Introduction

Ankyloglossia, commonly known as tongue-tie, is an anatomic variation that restricts tongue mobility caused by a restrictive lingual frenulum,1 best described as a midline fold.2 The structure of the lingual frenulum consists of a “dynamic three-dimensional structure that varies in morphology on a spectrum” (ref. 2, p. 760). This anatomical variation of the lingual frenulum can be classified by height of the facial attachment to the ventral portion of the tongue, height of facial attachment to the tongue, height of attachment to the mandible, and length of the frenulum.2 This variability is classified by how the fibers of the lingual frenulum mobilize when the fascia is placed under tension.2 The restricted motion of the tongue can alter oral latch onto a breast or bottle nipple and impact swallowing, which may cause ineffective transfer of breast milk or formula.3 Improper latch to the breast or bottle can impact the infant’s ability to obtain adequate nutrition for growth and development and, during breastfeeding (BF), can cause maternal nipple pain and result in early cessation of BF.4 Prevalence of tongue-tie is not well understood and has been cited as low as 0.3%5 to as high as 16%.6

Professional consensus on the definition of tongue-tie and how it relates to symptoms of problematic feeding has been developed, noting that ankyloglossia is one possible reason for BF pain and ineffective latch.7 However, the use of the terms “posterior ankyloglossia” and “lip tie” have not reached professional consensus. In addition, the determination of which infants benefit most from frenotomy to correct tongue-tie remains unknown, largely due to the low level of evidence available, the variability in screening for tongue-tie, and the absence of a psychometrically sound diagnostic tool.8 Despite the lack of strong evidence, recognition of tongue-tie has increased, with a dramatic rise in treatment via frenotomy of 866% over the past two decades.9 The purpose of this study was to quantitatively synthesize the prevalence of tongue-tie in children aged <1 year and to examine the psychometric properties of the assessment tools used for diagnosing tongue-tie in these studies.

Methods

In February 2020, a literature search was conducted to review the published prevalence rates of tongue-tie. Databases searched include the Cumulative Index of Nursing and Allied Health Literature (CINAHL) and PubMed. The search terms ((tongue-tie) or (tongue tie) or (ankyloglossia)) and ((prevalence) or (incidence)) were entered into each database. The search was limited to articles involving humans and published in the English language with a study population of children aged <1 year. No limit was placed on the time of publication. Duplicate articles were removed.

Study selection and data extraction

To be included, articles must have been original research, reporting the prevalence of tongue-tie by screening a larger population. Articles were excluded if they did not include a prevalence rate, if the sample was only a subset of the population that already had a diagnosis of tongue-tie, if participants were aged >1 years, or if the study did not directly assess the sample for tongue-tie (e.g., a chart review or examination of Tenth Revision of the International Statistical Classification of Diseases and Related Health Problems diagnosis codes). Titles and abstracts were screened independently by two authors (R.R.H. and B.F.P.) using the inclusion and exclusion criteria. Articles that met inclusion and exclusion criteria or could not be evaluated based on the abstract were reviewed in full text. Both authors then reviewed the full-text articles independently. Any article that one author chose to include were reviewed by both authors for final decision.

The individual study characteristics were examined in the studies that met criteria for inclusion. The relevant features extracted included the study sample size and characteristics, region of the world where the study took place, diagnostic method used to identify tongue-tie, and reported prevalence of tongue-tie within the sample. Prevalence reported by sex was extracted, if available. The authors conducted this meta-analysis in concordance with the Preferred Reporting Items of Systematic Review and Meta-Analyses (PRISMA) standards of quality for reporting meta-analyses10 and the guidelines for Meta-Analysis and Systematic Reviews of Observational Studies.11

Examination of psychometric properties of diagnostic assessments

Following identification of the diagnostic tools used to identify tongue-tie, a review of the literature was conducted for each of the assessment tools to identify and review their psychometric properties. A search was conducted in both CINAHL and PubMed using the name of each tool separately (Coryllos, Kotlow, and the Assessment Tool for Lingual Frenulum Function) and ((psychometrics) or (reliability) or (validity) or (psychometric properties)). This search was limited to humans and English language. Articles reporting the psychometric properties of the assessment tools were examined to review the quality of the assessments.

Data analysis

An overall prevalence of tongue-tie across studies was examined using a random-effects meta-analysis of proportions approach. Prevalence was represented as a binomial distribution (tongue-tie present or absent). Subgroup meta-analysis was performed to examine the influence of sex (male or female) and diagnostic method on the prevalence of tongue-tie with studies as the unit of analysis. Random-effects tests of difference were then used to compare prevalence between studies of male and female infants and studies diagnosing tongue-tie by visual exam only compared with those using visual exam in addition to a standardized diagnostic measure.

For all analyses, the use of the random-effects model allowed examination of heterogeneity in and between studies, the appropriate method given that diagnosis is variable and may be subjective. Weighted estimates, 95% confidence intervals (CIs), z-tests (i.e., summary estimate divided by standard error of the summary estimate), and p values were calculated using Stata v16 (College Station, TX) and the metaprop command. Dispersion in effect size across studies (Q) along with an associated p value and variation in observed estimates attributable to heterogeneity (I2) were calculated. An alpha of 0.05 was considered statistically significant.

Assessment of risk of bias

Risk of bias in measurement was assessed through the evaluation of the psychometric properties of the assessment tools used to diagnose tongue-tie. The psychometric properties were extracted by one author (R.R.H.) and evaluated by two authors (R.R.H. and B.F.P.). Risk of bias in prevalence reporting was evaluated using an existing tool to examine external and internal validity and overall risk of bias.12 Overall bias was recorded as low, moderate, or high based on the 10 items critiqued within each study.

Results

The selection process for the literature included in this review is presented in the PRISMA flow diagram (Fig. 1). After screening titles and abstracts, 31 full-text articles were reviewed, and 15 were eligible for final inclusion in analyses. Included studies were conducted between 1990 and 2019 throughout the world, including in the United States, Thailand, Mexico, Spain, Turkey, Sweden, and the United Kingdom. Sample size ranged from 40013 to 3490.4 Table 1 presents relevant characteristics extracted from the included studies.

Fig. 1: PRISMA 2009 flow diagram.
figure 1

The PRISMA diagram details the search and selection process applied during our systematic literature search and critical review. From Moher et al.10. For more information, visit www.prisma-statement.org.

Full size image
Table 1 Study characteristics.
Full size table

Overall prevalence of tongue-tie

There were 15 studies reporting prevalence of tongue-tie for a collective total of 24,536 infants (Fig. 2). Irrespective of diagnostic method, the overall prevalence was 8% (95% CI 6–10%, z = 6.51, p = 0.00). There was significant (Q = 1617.37) and substantial (I2 = 99.13%) heterogeneity across studies reporting prevalence of tongue-tie.

Fig. 2: Overall prevalence of tongue-tie (N = 24,536).
figure 2

Forest plot of the studies documenting prevalence of tongue-tie. The analysis included 15 studies with a total of 24,536 cases.

Full size image

Prevalence by infant sex

Twelve studies reported infant sex. Of those, eight specifically compared prevalence between female and male infants. Four found a significantly higher incidence of tongue-tie in males,14,15,16,17 and four did not report a significant difference in rates by sex.4,13,18,19 Of the eight studies comparing incidence by sex, three were omitted from the analysis because a breakdown of distribution by sex within the full study sample was not provided, resulting in an inability to determine prevalence by sex in those studies.4,17,19 Of the 5 analyzed13,14,15,16,18 (n = 6319), prevalence of tongue was 7% in males (95% CI 2–12%, Q = 123.9) and 4% in females (95% CI 1–%, Q = 76.35). The difference in prevalence by infant sex was not statistically significant.

Prevalence by diagnostic method

Prevalence of tongue-tie in the 10 studies using visual examination alone (n = 14,426) was 7% (95% CI 4–10%, z = 4.47, p = 0.00, Fig. 3). The five studies using a more formal diagnostic method (n = 10,110) had a higher tongue-tie prevalence of 10% (95% CI 6–15%, z = 4.83, p = 0.00). The difference between these subgroups was not statistically significant (Q = 1.94, p = 0.16). Substantial variation remained across studies (visual examination method—Q = 1069.58, I2 = 99.16%; and standardized measure method—Q = 269.83, I2 = 98.52%).

Fig. 3: Prevalence of tongue-tie by diagnostic method (N = 24,536).
figure 3

Forest plot of studies comparing prevalence of tongue-tie by diagnostic method. The analysis included 15 studies with a total of 24,536 cases. Five studies (N = 10,110) used a formal diagnostic method; 10 studies (N = 14,426) diagnosed tongue-tie through visual examination.

Full size image

Psychometric properties of diagnostic assessments

Five studies diagnosed tongue-tie using a published assessment tool, the Hazelbaker Assessment Tool for Lingual Frenulum Function (ATLFF),4,18 Kotlow’s grading system,20,21 or Coryllos classification of tongue-tie severity.22 The majority of studies (n = 10) diagnosed based on visual examination, with variation in what criteria supported a diagnosis of tongue-tie. The ability of the infant to protrude the tongue out of the mouth was assessed in three studies,5,14,16 and length of frenulum attachment was measured in five studies without specific length necessary for diagnosis explicated.6,14,17,19,23 Two studies did not explicate criteria used for diagnosis.13,18

The ATLFF is a 12-item scale, with 5 items evaluating the appearance of the tongue and 7 evaluating tongue function. Scores range from 0 to 10 for appearance and from 0 to 14 for function, with higher scores indicating more optimal appearance and function. Hazelbaker recommends a diagnosis of tongue-tie for appearance scores ≤8 or function scores ≤11.24 Kotlow’s classification system requires the measurement of free tongue in millimeters, classifying tongue-tie severity of Class I–IV based on millimeters of free tongue.25 The Coryllos classification system defines four types of frenula based on the location of tongue anchoring. Coryllos recommends this evaluation be done alongside a BF assessment prior to making the decision to treat.26 Of these three tools, only one has published psychometric testing. The ATLFF has demonstrated moderate-to-excellent inter-rater agreement and correlated with the Bristol Breastfeeding Assessment Tool;27,28,29,30 however, function scores do not correlate with the number of mother–infant dyads who presented with BF problems caused by tongue-tie.4 The psychometric properties available on these tools are outlined in Table 2.

Table 2 Psychometric properties of assessment tools.
Full size table

Risk of bias

Risk of bias was noted to be moderate or high for all included studies, indicating that future research is likely to affect our confidence in prevalence estimates. While data were collected directly from subjects in all studies, sampling was not random, and it is impossible to determine whether the samples were representative of the population. The definition of tongue-tie differed between studies, inter-rater reliability was not reported, and there is no psychometric data on the instruments used to measure tongue-tie. Four of the studies failed to include a definition for tongue-tie, resulting in an inability to evaluate exactly what criteria were used to make a diagnosis. The remaining studies included a case definition; however, the studies were conducted prior to the development of professional consensus on the definition of tongue-tie, and uncertainty remains on the threshold of tongue-tie severity warranting treatment via frenotomy. This information is summarized in Table 3.

Table 3 Risk of bias in prevalence reporting by study.
Full size table

Egger’s test of bias interjected by small studies was significant, indicating that smaller studies reported larger estimates of the prevalence of tongue-tie (β = 13.2 ± 2.3; t = 5.76; p < 0.001). A funnel plot (Fig. 4) to evaluate possible publication bias did not demonstrate substantive asymmetry. Using meta-regression, there was no significance of year of study publication on the prevalence of tongue-tie reported in the literature (β = 0.002 ± 0.001; t = 2.02; p = 0.065, Fig. 5).

Fig. 4: A funnel plot of standard error by logit event rate.
figure 4

The analysis included 15 studies.

Full size image
Fig. 5: Prevalence estimates by year of publication.
figure 5

A figure demonstrating prevalence estimates of the 15 included studies by year of publication.

Full size image

Discussion

This meta-analysis of data from 24,536 infants found the overall prevalence of tongue-tie is 8% in children aged <1 year. The estimated predictive interval for overall prevalence suggests that future studies may expect to find a prevalence between 0 and 18% using similar approaches to those used in the included studies. Research by Mills and colleagues recognizes that the wide degree of variation in morphology limits our ability to determine a definitive classification for the diagnosis of abnormal frenulum anatomy.1 This may partly explain why prevalence of tongue-tie is variable between studies.

A genetic predisposition for tongue-tie has been identified, linked to “a mutated T-box transcription factor gene” (ref. 28, p. 46). The higher prevalence found in infants of male compared to female sex is consistent with the majority of studies published to date. The lack of statistical significance of this finding was likely due to the relatively small sample size in the subset of studies reporting prevalence by sex in conjunction with the high degree of heterogeneity in and between studies.

Although not statistically significant, the finding that prevalence of tongue-tie was numerically higher (10%) when a standardized assessment tool is used as compared to visual examination alone (7%) is an important finding. The main limitation in these analyses was that most of the studies used subjective visual examination to diagnose tongue-tie. Of the studies that used a standardized diagnostic tool, the psychometric properties of those tools are not adequate or have not been evaluated to date. The significant and substantive heterogeneity within and across studies is most likely due to poor measurement of the problem and indicates that differences are due to additional factors beyond the diagnostic method used. Direct comparisons of diagnostic evaluation studies are necessary to understand the clinical importance of the diagnostic method within the same study, not just by comparing studies with different measurement approaches.

Assessment of the quality of tongue-tie assessment tools includes reviewing reliability and validity of the measure. Using measures without strong psychometric properties may not appropriately or accurately measure the outcome variables of interest. Evidence of validity would support that the measure accurately diagnoses tongue-tie and not a related, but different construct (i.e., normal anatomical variation). Reliability supports that the score is appropriately and reasonably measuring tongue-tie severity and is the same over time and between alternate scorers. Failure to use a psychometrically sound instrument increases the likelihood that the problem under investigation is not accurately being measured. In this instance, the variation of prevalence rates of tongue-tie between studies may be partly due to the lack psychometric testing and utility of the ATLFF, Coryllos, and Kotlow instruments. It is also postulated that, in studies using visual examination alone, variability in cut-off scores and what constitutes a diagnosis of tongue-tie contributes to significant heterogeneity. These findings support the need for psychometric evaluation of available assessment tools or the development of a new, psychometrically sound, comprehensive tool to diagnose tongue-tie.

One universal diagnostic measure with accepted cut-off scores remains a major limitation in the published research on tongue-tie. Despite lack of psychometric evaluation, the use of any standardized assessment tool to diagnose tongue-tie resulted in an increased prevalence rate when compared to subjective visual examination alone. The ATLFF is the only diagnostic measure that has been evaluated psychometrically. The lack of psychometric evaluation limits the utility of the Kotlow and Coryllos scales to diagnose tongue-tie and future research is needed to assess these measures.

Understanding that tongue-tie diagnosis alone may not constitute difficulty feeding or maternal discomfort with feeding, a comprehensive, objective, and psychometrically sound evaluation of both infant feeding and maternal symptoms will help determine how tongue-tie impacts infant feeding, maternal comfort, and BF success. The consensus statement by Messner and colleagues was developed solely by otolaryngologists.1 More work is needed to incorporate the experience from other experts in the field of infant feeding, including lactation consultants, dentists, and speech and language practitioners. The creation of a diagnostic measure, taking into account the newly revised definition of tongue-tie and an interprofessional practice approach, is needed to more accurately capture the prevalence of tongue-tie and reduce heterogeneity between studies. Overall, the study of tongue-tie in infants has been limited by the absence of a universally accepted definition, lack of valid and reliable assessment measures, and the absence of comprehensive evaluation of infant feeding and maternal symptoms. Once these limitations are addressed, a large study is needed to determine the true prevalence of tongue-tie in infants and factors that may be associated with this anomaly.

Disclaimer

The material in this manuscript is original, has not been previously published, and has not been submitted for publication elsewhere.