Introduction
A cleft lip and palate (CLP) is a relatively common congenital malformation, with an incidence of about 1 in 700 newborns in the Caucasian population [
1]. While an isolated cleft lip primarily is an aesthetic problem, complete CLP may cause velopharyngeal insufficiency (VPI) which can interfere with speaking, breathing, and swallowing. Often CLP is associated with an articulation disorder generally regarded as compensation to an anatomical abnormity leading to VPI. This may lead to dysfunctions not only of the velopharyngeal sphincter, but also of the entire vocal tract [
2]. CLP-associated VPI typically leads to deviations in the resonance such as hypernasality or hyponasality, nasal air emission and weak pressure consonants, and compensatory articulation [
3].
VPI is an eminently clinical diagnosis. Any surgical intervention to correct the underlying anatomy should be planned, based on the combination of video-naso-pharyngoscopy and multiplanar videofluoroscopy. Magnetic resonance imaging is an emerging diagnostic tool, however, to date not widely adopted [
4,
5].
Typically, speech and language pathologists assess articulation placement and manner, and examine the oral cavity and the pharynx through direct vision and palpation of the hard and soft palate [
6]. However, the most important diagnostic procedure is a subjective evaluation by speech and language pathologists, who assess hyper-or hyponasality, nasal air emission and/or turbulances, consonant production errors, and voice disorders. For this subjective evaluation, the patients’ language background and age are important.
A more objective method may be the so-called nasometry, which measures nasalance instrumentally and can provide objective data for evaluating nasal resonance.
Different centers may use different speech parameters in testing, and therefore, they are not always comparable [
7]. Several assessment protocols have been described [
8‐
10], but none of them was widely adopted. The most accepted assessment protocol was developed by an international working group. Henningsson et al. [
11] reported an universal system for reporting speech outcome measures. The system includes five universal characteristics like hypernasaly, hyponasality, nasal air emission and/or turbulances, consonant production errors, and voice disorders. Concerning the grading of hypernasality, mild forms are appreciated primarily in certain vowels and may not be socially disturbing to the patient or family. Moderate hypernasality is audible with most vowels and deemed socially unacceptable. Severe hypernasality would usually prompt a recommendation for intervention by the patient and the clinician, because speech intelligibility is significantly diminished [
11]. A standard test regimen for the perceptual nasality evaluation is routinely performed using specific test items, like those from the Heidelberg Rhinophonia assessment form [
12]. Typically, for perceptive nasality assessment, a scale with three to five grades is used. Recently, Baylis et al. [
13] described that the use of analog visual scales is more accurate than the nasality documentation within a few categories only. Many studies investigate the use of instrumental objective diagnosis, which typically provide higher diagnostic resolutions compared to only few assessment classes diagnosed by speech therapists [
1,
14‐
16].
Our goal was to determine possible differences between perceptual and instrumental measurements in the east Austrian area. We selected various items of the Heidelberg Rhinophonia assessment form and determined their nasalance scores on the NasalView® System to explore their potential to assist the perceptive nasality assessment.
Discussion
By comparing the gold standard “perceptual diagnosis” with instrumental measurements, we found speech items, suitable for the instrumental follow-up assessments in patients with diagnosed hypernasality. The sensitivity of only four specific speech items is superior to the averaged sensitivity using additional speech items. Therefore, our findings may provide clinicians with a strategy to increase the sensitivity in the follow-up of patients with perceptively diagnosed nasality.
Considering the reliability of perceptually diagnosed nasality, multiple raters may be preferred compared to the diagnosis by only one evaluator [
20]. Therefore, we engaged two speech therapists for the perceptive assessment: an experienced speech therapist and a trainee. It is clear that a comparison between two speech therapists cannot be statistically significance, and therefore, our specific results cannot be extrapolated to other clinical centers. The comparison of the vowel [i:] and [a:] reflects a possible compensatory function of the tongue. During [i:] phonation, the tongue is positioned high close to the soft palate. During phonation of the vowel [a:], the tongue lies deep and far back [
21]. Our findings that the use of a nasal and oral sentence can provide clinically useful results confirm Bressmann [
18], who described that the use of only one short nasal and one short oral sentence does not compromise the validity of the examination. In addition, Watterson, et al. [
22] described that speech items with a minimum of six syllables are sufficient for valid determination of nasalance. Therefore, we selected a short sentence without nasal consonants (Table
1; item #7) and another with only nasal consonants (Table
1, item #8). To reveal possible additional information, we computed the so-called nasalance distance (ND) and nasalance ratio (NR) from the two sentences (item #7 and #8). The nasal sentence reveals the nasalance maximum, while the oral sentence indicates the individuals’ nasalance minimum [
23].
When plotting the four perceptually diagnosed groups against the NasalView
® measures, we observed a good correlation with some but not all individual speech items. The positive discrimination was best in the speech items 2 and 4 (Fig.
1). The ND and NR between speech item #7 and #8 corresponded rather poorly to the perceptual grading (Fig.
2). The mean of grouped speech items (four oral, four nasal, all 8 speech items) discriminated rather well between the perceptual and instrumental measures (Fig.
2).
The comparison of perceptive and instrumental assessments is a common approach to investigate the theory and praxis of speech assessment [
24]. Using ROC curves, we revealed that the nasalance associated with individual items and item groups corresponded at various degrees to the perceptual assessments.
NasalView
® measurements can correlate with perceptual assessments [
16]. Single speech items can correlate better than item groups (Table
2). However, it would be a key flaw to adopt our specific results to other clinical centers without further tests, simply because a comparison of only one speech language pathologist with an instrumental diagnostic method cannot reveal statistically significant results.
Each of the single nasal speech items #2, #4, #5, and the oral speech item #7, provided the same or better AUC measures compared to the four oral, or the four nasal speech items combined, or all 8 speech items in total (compare Fig.
3, with Fig.
4). The computed sensitivity and specificity for each single item analysis of speech item #2, #4, #5, and #7 score highest for sensitivity (96.43–100%), while the averaged “four oral” “four nasal” or “all eight speech items together” scored lower (89.29–96.43%; Table
2). The specificity of all results was below acceptable standards (Table
2).
Our findings with the NasalView System
® are only partially transferable to other systems due to differences between instruments [
25]. Because, in our study, the specificity of the instrumental nasalance measures was generally low, our findings support the previously published opinion, that instrumental assessment can never substitute, but only complement perceptual evaluation [
7].
Socio-cultural and regional slang affects the comparability between studies [
26,
27]. Seaver et al. [
28] and Watterson et al. [
29] considered regional differences and proposed the need for standardization for different regions. The usability of specific speech items may depend on the cultural and linguistic background of the assessed person [
30].
Nasalance can vary between individual speakers and regional dialects [
26]. However, in the follow-up situation where every patient provides his/her personal baseline and only intraindividual comparisons are relevant, a repeated instrumental assessment can document subtle changes. Therefore, we postulate that the instrumental assessment can be used independently of the patients´ specific linguistic background.
Bressmann et al. [
23] described the ND and the NR as useful values, which can provide additional nasalance information. In our study, the sensitivity and specificity of ND and NR between speech item #7 and #8 were low (Table
2; AUC < 0.73). However, ND and NR may depend on the individual test person and specific test items.
Instrumental measures could be superior to perceptual examination in two aspects: finer scale (0–100%, instead of 0–3 in perceptual assessments) and objectivity of the instrument. As Baylis et al. [
13] described that the use of a finer scale can provide more accurate documentation, instruments may provide better opportunities to quantitatively describe subtle improvements during follow-ups after therapeutic interventions.