using random-effects meta-analysis.

Main Outcomes and Measures

The primary outcome was the difference in speaking fundamental frequency (F₀) between HRT users and nonusers. Subgroup analysis of the primary outcome was based on body mass index (BMI) using a cutoff value of 25. A BMI of 18.5 to 24.9 is considered normal, and a BMI of at least 25 is considered overweight. Secondary outcomes included other objective measurements of voice, including jitter and shimmer.

Results

A total of 937 records were screened, 18 full texts were assessed, and 11 studies were included in the final review. All 11 studies were case-control studies and were rated from fair to good quality based on the Newcastle-Ottawa Scale. A total of 5 studies including 154 participants who received HRT and 154 controls were used for meta-analysis. The mean (SD) F₀ was 185.9 (8.6) Hz for HRT users compared with 174.6 (6.6) Hz for nonusers. Hormone replacement therapy was associated with a higher mean F₀ by a difference of 11.85 Hz (95% CI, 7.35-16.36 Hz). Subgroup analysis showed that the change in F₀ was significant in postmenopausal women with a normal body mass index (mean difference, 13.47 Hz; 95% CI, 8.48-18.46 Hz) but not in those with a high body mass index.

Conclusions and Relevance

Existing evidence suggests that HRT is associated with a higher F₀ in postmenopausal women. The effectiveness of the treatment appeared to be more pronounced in women with a normal body mass index.

This systematic review and meta-analysis compares the voice frequency of postmenopausal women with normal body mass index (BMI) who used hormone replacement therapy (HRT) and those who did not use HRT.

Introduction

In postmenopause, a period of a woman’s life that follows the final menstruation, cessation of estradiol production occurs, and long-term manifestations of estrogen deprivation ensue.^1,2 There is a general perception that hormonal changes in menopause are associated with significant changes in voice. In a national study of more than 60 000 people older than 65 years, women were found to have increased odds of developing nonspecific hoarseness compared with men.³ Voice conditions have been found to significantly interfere with quality of life for women older than 60 years.⁴ Given that voice symptoms can have a substantial impact on work performance, family life, and social situations, the postmenopausal effect on voice cannot be overlooked.

Vocal folds and the voice production mechanism are known to be influenced by hormones.⁵ Hormonal receptors have been found in laryngeal mucosa and vocal folds.^6,7 Cytologic changes in the vocal folds in menopause have been shown to mirror changes in the cervix.⁵ Therefore, cessation of ovarian production of estradiol in menopause results in multiple changes to the vocal folds and subsequently the voice. Estrogen is essential for maintaining muscle function and its deficiency leads to muscle atrophy and reduced muscle strength.⁸ This leads to reduced vocal fold muscle bulk and incomplete glottal closure on phonation. Clinically, patients will report reduced vocal volume, diminished vocal range, and worsening vocal fatigue. Furthermore, estrogen deficiency results in a smaller number of glandular cells in the subepithelial layer of the vocal folds.⁵ Therefore, vocal folds will be dryer and less lubricated, leading to disruption of their biomechanical vibratory properties and hoarseness.⁹

Hormone replacement therapy (HRT) includes either estrogen therapy or a combination of estrogen and progesterone therapy.¹⁰ It remains the most effective treatment of symptoms relating to menopause, such as hot flashes and sweating, as well as genitourinary syndrome of menopause.¹⁰ In the long term, HRT has been shown to prevent bone loss, reduce coronary heart disease, and even extend life expectancy for postmenopausal women.^10,11 Although HRT is primarily used to manage systemic symptoms of menopause, its unintended effect on voice in postmenopausal women has not been investigated through large-scale studies, to our knowledge. Given the recent emergence of evidence on the benefits and safety of HRT, this treatment may become more commonly prescribed for postmenopausal women. The objective of the present study is to evaluate the association of HRT with voice in postmenopausal women through a systematic review and meta-analysis.

Methods

This protocol-based systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline.¹² Approval by a research ethics board was not required for this type of study at our local institution.

Information Sources

A literature search was performed in PubMed, Ovid MEDLINE, CINAHL, Cochrane EBM Reviews, and Embase from 1946 to February 19, 2020. A health sciences librarian developed, piloted, and executed the searches. The search strategy is included in the eAppendix in the Supplement.

Study Inclusion Criteria

Published clinical studies of humans, including randomized clinical trials, cohort studies, case-control studies, and case series, comparing voice outcomes among postmenopausal women who were or were not receiving either estrogen therapy or estrogen-progesterone therapy as HRT for treatment of climacteric symptoms relating to menopause were included in the final review.

Study Exclusion Criteria

Non-English articles, case reports, conference abstracts, letters and correspondence, expert opinions, and review articles were excluded. Grey literature, such as government reports, policy statements, and conference proceedings, were not considered.

Study Selection and Data Extraction

Two investigators (including T.W.) independently evaluated the titles and abstracts for inclusion eligibility. Disagreements were resolved by discussion with the senior investigator (R.J.L.). Data were independently extracted by 2 investigators (including R.J.L.) from each included study using structured, customized forms. Every effort was made to contact the original study author when data clarification or additional data were required. The following data were extracted: study design, treatment and control group sizes, patient characteristics, HRT dosing regimen, route and duration of administration, definition of outcome measures, and adverse events.

Definition of Outcome Measures

The primary outcome measure of the review was speaking fundamental frequency (F₀), defined as the frequency of vibration of vocal folds during connected speech or during sustained vowel phonation at a comfortable pitch and loudness. Secondary outcome measures included other objective measurements of voice, including jitter and shimmer.

Risk-of-Bias Assessment

Quality assessment of randomized clinical trials was rated using the Cochrane GRADE (Grading of Recommendations, Assessment, Development and Evaluations) tool,¹³ whereas quality assessment of cohort studies and case-control studies was rated using the Newcastle-Ottawa Scale.¹⁴ Publication bias was assessed using a funnel plot.

Data Synthesis

A narrative synthesis was conducted for all included studies. Studies with comparable baseline characteristics between intervention and control groups, outcome measures, and risk of bias were selected for meta-analysis. To account for heterogeneity across studies, a random-effects model was constructed using RevMan, version 5.0 (Cochrane Collaboration). All significant tests were 2-tailed, with P < .05 considered statistically significant. Between-study heterogeneity was quantified by the I² statistic and interpreted qualitatively as low (25%), moderate (50%), and high (75%).¹⁵ Mean differences were used as the treatment effect for the aforementioned voice outcomes. Subgroup analysis of the primary outcome was conducted based on the World Health Organization classification of body mass index (BMI; calculated as weight in kilograms divided by height in meters squared) using a cutoff value of 25. A BMI of 18.5 to 24.9 is considered normal, and a BMI of at least 25 is considered overweight.

Results

Study Selection and Characteristics

Searches identified 1643 records. After the removal of 706 duplicates, 937 unique records were identified, of which 18 were potentially relevant based on initial title and abstract screening (Figure 1). A total of 11 articles were included in the final systematic review,^{16,17,18,19,20,21,22,23,24,25,26} and 5 were eligible to be included in a meta-analysis.^{17,19,20,22,26} Study characteristics are shown in Table 1.

Figure 1. PRISMA Flow Diagram.

Table 1. Summary of Included Studies^a.

Source	Definition of menopause	Age, mean (SD), yb	No. of smokers	BMI, mean (SD)	No. of patients		HRT	HRT treatment regimen	Primary outcome measure	Adverse events	Summary of study findings
					HRT group	No-HRT group
Caruso et al,16 2000	Surgical menopause by ovariectomy	HRT: 53.1; no HRT: 51.9	0	23.6 (1.2)	48	36	Transdermal ET	Patch change once or twice a week; gel applied once daily	Comparison of cytologic findings from laryngeal smears and Papanicolaou tests	Zero	HRT prevents dystrophic changes of vocal folds, which correlate with Papanicolaou test findings
D’haeseleer et al,17 2011	12 Consecutive months of amenorrhea or surgically induced menopause via ovariectomy	HRT: 54.8; no HRT: 55.5	0	HRT: 23.7; no HRT: 24.1	26	38	Not provided	NA	Mean (range) F0: HRT, 187.7 Hz (148.9-236.7 Hz); no HRT, 178.9 Hz (153.9-202.2 Hz)	NA	In postmenopausal women without HRT, increasing BMI is associated with increasing F0; the same correlation is not seen in those receiving HRT
D’haeseleer et al,18 2012	12 Consecutive mo of amenorrhea or surgically induced menopause via ovariectomy	HRT: 57.2 (4.8); no HRT: 58.6 (5.4)	Not provided	Not provided	63	42	ET or EPT	NA	Nasal resonance: mean (SD) F0: HRT: 184 (16.8) Hz; no HRT: 178 (19.9) Hz	NA	No difference in nasal resonance between groups
D’haeseleer et al,19 2012	12 Consecutive mo of amenorrhea or surgically induced menopause via ovariectomy	HRT: 57.6 (4.5); no HRT: 58.5 (5.1)	0	HRT: 24.6 (3.4); no HRT: 24.4 (3.3)	59	46	Oral or transdermal ET; oral EPT	ET: oral or transdermal, mean (SD), 1.44 (0.44) mg; EPT: oral, mean (SD), 1.22 (0.54) mg for estradiol and 92.0 mg or 3.9 mg for natural or synthetic progestogen	Mean F0: HRT: 190.6 (18.8) Hz; no HRT: 176.4 (21.2) Hz	NA	Postmenopausal women without HRT had a significantly lower F0 compared with those receiving HRT; there was no difference between types of HRT
D’haeseleer et al,20 2013	12 Consecutive mo of amenorrhea or surgically induced menopause via ovariectomy	HRT and low BMI: 57.5 (5.0); HRT and high BMI: 56.7 (4.1); no HRT and low BMI: 58.5 (5.5); no HRT and high BMI: 59.4 (5.4)	0	HRT and low BMI: 22.4 (1.9); HRT and high BMI: 27.3 (1.8); no HRT and low BMI: 22.5 (1.7); no HRT and high BMI: 28.0 (2.2)	HRT and low BMI: 35; HRT and high BMI: 19	HRT and low BMI: 28; HRT and high BMI: 12	ET or EPT	NA	Mean F0: HRT and low BMI: 188.7 (18.2) Hz; HRT and high BMI: 187.8 (22.6) Hz; no HRT and low BMI: 174.5 (17.8) Hz; no HRT and high BMI: 183.0 (11.3) Hz	NA	Postmenopausal women not receiving HRT and with a low BMI have significantly lower F0
Firat et al,21 2009	Underwent hysterectomy and oophorectomy	HRT: 47.8 (4.9); no HRT: 45.4 (7.9)	0	NA	23	9	Intranasal or oral ET	Intranasal (300 μg 17-β-estradiol twice a day) or oral (2 mg estradiol once daily)	Difference in Turkish VHI score at baseline and 12 mo after HRT	NA	HRT had significantly improved VHI score at 12 mo; more pronounced improvement with intranasal estrogen
Hamdan et al,22 2018	Not provided	HRT: 53.0 (6.8); no HRT: 53.5 (5.6)	HRT: 4/19; no HRT: 13/34	HRT: 25.9 (5.4); no HRT: 25.8 (4.3)	19	34	ET, PT, or EPT	NA	VHI-10 score and habitual pitch (Hz)	NA	No difference in VHI-10 score between groups; significantly lower habitual pitch in no-HRT group compared with HRT group
Lindholm et al,23 1997	Surgically induced menopause or amenorrhea for 6-24 mo	51.6	Not provided	Not provided	29	13	Oral ET or EPT for ≥12 mo	ET (2 mg of estradiol orally per day for 1 y); EPT (2 mg of 17-β-estradiol and 1 mg of norethisterone acetate orally daily for 1 y)	Mean F0	NA	F0 significantly lower in no-HRT group during spontaneous speech and reading
Mendes-Laureano et al,24 2006	Menopause of ≥2 y	HRT: 54.5; no HRT: 56.5	0	<30	45	15	Oral EPT for ≥6 mo	1 mg of estradiol valerate and 90 μg norgestimate orally for ≥6 mo	Mean F0	NA	No difference in F0 between groups
Mendes Laureano et al,25 2007c	Menopause of ≥2 y	HRT: 54.5; no HRT: 56.5	0	<30	45	15	Oral EPT for ≥6 mo	1 mg of estradiol valerate and 90 μg norgestimate orally for ≥6 mo	HNR	NA	No difference in HNR between groups
Mendes Laureano et al,26 2009	Natural menopause of ≥2 y	HRT: 54.5; no HRT: 56.5	0	Not provided	15	15	Oral EPT for ≥6 mo	1 mg of estradiol valerate and 90 μg norgestimate orally for ≥6 mo	Jitter and shimmer	NA	No difference in jitter and shimmer between groups

Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); EPT, estrogen-progesterone therapy; ET, estrogen therapy; F₀, fundamental frequency; HNR, harmonic to noise ratio; HRT, hormone replacement therapy; NA, not available; PT, progesterone therapy; VHI, Voice Handicap Index; VHI-10, Voice Handicap Index-10.

^a All included studies were case-control studies.

^b Standard deviation of age was included if reported in the original study.

^c This study was a letter responding to reviewer comments to the authors’ study in 2006.²⁴ The HNRs were analyzed from voice recording samples from the same patients enrolled in the 2006 study.

Meta-analysis

Association of HRT With F₀

The association of HRT with F₀ was evaluated by 4 studies, all of which were of case-control design (Table 1).^17,19,20,22 The F₀ in each woman in these studies was measured during sentence reading at habitual pitch and loudness in a computerized speech laboratory using well-known automated voice analysis software, such as the Real Time Pitch program (KayPENTAX) and the Dr Speech Science program (Tiger DRS Inc). Data from these studies were combined for meta-analysis, resulting in 154 postmenopausal women who received HRT and 154 postmenopausal women who did not receive HRT. D’haeseleer et al¹⁷ reported mean F₀ values with a range, but SDs were required for meta-analysis. Therefore, the formula of (maximum − minimum)/4 was used to convert the range values to approximate the values of SDs.²⁷ Overall, data synthesis showed that the mean (SD) F₀ was 185.9 (8.6) Hz in HRT users compared with 174.6 (6.6) Hz in nonusers. The mean F₀ was 11.85 Hz higher in those who were receiving HRT (95% CI, 7.35-16.36 Hz; heterogeneity, I² = 0%) (Figure 2).^17,19,20,22 The I² statistic was 0%, which indicates a low level of heterogeneity among the studies. Three of these studies scored at least 8 of 9 asterisks on the Newcastle-Ottawa Scale, indicating good quality (Table 2).^17,19,20 One study scored 7 of 9 asterisks, indicating fair quality.²²

Figure 2. Forest Plot of Comparison of Mean Fundamental Frequency (F₀) in Postmenopausal Women Who Were or Were Not Receiving Hormone Replacement Therapy (HRT) With Normal Body Mass Index (BMI) or High BMI.

Women receiving HRT had a significantly higher mean F₀ than those who were not receiving HRT. The difference was significant in the group with a normal BMI but not in the group with a high BMI. A BMI of 18.5 to 24.9 is considered normal, and a BMI of at least 25 is considered overweight. Random-effects models were used for all analyses with inverse-variance weighting methods.

Table 2. Risk-of-Bias Assessment in Included Human Studies.

Source	Newcastle-Ottawa Scale for case-control studiesa
	Selection				Comparability	Exposure
	Is case definition adequate?	Representativeness of cases	Selection of controls	Definition of controls	Comparability of cases and controls on basis of design or analysis	Ascertainment of exposure	Same method of ascertainment for cases and controls	Nonresponse rate
Caruso et al,16 2000	*	*	*	*	**	*	*	*
D’haeseleer et al,17 2011	*	*	*	*	**	*	*	*
D’haeseleer et al,18 2012	*	*	*	*	*	*	*	*
D’haeseleer et al,19 2012	*	*	*	*	*	*	*	*
D’haeseleer et al,20 2013	*	*	*	*	*	*	*	*
Firat et al,21 2009	*	*	*	*	*	*	*	*
Hamdan et al,22 2018	No scoreb	*	*	*	*	*	*	*
Lindholm et al,23 1997	*	*	*	*	No scoreb	*	*	No scoreb
Mendes-Laureano et al,24 2006	No scoreb	*	*	*	*	*	*	No scoreb
Mendes Laureano et al,25 2007	No scoreb	*	*	*	*	*	*	No scoreb
Mendes Laureano et al,26 2009	No scoreb	*	*	*	*	*	*	No scoreb

^a This scale has a scoring system using asterisks based on 3 domains, including selection of study groups, comparability of groups, and ascertainment of exposure. A maximum of 4 asterisks could be given to the selection domain, 2 asterisks to the comparability domain, and 3 asterisks to the exposure domain. A greater number of asterisks indicates greater quality.

^b These studies did not score an asterisk in this category.

Subgroup analysis was performed based on the World Health Organization classification of BMI, using 25 as a cutoff. In the subgroup with normal BMI (18.5-24.9), the mean (SD) F₀ was 189.7 (1.9) Hz in HRT users and 175.2 (3.3) Hz in nonusers. The mean F₀ was 13.47 Hz higher in HRT users with a normal BMI (95% CI, 8.48-18.46 Hz; heterogeneity, I² = 0%) (Figure 2). This difference was not observed in the subgroup of HRT users with a high BMI (mean difference [MD], 4.75 Hz; 95% CI, −5.72 to 15.22 Hz; heterogeneity, I² = 0%).

Association of HRT With Jitter and Shimmer

The association of HRT with jitter and shimmer in postmenopausal women was assessed in 3 case-control studies (Table 1).^19,22,26 The combined data included 93 postmenopausal women receiving HRT and 95 postmenopausal women who were not receiving HRT. There was no difference in jitter or shimmer between the 2 groups (jitter: MD, 0.12; 95% CI, −0.03 to 0.28; heterogeneity, I² = 0%; shimmer: MD, −0.32; 95% CI, −1.07 to 0.43; heterogeneity, I² = 0%). The I² statistic was 0% for each parameter, which indicates a low level of heterogeneity among the studies. One study scored 8 of 9 asterisks on the Ottawa-Newcastle Scale, indicating good quality (Table 2).¹⁹ The remaining 2 studies scored 6 of 9 asterisks, indicating fair quality.^22,26

Qualitative Review of Studies Not Included in Meta-analysis

Six prospective case-control studies were not included in the data synthesis (Table 1).^{16,18,21,23,24,25} All participants in these 6 studies were women with either natural or surgically induced menopause. They were receiving HRT for climacteric symptoms associated with menopause. Caruso et al¹⁶ found a significantly higher proportion of nonusers reporting voice symptoms compared with HRT users using a general questionnaire. Laryngeal smears from the nonuser group showed cellular atrophy-dystrophy, which was comparable to findings from Papanicolaou tests. This type of cellular change was not seen in laryngeal smears or Papanicolaou test results in the HRT group. D’haeseleer et al¹⁸ examined nasal resonance and speaking F₀ between HRT users and nonusers, and no change in nasal resonance was found. Because BMI was not reported by this study, it was not included in the meta-analysis. Firat et al²¹ found that the HRT group had significantly improved Voice Handicap Index scores at 12 months after therapy initiation compared with baseline. This finding was in contrast to a marked deterioration of Voice Handicap Index scores at 12 months among patients not receiving HRT. Between-group comparisons were not made, so it was unclear how the Voice Handicap Index scores differed between the 2 groups. Lindholm et al²³ evaluated F₀ in HRT users and nonusers and found that F₀ was significantly lower in nonusers. However, this study did not account for BMI in its data analysis, and the comparability of the groups was not clear. Mendes-Laureano et al²⁴ found that there was no difference in F₀ between HRT users and nonusers in a 2006 study. This study did not have a clear definition of menopause and therefore was not included in the meta-analysis. The authors reanalyzed the data of the same patient cohort to include the harmonic to noise ratio in a 2007 study,²⁵ which found no between-group differences.

Adverse Events Associated With HRT

The study by Caruso et al¹⁶ reported 0 adverse events (Table 1). None of the other included studies provided rates of adverse events associated with HRT.

Risk of Bias Within Studies

A summary of risk of bias is presented in Table 2. All of these studies were assessed using the Newcastle-Ottawa Scale for case-control studies. The study by Hamdan et al²² did not exclude smokers from study participation. Although BMI was accounted for in their data analysis, smoking was not. As a result, the study only scored 1 asterisk in comparability and was subsequently categorized as fair quality. The study by Lindholm et al²³ was categorized as poor quality because it did not control for BMI in the data analysis. As a result, comparability between groups could not be confidently established, resulting in a 0 asterisk score. The 3 articles by Mendes Laureano et al^24,25,26 did not define menopause clearly, and they did not report on missing data. As a result, those articles were categorized as fair quality.

Publication Bias

A funnel plot for publication bias was generated using the studies included in the meta-analysis for F₀ (Figure 3).^17,19,20,22 The outer diagonal dashed lines in Figure 3 indicate the triangular region within which 95% of studies were expected to lie in the absence of both biases and heterogeneity.²⁸ The vertical dashed line indicates the mean effect size. The plot shows a clustering of studies near the top of the funnel and no studies at the bottom of the funnel, which suggests publication bias. However, these results were interpreted with caution because there were only a limited number of studies included in the analysis.

Figure 3. Funnel Plot Showing Publication Bias.

The outer dashed lines indicate the triangular region within which 95% of studies were expected to lie in the absence of both biases and heterogeneity. The vertical dashed line indicates the mean effect size. Zero on the x-axis in this case indicates no intervention effect. A BMI of 18.5 to 24.9 is considered normal, and a BMI of at least 25 is considered overweight. BMI indicates body mass index.

Discussion

The present systematic review and meta-analysis found that postmenopausal HRT users had a significantly higher F₀ compared with nonusers. This association was more pronounced in women with a normal BMI. As women age, their F₀ tends to decrease; this study suggests that HRT may help reverse this phenomenon.²⁹

Adipose tissue can be a significant extragonadal source of estrogen by converting androgen to estrones, which is a weak form of estrogen.³⁰ In postmenopausal women, a major site of estrone synthesis is the lipocytes.³¹ This is in keeping with our study finding that there was no significant change in F₀ between HRT users and nonusers with a high BMI. In these women, voice changes associated with menopause may not be as noticeable owing to a high level of endogenous estrogen secondary to higher volumes of adipose tissue. As a result, their voice is less dependent on HRT compared with those with a normal BMI.

We found no difference between HRT users and nonusers in acoustic variables such as jitter and shimmer, which are traditional time-based measurements of voice. In recent years, frequency-based acoustic evaluations of voice have emerged and have been found to have a greater sensitivity in detecting voice changes and voice disorders. These include cepstral peak prominence and the cepstral spectral index of dysphonia measured during sentence readings.^32,33 These measurements have been shown to be better than the time-based parameters at differentiating between dysphonia and normal voice and at defining the severity of dysphonia.³⁴ Future studies can focus on cepstral evaluations of voice to further provide objective evidence to support the positive effect of HRT for postmenopausal women with voice issues.

Only 2 included studies used patient-reported outcome measures, such as the Voice Handicap Index, as an outcome.^22,23 As a result, questions still remain regarding the complex relationships between the perception of a feminine voice as determined by a higher F₀ and a woman’s perception of her own voice and its effect on her quality of life. In addition to objective measurements of voice, future studies should also include subjective evaluations of voice using patient-reported outcome measures to determine the association between HRT and self-perceived voice quality.

Adverse Events

The benefits and risks associated with HRT are well documented in the menopause literature. In 2002, results of a large randomized clinical trial conducted by the National Institutes of Health, known as the Women’s Health Initiative, were published.³⁵ The study concluded that taking HRT increased women’s risks of heart attacks, strokes, venous thrombosis, and breast cancer. These conclusions had resulted in a dramatic reduction in the prescription of HRT in the United States in the early 2000s.^36,37 However, subsequent studies have shown that the benefits of HRT generally outweigh the risks among appropriate patients. A comprehensive overview of the findings from the Women’s Health Initiative trials with a 13-year cumulative follow-up found no significantly increased risk of heart disease in women receiving HRT who were younger than 60 years or within 10 years of menopause onset.³⁸ Although the relative risk of stroke increased with HRT use, it was a rare event in this age group, with an absolute attributable risk of fewer than 0.5 additional cases per 1000 women per year. Venous thrombotic events increased with oral estrogen use, but large observational studies did not identify an increased risk with transdermal estradiol use.³⁹ Breast cancer risk increased slightly with estrogen-progesterone therapy but not until 4 to 5 years of use.⁴⁰ The attributable risk of breast cancer with estrogen-progesterone therapy in the Women’s Health Initiative was less than 1 additional case per 1000 women per year. Last, no difference was found between HRT users and nonusers in all-cause mortality or cause-specific mortality, such as cardiovascular and cancer mortality, after an 18-year cumulative follow-up.⁴¹

Limitations

The included clinical studies are not of the highest quality. All studies were of case-control design with small sample sizes. Selection bias can be a particular problem inherent in case-control studies, where it gives rise to noncomparability between cases and controls. This problem was critically evaluated during this review. To ensure comparability and to minimize heterogeneity, only studies of clear case and control definitions and similar methods and quality were included in the meta-analysis. The most important confounder was BMI, and we made sure that the studies included in the meta-analysis accounted for BMI in their analyses. As a result, the level of between-study heterogeneity in the meta-analysis was low, as indicated by the I² statistics.

In the absence of publication bias, the studies will be distributed symmetrically about the mean effect size because the sampling error is random. Publication bias due to smaller studies without statistically significant results not being published will result in an asymmetrical funnel plot with a visible gap. In the present study, the included studies were clustering at the top of the funnel plot, resulting in a gap at the bottom and therefore suggesting publication bias (Figure 3). However, since there were fewer than 10 studies in the meta-analysis, statistical testing for funnel plot asymmetry could not be performed because the test power would be too low to distinguish chance from real asymmetry.²⁸ Therefore, these results should be interpreted with caution owing to the limited number of studies included in the analysis.

Conclusions

The present systematic review and meta-analysis has found that HRT is associated with a significantly higher F₀ in postmenopausal women who were receiving HRT compared with those who were not. The treatment effect appeared to be more pronounced in women with a normal BMI. The primary indication for HRT is for treatment of systemic symptoms relating to menopause, not voice issues. Therefore, the decision to initiate HRT is not the responsibility of otolaryngologists. However, a basic knowledge of the benefits and risks associated with HRT, as well as its potential effects on voice, should be within a voice clinician’s armamentarium. This information would be helpful in counseling postmenopausal women with voice issues who may be considering HRT for other reasons associated with menopause. Future studies with a larger patient cohort and control population using patient-reported outcome measures and more sensitive objective voice measurements will further shed light on how HRT could affect the voice.

Supplement.

eAppendix. Literature Search Strategies