Beyond Prevalence: Understanding the Relationship Between Early Anatomic Factors and the Likelihood for Cleft Speech Characteristics

ABSTRACT

Purpose

This study investigates early anatomic determinants influencing the likelihood of cleft speech characteristics (CSCs) in children with cleft palate with or without lip involvement (CP+/−L). The primary objective was to identify critical anatomic factors within the first year of life that impact the presence of CSCs at Age 3.

Methods

The Cleft Collective Longitudinal Cohort Study of children born with a CP+/−L was utilized. Data from 293 children were included in the analyses. Anatomic variables, including cleft type, pre‐operative cleft width and fistula status, were assessed in relation to CSCs. Logistic regression analyses were adjusted for confounding variables such as age, sex and hearing status.

Results

At 36 months, 64.9% of children with cleft palate exhibited CSCs. Anterior CSCs were most common (36.2%), while passive CSCs were least common (16.0%). Cleft type strongly influenced CSC prevalence, with UCLP and BCLP associated with higher odds of any CSC. Wider clefts were associated with passive, non‐oral and posterior CSCs. Fistula presence markedly increased the odds of CSCs, especially passive and posterior CSCs. Logistic regression analyses revealed strong associations between cleft type, cleft width, fistula and specific CSCs outcomes.

Conclusions

The multifactorial nature of CSC production in children with CP+/−L is underscored by distinct impacts of cleft type, pre‐operative cleft width and fistula status. Early, individualized interventions are paramount, emphasizing the need for proactive measures, including effective fistula management and speech therapy. Prioritizing prevention strategies for children with anatomic risk factors may reduce the likelihood of producing CSCs, contributing to more typical speech development and optimizing long‐term speech outcomes for affected children.

WHAT THIS PAPER ADDS

What is already known on this subject

• Children with cleft palate, with or without lip involvement (CP+/−L), frequently develop speech sound disorders due to anatomic variations affecting the craniofacial complex. Previous research has identified a range of speech difficulties in these children, with variability in the prevalence and types of cleft speech characteristics (CSCs). Less is known about the precise multifactorial impact of specific anatomic factors and their influence on the likelihood of developing CSCs, necessitating further investigation.

What this paper adds to existing knowledge

• This study identifies the multifactorial impact of early anatomic factors, including cleft type, cleft width and fistula status, on the likelihood of cleft speech characteristics (CSCs) at Age 3 in children with CP+/−L. This research enhances understanding of how these anatomical factors influence CSCs, offering valuable data to guide future studies and/or clinical practices aimed at optimizing speech development in children with cleft conditions.

What are the potential or actual clinical implications of this work?

• The findings emphasize the importance of early and individualized speech assessments and interventions for children with cleft conditions, particularly those with wider clefts and fistulas. Early detection and timely surgical and therapeutic interventions could reduce the likelihood of CSCs, improving long‐term speech outcomes and enabling more effective management plans tailored to the specific anatomic profiles for these children.

1. Introduction

Craniofacial anomalies, specifically cleft palate with or without lip involvement (CP+/−L), represent a group of congenital conditions that result in anatomic variations affecting both the form and function of the craniofacial complex. These anomalies often disrupt the normal function of the velopharyngeal mechanism, as well as the relationships between the maxilla and mandible. These anatomic differences, in turn, can have a substantial impact on speech development and production (Courtney et al. 1996; Hardin‐Jones and Jones 2005; Leavy et al. 2016; Molsted and Dahl 1990; Satoh et al. 2004). Speech production in children with CP+/−L may also be influenced by hearing function, as these children are at increased risk for conductive hearing loss, which can further impact speech perception and articulatory accuracy (Pereira and Sell 2024; Baker et al. 2021). Additionally, phonological delays and/or cognitive‐linguistic deficits can co‐occur in children with cleft conditions, contributing to the development of speech sound disorders (Southby 2024). These factors highlight the need for a multidimensional perspective when assessing and treating speech disorders in this population (Kummer 2014; Lancaster et al. 2020; Lien et al. 2023; Mason 2020).

The multifaceted nature of speech sound disorders and overall articulation outcomes in children with CP+/−L is further complicated by factors, such as the cleft type and pre‐operative cleft width, the timing of lip and palate surgery, the type of surgical intervention and the occurrence of post‐operative complications, such as fistula (Chapman et al. 2008; Gamble et al. 2023; Harding and Grunwell 1996; Hardwicke et al. 2014; Karling et al. 1993; Lam et al. 2012; Okhiria et al. 2022). Therefore, it is essential to gain insights into how these critical factors influence the development of speech sound production and specifically contribute to cleft speech characteristics (CSCs) (Sell 2005).

CSCs can include both active and passive speech errors, often observed in children with CP+/−L. This classification system is utilized as a component of speech analysis in the UK Speech Pattern Summary (John et al. 2006; Sell et al. 1999). Active CSCs are learned speech errors resulting from alternative articulatory targets in place of the intended target, while passive CSCs are an unintended product of a structural abnormality or dysfunction of the velopharyngeal region. Active CSCs are further categorized as anterior, posterior and non‐oral and relate to where articulation errors are made (Harding and Grunwell 1996; Hutters and Brøndsted 1987; Sell 2005). Anterior CSCs typically include dentalisation, lateralization or palatalization of the oral pressure consonants (Pereira and Sell 2024). Posterior CSCs involve the backing of sounds to velar or uvular places of articulation (Pereira and Sell 2023). Double articulation, when two articulatory placements are used simultaneously, is also considered a posterior CSC (Pereira and Sell 2024). Non‐oral CSCs are errors of articulatory placement occurring outside the oral cavity and include pharyngeal and glottal articulation, as well as nasal fricatives (Pereira and Sell 2024). In contrast, passive CSCs are alterations in the manner of production and often stem from structural issues (Sell 2005; Harding and Grunwell 1996; Hutters and Brøndsted 1987). These include weak oral pressure consonants and nasal realizations of oral pressure consonants. These speech sound errors often lead to reduced speech intelligibility (Chapman et al. 2016; John et al. 2006; Lien et al. 2023; Sell et al. 1999).

Importantly, the classification of CSCs has significant clinical implications. Passive CSCs are often indicative of anatomic deficits, secondary to velopharyngeal dysfunction (VPD) or altered dental/occlusal or fistula status, and may necessitate surgical intervention to correct. In contrast, active CSCs often require direct speech therapy to remediate. Active CSCs can develop in response to an underlying VPD, fistula, or dental/occlusal status but often persist even after surgical correction to the VPD has occurred since altered articulatory gestures are part of the motor plan for the intended target (Sell 2005; Lien et al. 2023; Kummer 2014). This distinction is critical, as effective treatment planning depends on identifying if a given CSC stems from an anatomical difference (i.e., velopharyngeal insufficiency), active/learned articulation patterns or a combination of both.

The impact of cleft type on speech outcomes has been reported to vary across studies. A recent meta‐analysis of patient factors influencing velopharyngeal function in children with CP+/−L found strong statistical evidence that rates of VPD (and secondary surgery) varied according to cleft phenotype, Robin Sequence and syndrome diagnosis, but noted overall poor data quality and follow‐up (Sainsbury et al. 2023). Prior studies have suggested no independent effect of cleft type (Brunnegård and Lohmander 2007; Pulkkinen et al. 2001; Timmons et al. 2001), while others document poorer speech outcomes in general for children with bilateral cleft lip and palate (BCLP) (Butterworth et al. 2023; Klintö et al. 2018; Mahoney et al. 2013; Nyberg et al. 2010). Prior research has also demonstrated that wider clefts are associated with a higher likelihood of requiring secondary surgeries (Bicknell et al. 2002; Yuan et al. 2016). Those with unilateral and BCLP are often reported as having worse speech compared to those with cleft palate only (CPO) (Butterworth et al. 2023; Timmons et al. 2001). Further, the presence of oronasal fistulae post‐primary palatoplasty can result in hypernasal speech, nasal air emission and articulation errors during speech production (Karling et al. 1993). These differences may be related to terminology used to classify speech outcomes and approaches to analyses, which also vary substantially across the literature (Kummer et al. 2012; Pereira and Sell 2024). Some studies focus on velopharyngeal insufficiency as a speech outcome, while others focus on resonance, airway patency and/or general characteristics of speech sound production (Allori et al. 2017; de Blacam et al. 2018; de Blacam et al. 2022; Pereira and Sell 2024; Sell et al. 2015).

Numerous investigations have additionally sought to identify the prevalence of speech errors in children with CP+/−L and these studies have collectively documented the occurrence of speech difficulties (Dalston et al. 1992; Harding and Grunwell 1996; Hardin‐Jones and Jones 2005; LeBlanc and Cisneros 1995; Lien et al. 2023; Pereira and Sell 2024). These investigations have identified a variable landscape, with reported prevalence rates of speech sound errors fluctuating from 10% to over 25% in children with repaired cleft palate, and considerably higher rates, reaching up to 47%, in individuals with VPD (Hardin‐Jones and Jones 2005; Nyberg et al. 2014; Sell et al. 2015; Willadsen et al. 2017). Rates as high as 74% in adults with unrepaired or poorly repaired cleft palate have also been reported (Lou et al. 2021).

Despite the documented prevalence rates, the precise influence of specific early anatomic factors on articulation outcomes, specifically, the likelihood of developing CSCs, remains relatively unknown (Bicknell et al. 2002; Yuan et al. 2016; Butterworth et al. 2023; Timmons et al. 2001; Karling et al. 1993). Therefore, the purpose of this study is to identify how the multifactorial presence of cleft type, pre‐operative cleft width and post‐operative fistula status may influence the presence of CSCs at Age 3 and to identify the likelihood of developing specific CSCs given patient‐specific anatomy.

2. Materials and Methods

A collaborative study was developed between the University of Virginia, the University of Wyoming and the University of Bristol for analyses of data from the Cleft Collective Cohort Study (Project Number CC047). The Cleft Collective Cohort Study is an observational study in which data are collected from routine clinical care and from participant self‐report.

All ethical approvals were obtained prior to study initiation. The Cleft Collective Cohort Study obtained ethical approvals for recruitment and data collection through the South West Central Bristol Ethics (REC approval 13/SW/0064). Ethical approval for the secondary analysis of data reported in this study and collaborative institutional affiliation agreements were obtained (IRB‐SBS#5168). The resource is available for clinical and academic communities to address a range of cleft‐related research questions. Details related to the data are available through an online data dictionary (http://www.bristol.ac.uk/cleft‐collective/professionals/access/).

2.1. The Cleft Collective

The Cleft Collective is a large prospective observational cohort study of children with CP+/−L, investigating causes of cleft, the best treatments and the impacts of clefting on individuals affected and their families. Children are recruited to the Cleft Collective from the 16 regional cleft centres in the United Kingdom. These 16 regional centres provide specialist care from a multidisciplinary team comprising surgeons, speech and language therapists (SLTs), dentists, orthodontists, geneticists, psychologists, nurses and audiologists. The centres work closely together to ensure consistency in service provision and to monitor outcomes (Davies et al. 2024). Where necessary, the specialist teams liaise with local providers to support the delivery of community‐based SLT, dental and other services as needed. The Cleft Collective includes data from children recruited from all 16 centres and their families.

Children are recruited either just before or soon after birth (the Birth Cohort) or at the 5‐year audit clinic (5‐year cohort). Data collected in the study includes parent and child self‐report questionnaires, clinical data from surgeons and SLTs and biological samples. The resource includes a nested substudy, the Cleft Collective Speech and Language study (CC SL). Children recruited to the Cleft Collective Birth Cohort who were born with CP ± L are eligible to be consented to the CC SL (Wren et al. 2018). More detailed data on speech are collected in the CC SL study. These data include recordings for transcription and analysis. Eligibility criteria for participation in the CC SL study are broad, and there has been no reported selection bias by those recruiting at the clinical settings.

Clinical data on the children recruited to the CC SL are collected at Ages 18–24 and 36 months. Children with CP+/−L in the United Kingdom are invited for an assessment at 3 years of age to monitor progress in speech and language. During this assessment, the presence of each type of CSC is documented.

2.2. Participants and Variables of Interest

Children were eligible for inclusion if they had a diagnosis of CP ± L and speech assessment data recorded at 3 years of age. Data from children meeting inclusion criteria were extracted from the Cleft Collective resource and included demographic, surgical and speech data, consistent with the Cleft Collective protocol outlined by Davies et al. (2024). A flow diagram illustrating the derivation of participants meeting inclusion criteria from the Cleft Collective is provided in the Supporting Materials (Figure S1). A total of 68 participants who were over 3 years old but whose 36‐month SLT assessment data had not yet been shared with the CC SL study team were excluded. Those with concomitant syndromic diagnoses were excluded as well (see Figure S1). All participants within the sample underwent primary palatoplasty within a similar timeframe and received similar surgical repair for primary palatoplasty. The median age at palatoplasty was 10 months, with an interquartile range between 8 and 11.4 months. Data were available for 199 children from either the surgeon or parental report. Data on the surgical technique used for palatoplasty were available for 129 children, of which 89.9% received an intravelar veloplasty.

Data from the 36‐month speech evaluation was selected as a time point for the CSC outcome variable, given that speech outcomes at Age 3 are a critical early indicator of a child's overall speech development trajectory and can provide insights into the success of early surgical interventions (Persson et al. 2006). By this age, children typically have undergone significant development in speech and language skills, making it an appropriate time to document the presence of CSCs (Lancaster et al. 2020). To assess the influence of anatomic factors in the first year of life on speech outcomes at Age 3, key anatomic variables were included in the analyses. Anatomic variables of interest included the following: cleft type (CPO, unilateral cleft lip and palate (UCLP), BCLP), pre‐operative palatal cleft width (pre‐operative measurement in millimetres between soft edge width of cleft) and post‐palatoplasty fistula status (present/absent). These variables, along with age in months at speech assessment, sex, hearing status and information on the type and presence of CSCs, were extracted from the Cleft Collective resource.

Data on CSC presence and type were collected by SLTs live during routine clinical encounters, which followed standardized clinical protocols. All SLTs were trained in speech assessment methods and skilled in phonetic transcription and classifying CSCs. These clinicians conducted speech assessments across multiple environments (e.g., regional cleft clinic, in‐home environment, local clinics, etc.). CSCs were assessed using both formal and informal assessments, including word/sentence repetition tasks, conversational speech tasks and single‐word articulation tests. Formal assessments of speech sound productions included the Great Ormond Street Speech Assessment (GOS.SP.ASS) (John et al. 2006; Sell et al. 1994; Sell et al. 1999) and the Phonological Assessment of Child Speech (PACS) (Grunwell 1985). All CSCs were classified according to the standardized Cleft Audit Protocol for Speech Augmented (CAPS‐A) framework (John et al. 2006).

Data on cleft type were obtained from multiple sources in the Cleft Collective, including parental questionnaires and surgical forms. Parental report on cleft type detailed whether the child was born with a CPO, a UCLP or a BCLP. Surgical report of cleft classification was obtained using LAHSHAL coding, which details the extent and location of the cleft (Mcbride et al. 2016). LAHSHAL data were available for 84% of the sample, where data on cleft type had been obtained from both parental questionnaire, and surgical report data were validated between sources. Data on pre‐operative palatal cleft width, specifically at the hard/soft palate junction, were completed at the time of the child's surgery and collected from surgical forms. The Cleft Collective worked closely with surgeons from across the United Kingdom to agree on a process for obtaining this measurement. A diagram specifying where the measurements should be taken is displayed on the reverse of the surgical forms. The presence of a post‐operative fistula was assessed by the SLT at the 36‐month assessment and recorded on the 36‐month assessment forms as diagnosed, suspected, none or unable to see. Hearing status was obtained from the 36‐month assessment forms completed by the SLTs. SLTs were asked whether the child had any history of hearing loss, based on their clinical knowledge and case reports of the child and on parent report. This was recorded as a binary (yes/no) variable.

2.3. Statistical Analyses

Data analyses were conducted using Stata v17. Descriptive statistics were used to identify the prevalence of CSCs relative to patient‐specific anatomic factors. Shapiro–Wilk tests of normality were performed on pre‐operative cleft width and the child's age at speech assessment. A Levene's test of homogeneity was further performed on pre‐operative cleft width by each CSC outcome of interest. Associations between the patient‐specific factors and the CSCs were analysed using Chi‐square tests of association, Student t‐tests and Wilcoxon rank sum tests. Binary logistic regression was used to determine the effect size between each of the patient‐specific factors and the CSCs. Outcomes explored included (1) the presence of any CSC, (2) presence of at least one anterior CSC, (3) presence of at least one posterior CSC, (4) presence of at least one non‐oral CSC and (5) presence of at least one passive CSC. Exposures of interest comprised cleft subtype, pre‐operative cleft width and fistula status. Logistic regression models were adjusted for the child's age at speech evaluation, biological sex, cleft subtype and hearing status (Lithovius et al. 2015; Sainsbury et al. 2023; Schönweiler et al. 1999; Baker et al. 2021). For each logistic regression analysis, Model 1 was unadjusted and only accounted for the outcome and exposure. Model 2 was minimally adjusted to account for cleft type (where cleft type was not the exposure), biological sex and age at speech assessment. Model 3 was fully adjusted to account for the variables included within Model 2 and for hearing status. A multivariable model was additionally completed to assess the combined influence of all anatomic predictors, inclusive of cleft type, pre‐operative cleft width and fistula status, on CSC outcome. The multivariable analysis (Model 4) was also fully adjusted to include the child's age at speech evaluation, biological sex and hearing status. As is common with observational cohort studies, the inclusion of different exposures and confounders resulted in differing sample sizes per model. To further account for the differing sample sizes throughout the paper and, in turn, the different sample characteristics that were potentially unaccounted for within the logistic regression analyses, a sensitivity analysis was additionally performed with the same sample used in Model 4 (the smallest sample size, n = 116). Results of the sensitivity analysis are presented within Supporting Materials (Tables S1–S3). Model confounders are summarized in Table S4. For all analyses, we report p values and describe the strength of evidence against the null hypothesis on a continuum, rather than using a binary ‘significant/non‐significant’ threshold. Following guidance by Sterne and Davey Smith (2001), we interpret smaller p values as stronger evidence and use the following descriptors for clarity: p < 0.01 strong, 0.01–<0.05 moderate, 0.05–<0.10 limited/weak and ≥0.10 little/no evidence.

3. Results

3.1. Sample Description

The sample comprised data from 293 children who were part of the CC SL substudy. Data on cleft subtype and biological sex were available for all participants. Within the sample, 129 children (44.0%) were born with CPO, 115 children (39.3%) were born with a UCLP and 49 children (16.7%) were born with a BCLP. The sample comprised more males (n = 172, 58.7%) than females (n = 121, 41.3%). Pre‐operative cleft width measurements were available for 167 children and were approximately normally distributed (W = 0.99, p = 0.465). The mean pre‐operative cleft width measurement was 10.9 mm with a standard deviation of 3.83 mm. When the sample was split by the presence of each CSC outcome, the variance for pre‐operative cleft width was homogenous between groups (p values ranged between 0.121 and 0.537). Data on fistula status at the time of speech assessment were available for 235 children, of whom 50 (21.3%) children had a diagnosed fistula. Data on the child's age at speech assessment were available for 270 children. Data on age at assessment were skewed (W = 0.81, p < 0.001), the median age was 37 months with an interquartile range of 36–39 months. Data on hearing status were available for 276 children, of whom 159 (57.6%) were reported as having a history of diagnosed hearing loss.

3.2. Type and Distribution of CSC Production at 36 Months of Age

At least one CSC was documented in 64.9% of cases (n = 190). The most common CSC present was an anterior CSC, reported in 36.2% of the sample. The least common CSC present was a passive CSC, reported in 16.0% of the sample. Figure 1 demonstrates the distribution and prevalence of CSCs in the study sample, as a whole, at 36 months of age.

FIGURE 1.

Distribution and prevalence of cleft speech characteristics (CSCs) at 36 months of age.

3.3. Influence of Cleft Type on CSCs

Descriptive statistics suggest an increased prevalence of CSCs, the more involved the cleft (Figure 2). Strong statistical evidence was found to suggest a difference in the prevalence of having at least one CSC present between cleft types (p < 0.001). The Pearson Chi‐square (χ ²) test of association identified strong statistical evidence to suggest a difference in the prevalence of having passive CSCs or posterior CSCs present between cleft types (p = 0.016; p < 0.001, respectively). Weak statistical evidence was found to suggest a difference in the prevalence of non‐oral CSCs between cleft types (p = 0.088). Statistical evidence was lacking to suggest a difference in the prevalence of anterior CSCs between cleft types (p = 0.238) (see Table 1: presence of CSCs by Cleft Type).

FIGURE 2.

Prevalence of CSCs by cleft type at 36 months of age. BCLP, bilateral cleft lip and palate; CPO, cleft palate only; CSCs, cleft speech characteristics; UCLP, unilateral cleft lip and palate.

TABLE 1.

Presence of CSCs by cleft type.

Any CSC present by cleft type
	No CSCs present	CSCs present	Total
CPO	62 (48.1%)	67 (51.9%)	129
UCLP	33 (28.7%)	82 (71.3%)	115
BCLP	8 (16.3%)	41 (83.7%)	49
Pearson χ 2 = 19.15 Pr < 0.001

Anterior CSCs by cleft type
	No CSCs present	CSCs present	Total
CPO	89 (69.0%)	40 (31.0%)	129
UCLP	70 (60.9%)	45 (39.1%)	115
BCLP	28 (57.1%)	21 (42.9%)	49
Pearson χ 2 = 2.87 Pr = 0.238

Posterior CSCs by cleft type
	No CSCs present	CSCs present	Total
CPO	115 (89.2%)	14 (10.9%)	129
UCLP	81 (70.4%)	34 (29.6%)	115
BCLP	29 (59.2%)	20 (40.8%)	49
Pearson χ 2 = 22.18 Pr < 0.001

Non‐oral CSCs by cleft type
	No CSCs present	CSCs present	Total
CPO	98 (76.0%)	31 (24.0%)	129
UCLP	74 (65.4%)	41 (35.7%)	115
BCLP	31 (63.3%)	18 (36.7%)	49
Pearson χ 2 = 4.86 Pr = 0.088

Passive CSCs by cleft type
	No CSCs present	CSCs present	Total
CPO	117 (90.1%)	12 (9.3%)	129
UCLP	92 (80.0%)	23 (20.0%)	115
BCLP	37 (75.5%)	12 (24.5%)	49
Pearson χ 2 = 8.29 Pr = 0.016

Abbreviations:  χ ², Pearson Chi square; BCLP, bilateral cleft lip and palate; CPO, cleft palate only; CSC, cleft speech characteristics; Pr, probability value; UCLP, unilateral cleft lip and palate.

3.3.1. Logistic Regression Exploring the Relationship Between Cleft Type and CSCs

Logistic regression for each CSC outcome (any CSC, passive, non‐oral, anterior, posterior CSCs) was explored using the three models described above. Model 1 comprised n = 293 children, Model 2 comprised n = 270 children and Model 3 comprised n = 255 children. Across all models, strong evidence was found to suggest that having at least one CSC present was associated with cleft type (see Table 2). When adjusting for biological sex, age at assessment and hearing status Model 3 suggested the odds of having at least one CSC present increased by 112% for UCLP cases in comparison to CPO cases (OR_model3: 2.12, 95% CIs: 1.16–3.90, p = 0.015). The odds of having at least one CSC present were increased by 399% for children with BCLP in comparison to children with CPO (OR_model3: 4.99, 95% CIs: 1.89–13.18, p = 0.001). Strong statistical evidence was found to suggest that having at least one posterior CSC was associated with cleft type. Model 3, the fully adjusted model, suggested the odds of having at least one posterior CSC increased by 270% for UCLP cases in comparison to CPO cases (OR_model3: 3.70, 95% CIs: 1.72–7.96, p = 0.001). The odds of having at least one posterior CSC were increased by 627% for BCLP cases in comparison to CPO cases (OR_model3: 7.27, 95% CIs: 2.95–17.90, p < 0.001). Table 2 highlights the results for each model.

TABLE 2.

Regression models assessing cleft type and CSC outcome (reference category cleft palate only).

		OR	p	95% CIs
At least one CSC reported
Model 1 (n = 293)	UCLP	2.30	0.002	1.35	3.91
	BCLP	4.74	<0.001	2.06	10.90
Model 2 (n = 270)	UCLP	2.28	0.005	1.28	4.07
	BCLP	4.80	0.001	1.94	11.86
Model 3 (n = 255)	UCLP	2.12	0.015	1.16	3.90
	BCLP	4.99	0.001	1.89	13.18
Model 4 (n = 116)	UCLP	1.37	0.511	0.53	3.54
	BCLP	3.39	0.112	0.75	15.32
Anterior CSCs
Model 1 (n = 293)	UCLP	1.43	0.184	0.84	2.43
	BCLP	1.67	0.139	0.85	3.29
Model 2 (n = 270)	UCLP	1.58	0.118	0.89	2.82
	BCLP	1.70	0.158	0.81	3.56
Model 3 (n = 255)	UCLP	1.59	0.139	0.86	2.92
	BCLP	1.72	0.170	0.79	3.73
Model 4 (n = 116)	UCLP	1.31	0.574	0.51	3.39
	BCLP	1.59	0.470	0.45	5.63
Posterior CSCs
Model 1 (n = 293)	UCLP	3.45	<0.001	1.74	6.83
	BCLP	5.67	<0.001	2.56	12.55
Model 2 (n = 270)	UCLP	3.52	0.001	1.69	7.33
	BCLP	7.24	<0.001	3.06	17.15
Model 3 (n = 255)	UCLP	3.70	0.001	1.72	7.96
	BCLP	7.27	<0.001	2.95	17.90
Model 4 (n = 116)	UCLP	2.89	0.082	0.88	9.53
	BCLP	5.41	0.020	1.31	22.35
Non‐oral CSCs
Model 1 (n = 293)	UCLP	1.75	0.0480	1.00	3.05
	BCLP	1.84	0.092	0.90	3.72
Model 2 (n = 270)	UCLP	1.50	0.178	0.83	2.73
	BCLP	1.71	0.162	0.81	3.62
Model 3 (n = 255)	UCLP	1.46	0.236	0.78	2.75
	BCLP	1.72	0.176	0.78	3.80
Model 4 (n = 116)	UCLP	1.25	0.682	0.43	3.62
	BCLP	1.38	0.639	0.36	5.22
Passive CSCs
Model 1 (n = 293)	UCLP	2.44	0.020	1.15	5.16
	BCLP	3.16	0.010	1.31	7.63
Model 2 (n = 270)	UCLP	2.18	0.052	0.99	4.79
	BCLP	2.69	0.038	1.06	6.88
Model 3 (n = 255)	UCLP	2.08	0.081	0.91	4.75
	BCLP	2.65	0.050	1.00	7.01
Model 4 (n = 116)	UCLP	1.14	0.876	0.22	5.83
	BCLP	3.18	0.181	0.58	17.26

Note: Where Model 1 is unadjusted; Model 2 is adjusted for age at speech assessment and biological sex; Model 3 is adjusted for age at speech assessment, biological sex and hearing status; Model 4 is adjusted for pre‐operative cleft width (mm), fistula status, age at speech assessment, biological sex and hearing status.

Abbreviations: BCLP, bilateral cleft lip and palate; CIs, confidence intervals; CSC, cleft speech characteristics; OR, odds ratio; p, p value; UCLP, unilateral cleft lip and palate.

3.4. Influence of Cleft Width on CSCs

Mean cleft width and standard deviation associated with the presence or absence of each type of CSC are reported in Table 3. A wider mean cleft width was reported in children who presented with at least one passive CSC, posterior CSC, non‐oral CSC or any CSC compared to those children who did not present with the specific CSC. Student t‐tests suggested strong statistical evidence of a difference in mean cleft width between children who had at least one passive CSC compared to those who did not (p = 0.002). Moderate statistical evidence was found to suggest a difference in mean cleft width between children who had at least one posterior CSC compared to those children who did not (p = 0.040) and for those children who had at least one non‐oral CSC compared to those children who did not (0.022). Weak statistical evidence was found to suggest a difference in mean cleft width between children who had at least one CSC compared to children who did not have any CSCs present (p = 0.060).

TABLE 3.

Influence of cleft width on CSCs.

	Mean soft edge width of cleft in millimetres (SD)		p value
	No CSC present	CSC present
Any CSC	10.18 mm (4.07 mm)	11.35 mm (3.65 mm)	0.060
Anterior CSC	11.08 mm (3.91 mm)	10.75 mm (3.71 mm)	0.598
Posterior CSC	10.61 mm (3.85 mm)	12.05 mm (3.59 mm)	0.040 *
Non‐oral CSC	10.51 mm (3.87 mm)	12.00 mm (3.55 mm)	0.022 *
Passive CSC	10.56 mm (3.81 mm)	13.16 mm (3.17 mm)	0.002 *

Abbreviation: CSCs, cleft speech characteristics.

^* α = 0.05.

3.4.1. Logistic Regression Exploring the Relationship Between Cleft Width and CSCs

Model 1, the unadjusted model, comprised n = 167 children, Model 2, adjusted for cleft subtype, biological sex and age at assessment, comprised n = 151 children and Model 3, fully adjusted for variables detailed in Model 2 as well as hearing status, comprised n = 145 children. Strong statistical evidence was found to suggest that having at least one passive CSC was associated with the soft edge width of the cleft. Model 3 suggested the odds of having at least one passive CSC increased by 23% for each mm increase in soft edge width of cleft (OR_model3: 1.23, 95% CIs: 1.06–1.42, p = 0.006). Results for each model are provided in Table 4.

TABLE 4.

Regression models assessing cleft width and CSC outcome.

	OR	p	95% CIs
At least one CSC reported
Model 1 (n = 167)	1.09	0.062	1.00	1.18
Model 2 (n = 151)	1.08	0.091	0.99	1.19
Model 3 (n = 145)	1.09	0.084	0.99	1.21
Model 4 (n = 116)	1.05	0.391	0.94	1.17
Anterior CSCs
Model 1 (n = 167)	0.98	0.595	0.90	1.06
Model 2 (n = 151)	0.99	0.831	0.90	1.08
Model 3 (n = 145)	1.00	0.949	0.91	1.10
Model 4 (n = 116)	0.96	0.403	0.86	1.06
Posterior CSCs *
Model 1 (n = 167)	1.11	0.042	1.00	1.22
Model 2 (n = 151)	1.11	0.080	0.99	1.25
Model 3 (n = 145)	1.10	0.122	0.97	1.24
Model 4 (n = 116)	1.07	0.318	0.93	1.23
Non‐oral CSCs **
Model 1 (n = 167)	1.11	0.024	1.01	1.22
Model 2 (n = 151)	1.10	0.048	1.00	1.22
Model 3 (n = 145)	1.12	0.039	1.01	1.25
Model 4 (n = 116)	1.16	0.021	1.02	1.32
Passive CSCs **
Model 1 (n = 167)	1.22	0.003	1.07	1.39
Model 2 (n = 151)	1.22	0.005	1.06	1.41
Model 3 (n = 145)	1.23	0.006	1.06	1.42
Model 4 (n = 116)	1.24	0.029	1.02	1.50

Note: Where Model 1 is unadjusted; Model 2 is adjusted for cleft type, age at speech assessment and biological sex; Model 3 is adjusted for cleft type, age at speech assessment, biological sex and hearing status; Model 4 is adjusted for cleft type, fistula status, age at speech assessment, biological sex and hearing status.

Abbreviations: CIs, confidence intervals; CSC, cleft speech characteristics; OR, odds ratio; p, p value.

^* p ≤ 0.05.

^** p ≤ 0.01.

3.5. Influence of Fistula on CSCs

The timing of assessing the presence of a post‐operative fistula by the SLT varied, but was approximately 24–30 months at primary palate repair. Descriptive statistics suggest that the presence of CSCs is more likely when a fistula has been diagnosed (Figure 3). Strong statistical evidence was found to suggest a difference in the prevalence of having at least one CSC present with a diagnosed fistula (p < 0.001). Strong statistical evidence was also found to suggest a difference in the prevalence of having passive CSCs or posterior CSCs present when a fistula diagnosis was also present (p < 0.001, p = 0.002, respectively). Moderate statistical evidence was found to suggest a difference in the prevalence of having non‐oral CSCs and fistula diagnosis as well (p = 0.013). Weak statistical evidence was found to suggest a difference in the prevalence of having anterior CSCs with a diagnosed fistula (p = 0.063) (Table 5).

FIGURE 3.

Prevalence of cleft speech characteristics (CSCs) based on the presence or absence of fistula.

TABLE 5.

Influence of fistula on CSCs.

Any CSCs present by fistula diagnosis
	No CSCs present	Any CSCs present	Total
No fistula	78 (42.6%)	107 (57.8%)	185
Fistula diagnosed	7 (14.0%)	43 (86.0%)	50
Pearson χ 2 = 13.5217 Pr < 0.001

Anterior CSCs by fistula diagnosis
	No CSCs present	Anterior CSCs present	Total
No fistula	126 (68.1%)	59 (31.9%)	185
Fistula diagnosed	27 (54.0%)	23 (46.0%)	50
Pearson χ 2 = 3.4486 Pr = 0.063

Posterior CSCs by fistula diagnosis
	No CSCs present	Posterior CSCs present	Total
No fistula	150 (81.1%)	35 (18.9%)	185
Fistula diagnosed	30 (60.0%)	20 (40.0%)	50
Pearson χ 2 = 9.7580 Pr = 0.002

Non‐oral CSCs by fistula diagnosis
	No CSCs present	Non‐oral CSCs present	Total
No fistula	137 (74.1%)	48 (26.0%)	185
Fistula diagnosed	28 (56.0%)	22 (44.0%)	50
Pearson χ 2 = 6.1345 Pr = 0.013

Passive CSCs by fistula diagnosis
	No CSCs present	Passive CSCs present	Total
No fistula	163 (88.1%)	22 (11.9%)	185
Fistula diagnosed	32 (64.0%)	18 (36.0%)	50
Pearson χ 2 = 16.1972 Pr < 0.001

Abbreviations: χ ², Pearson Chi‐square; CSCs, cleft speech characteristics; Pr, probability value.

3.5.1. Logistic Regression Exploring the Relationship Between Fistula Status and CSCs

Model 1, unadjusted model, comprised n = 235 children, Model 2, adjusted for cleft subtype, biological sex and age at assessment, comprised n = 216 children and Model 3, fully adjusted for variables detailed in Model 2 as well as hearing status, comprised n = 209 children. Moderate statistical evidence was found to suggest that having at least one CSC present was associated with fistula diagnosis. Model 3 suggested the odds of having at least one CSC present increased by 212% for those diagnosed with a fistula compared to those without a fistula (OR_model3: 3.12, 95% CIs: 1.26–7.73, p = 0.014). Strong statistical evidence was found to suggest that having at least one passive CSC was associated with fistula diagnosis. Model 3 suggested the odds of having at least one passive CSC increased by 204% for those diagnosed with a fistula compared to those without a fistula (OR_model3: 3.04, 95% CIs: 1.37–6.75, p = 0.006). Results for each model are provided in Table 6.

TABLE 6.

Regression models assessing fistula status and CSC outcome (reference category fistula did not form).

	OR	p	95% CIs
Any CSC reported
Model 1 (n = 235)	4.48	0.001	1.91	10.48
Model 2 (n = 216)	3.30	0.009	1.35	8.03
Model 3 (n = 209)	3.12	0.014	1.26	7.73
Model 4 (n = 116)	2.75	0.083	0.88	8.64
Anterior CSCs
Model 1 (n = 235)	1.82	0.065	0.96	3.44
Model 2 (n = 216)	1.62	0.165	0.82	3.21
Model 3 (n = 209)	1.68	0.154	0.82	3.41
Model 4 (n = 116)	1.24	0.661	0.47	3.29
Posterior CSCs
Model 1 (n = 235)	2.86	0.002	1.45	5.61
Model 2 (n = 216)	2.05	0.053	0.99	4.24
Model 3 (n = 209)	2.22	0.036	1.05	4.66
Model 4 (n = 116)	2.10	0.146	0.77	5.71
Non‐oral CSCs
Model 1 (n = 235)	2.24	0.015	1.17	4.29
Model 2 (n = 216)	1.78	0.102	0.89	3.54
Model 3 (n = 209)	1.67	0.163	0.81	3.41
Model 4 (n = 116)	1.60	0.346	0.60	4.28
Passive CSCs
Model 1 (n = 235)	4.17	0.000	2.01	8.64
Model 2 (n = 216)	3.17	0.003	1.46	6.88
Model 3 (n = 209)	3.04	0.006	1.37	6.75
Model 4 (n = 116)	4.68	0.017	1.31	16.71

Abbreviations: CIs, confidence intervals; CSC, cleft speech characteristics; OR, odds ratio; p, p value.

Note: Model 1 is unadjusted; Model 2 is adjusted for cleft type, age at speech assessment and biological sex; Model 3 is adjusted for cleft type, age at speech assessment, biological sex and hearing status; Model 4 is adjusted for cleft type, pre‐operative cleft width (mm), age at speech assessment, biological sex and hearing status.

3.6. Combined Influence of Anatomic Factors on CSCs

The multivariable model (Model 4) demonstrated the smallest sample size, given the number of factors assessed (n = 116). When considering the combined analysis of differing anatomic profiles, inclusive of cleft type, cleft width and fistula status, Model 4 is suggestive that cleft type and fistula status are the primary factors driving CSCs, with increased odds for the presence of any CSC. Findings indicate that the odds of having CSCs present increase by 175% when a child has been diagnosed with a fistula [OR_model4: 2.75, 95% CIs: 0.88–8.64, p = 0.083]. Cleft width demonstrated little involvement with this association in Model 4. However, the lack of statistical evidence for the impact of cleft width in the results for Model 4, compared to the other models, is likely due to the reduced sample size and subsequently reduced power of this multivariable model. In contrast, the presence of passive CSCs was associated with both increased cleft width and having a fistula diagnosis within this multivariable model. The odds of passive CSCs being present increase with wider pre‐operative cleft widths (increased odds of 24% for each millimetre increase in width [OR_model4: 1.24, 95% CIs: 1.02–1.50, p = 0.029]) and when a fistula has been diagnosed (increased odds of 368% when a fistula has been diagnosed [OR_model4: 4.68, 95% CIs: 1.31–16.71, p = 0.017]). Cleft type appeared to have little impact on the presence of passive CSCs in this model. The presence of non‐oral CSCs was also associated with having a wider cleft. The odds of a non‐oral CSC being present increase by 16% for each millimetre increase in pre‐operative cleft width [OR_model4: 1.16, 95% CIs: 1.02–1.32, p = 0.021]. The presence of anterior CSCs did not appear to be associated with any anatomical factor within Model 4. Similarly, this lack of association was seen across Model 3 analyses. The presence of posterior CSCs was associated with cleft type within Model 4. The odds of having a posterior CSC were increased by 189% for UCLP cases compared to CPO cases and increased by 441% for BCLP cases compared to CPO cases [UCLP OR_model4: 2.89, 95% CIs: 0.88–9.53, p = 0.082; BCLP OR_model4 5.41, 95% CIs: 1.31–22.35, p = 0.020].

3.6.1. Sensitivity Analysis

Similar patterns were observed in the main analyses (Models 1–3) and the sensitivity analysis, suggesting that there is little difference in the unseen characteristics of the samples used for the main analyses (Models 1–3). However, both the strength of the associations in the sensitivity analysis and the statistical evidence were affected by the reduction of the sample size, reducing precision and resulting in reduced power to detect genuine differences between the exposures and outcomes (see Tables S1–S3). Therefore, results of Model 4 (with the same N as the sensitivity analysis) should be interpreted cautiously, and this sensitivity analysis provides further statistical evidence to support findings from Models 1–3 reported in this study.

4. Discussion

This study provides information regarding the likelihood of developing CSCs by Age 3 based on patient‐specific anatomic factors. By exploring the influence of anatomic variables present within the first year of life and their impact on speech outcomes, in both univariable and multivariable models, findings from this study identify potential early anatomic determinants for developing CSCs in children with CP+/−L. Findings may improve clinical interventions and support the development of tailored management plans. Results from Model 3 revealed the strongest statistical evidence, and these findings are discussed below.

4.1. Influence of Cleft Type

Consistent with prior work that suggests an influence of cleft type on speech production, logistic regression analyses for the present study demonstrated robust associations between cleft type and the presence of at least one CSC. Notably, Model 3, the fully adjusted model with the larger sample size, demonstrated a 112% increase in the odds of having any CSC for children with UCLP diagnoses compared to CPO. Similarly, cases of BCLP exhibited a substantial 399% increase in odds. Taken together, children with CPO appear less likely to demonstrate CSCs than those with UCLP or BCLP, in general.

When looking more granularly at CSC outcomes and stratifying by CSC error type, cleft type independently demonstrated strong statistical evidence for posterior CSCs, with a substantially increased likelihood (7.27 times greater chance and 627% increased occurrence rate) to produce these errors when BCLP was present. Children with UCLP additionally demonstrated increased odds of producing posterior CSCs and a 270% increased occurrence rate compared to those with CPO. These findings are consistent with prior studies that focus on the severity of clefting, specifically cleft type, and align with results that clefts which affect the soft palate only demonstrate better speech outcomes compared to clefting which impacts both the hard and soft palate (Bicknell et al. 2002; Hardin‐Jones and Jones 2005; Klintö et al. 2018; Lohmander et al. 2002; Marrinan et al. 1998; Mason 2020).

Interestingly, there was little‐to‐no statistical evidence to suggest a greater prevalence of non‐oral CSCs across differing cleft types, meaning non‐oral CSCs were similar for children with CPO, UCLP and BCLP. Thus, raw data across cleft type demonstrated a similar occurrence rate for non‐oral speech errors. However, Model 3 showed increased odds, demonstrating a slightly greater likelihood for developing non‐oral and anterior CSCs for those with UCLP and BCLP compared to those with CPO, however, there was no statistical evidence to support this.

4.2. Influence of Pre‐Operative Cleft Width

When further assessing the impact and severity of clefting and its influence on speech outcomes, pre‐palatoplasty cleft width presented as a risk factor for passive and non‐oral CSCs at Age 3. For each millimetre increase in pre‐operative cleft width, the odds of demonstrating passive CSCs increased by 23%, and the odds of demonstrating non‐oral CSCs increased by 12%. However, no statistical evidence was found to suggest a difference in the likelihood to produce anterior or posterior CSCs.

Differences in the relationships between the maxillary anatomy, secondary to cleft width, provide possible explanations for these differing CSC outcomes. More severe forms of clefting (e.g., BCLP) are often associated with a wider palatal cleft width. This, in turn, has the potential to result in a greater likelihood for production of CSCs as well as a higher likelihood for development of post‐operative velopharyngeal insufficiency. Further, in complex cases involving wide palatal clefts, the surgical repair may fail to restore palatal anatomy, leading to the potential for the development of speech disorders secondary to post‐operative complications, such as fistula formation (Smyth and Wu 2019).

Similarly, Wu et al. (2017) documented increased cleft width as a predictor for future speech disorders. Those with a cleft width of 15.5 mm have been reported to be more likely to demonstrate hypernasality and speech disorders; however, details related to the specific type of speech issues were not assessed in previous work (Lam et al. 2012; Wu et al. 2017). Within the present study, cleft width was predictive, not only for CSCs in general, but specifically for passive and non‐oral CSCs. Further, a greater incidence of passive, non‐oral and posterior CSCs was documented within the study sample for individuals with cleft widths ranging from approximately 12–13 mm (±3.5 mm). Interestingly, no statistical relationship was found for anterior CSCs based on cleft width in this cohort. One possible explanation is that, within the current 3‐year‐old study cohort, dental or occlusal issues (e.g., Class III malocclusion) have not fully developed, which are commonly associated with anterior oral CSCs, such as dentalisation. These errors typically emerge or become more evident as occlusal status worsens over time. Consequently, future studies could explore whether older children, once their occlusal relationships are more established, also show a higher prevalence of anterior oral CSCs, taking into consideration dental/occlusal status as an additional anatomic predictor. This may yield an even more comprehensive understanding of how these speech errors develop and persist.

4.3. Influence of Fistula Status

Our results demonstrated a higher prevalence of CSCs, particularly passive and posterior CSCs, in individuals with a diagnosed fistula. Logistic regression analyses supported these findings, revealing strong associations between fistula diagnosis and the likelihood of developing CSCs. In Model 3, the odds of CSCs were markedly increased for those with a diagnosed fistula compared to those without. Specifically, a 204% greater likelihood was observed for the production of passive CSCs, such as nasalized, weak or absent pressure consonants. For those with a diagnosed fistula, there was a 67% greater likelihood for production of non‐oral CSCs, a 68% greater likelihood for production of anterior CSCs and a 122% greater likelihood for production of posterior CSCs. This variability observed in the likelihood for CSCs, particularly for active CSCs, is likely related to variability in fistula size and location. However, specifics related to fistula location were not documented in the available dataset, and the presence of fistula was determined based on visualisation by the SLT in the clinic.

Despite this, it is possible to assume that differences in the existence of CSCs may be tied not only to the presence of a fistula, but also its size and location (Karling et al. 1993). Further, it is likely that anterior or hard palate fistula often influences the speech production of phonemes with an alveolar or postalveolar place of articulation (Howard et al. 2019; Karling et al. 1993; Mason 2020; Parwaz et al. 2009). In contrast, posteriorly positioned fistulae are likely to have a greater influence on passive CSCs, such as absent, weak or nasalized pressure consonants (Gustafsson et al. 2023; Smyth and Wu 2019). Smyth and Wu (2019) additionally found the prognostic impact of palatal fistula to be highly significant for the development of VPD. Therefore, it is important to also consider the cascading impact of a diagnosed fistula on velopharyngeal function when weighing surgical management options and the impact on speech. This emphasizes the critical role of fistula diagnosis, in addition to understanding VPD status, in predicting and understanding CSCs in children with cleft conditions. However, we realize that fistula management is a complex process that requires the input of several members of the patient's care team, and speech is not the only consideration to take into account when considering a surgical repair.

4.4. Clinical Implications

Taken together, these findings highlight that CSCs in individuals with cleft palate are multifactorial and influenced by cleft type, cleft width and presence or absence of a fistula. These findings support the need for early assessment along with consideration of timely repair for symptomatic fistula, with statistical evidence that repair prior to Age 3 may be beneficial. This may facilitate a reduction in CSCs and help establish normal articulatory function for children with cleft conditions. However, any decision to repair a fistula must be preceded by multidisciplinary assessment with instrumental and imaging evaluation, as appropriate, given the potential co‐occurrence of posterior fistulae and VPD. When co‐occurring VPD is likely, coordinated (combined) surgery should be considered to minimise repeat anaesthesia and optimise outcomes.

Additionally, identifying these anatomic factors may support early identification and monitoring of children who are at a greater risk for developing CSCs. While children with cleft conditions are typically monitored yearly through cleft team visits, the presence of specific anatomic factors may encourage more frequent visits for speech monitoring. Early identification, beginning as young as 1 year of age, may also then allow for the development of tailored patient and family supports focused on preventing active CSCs and initiating speech therapy when needed.

These results emphasize the importance of considering the influence of cleft type in clinical speech assessments, but more importantly, using this knowledge to support accurate identification of CSCs when developing treatment plans. Knowledge that the frequency of non‐oral and anterior CSCs often occurs similarly across cleft types highlights this as a key focus area for articulatory and phonological interventions. For those with UCLP and BCLP, the higher likelihood to produce posterior errors, in addition to anterior and non‐oral errors, stresses the importance of both prevention of active speech errors and remediation, with a focus on improved place of articulation, when present. Furthermore, when passive and non‐oral errors are present, clinicians should also rule out the co‐occurrence of any underlying VPD to determine if the most appropriate treatment route is therapeutic and/or surgical (Chee‐Williams et al. 2025).

Incorporation of clinical imaging data and quantitative analyses in speech assessments, such as those provided by MRI, nasendoscopy or multiview videofluoroscopy, may also be beneficial prior to Age 3 (Henningsson and Isberg 1991; Kotlarek et al. 2023; Lam et al. 2006; Mason and Perry 2017). In a group of children with repaired cleft palate between 3 and 39 years of age, MRI revealed a shorter total and effective velum, as well as a higher incidence of levator veli palatini muscle discontinuity when velopharyngeal insufficiency was present (Sitzman et al. 2024). While current investigations of children under Age 3 are limited (Mason et al. 2025), future applications of MRI may further support the identification of additional anatomic factors related to the velopharyngeal musculature that may be predictive for the development of VPD, CSCs or both. Further research is needed to explore factors influencing CSCs and refine knowledge related to predictors that influence speech development for children with cleft conditions.

4.5. Strengths, Limitations and Future Directions

Strengths of the study include prospective, longitudinally collected data drawn from a large multicentre, national study. Collecting data from routine clinical care and from participant self‐report allows for the collection of a broad and extensive dataset, which can be mined to address a wide range of clinically relevant questions. The use of routine clinical data limits the potential for measurement of the reliability of some of the data collected, and this is a limitation. However, the benefit is the potential to collect data from a large and diverse population, where the burden to the participant is minimal. Cohort studies are powerful in providing data for preliminary analyses to explore questions of interest and to determine whether a future trial, which focuses typically on one primary research question, can be justified.

As is typical for cohort studies with multiple variables of interest, the fully adjusted models in our analyses resulted in fewer participants available in order to control potentially confounding variables such as sex, age and hearing status. Future studies should aim to identify larger populations, when available, to better generalize results. Detailed data on cleft classification were unavailable for the whole sample, resulting in the need to combine Veau I (cleft of the soft palate only) and Veau II (cleft of the hard and soft palate) into one category. Future research should explore whether differences in the presence of CSCs exist between a child born with a Veau I versus a Veau II cleft. Additionally, details related to the size and position of the fistula were absent for this dataset, resulting in a binary analysis (present vs. absent) for fistula diagnosis. Added details related to the size and location of the fistula may provide even greater predictive ability and facilitate the identification of more detailed patient profiles for those who are likely to develop CSCs. Knowledge of the impact of fistula location on subsequent CSCs could further influence tailored and timely management plans. It was not possible to include surgical timing within the analyses, as the inclusion of this variable reduced the sample considerably for some of the exposures. However, for those with available data (N = 199), the median age at palatoplasty was 10 months with an interquartile range between 8 and 11 months, indicating timely palatoplasty for a majority of the sample. Given this, palatoplasty timing is not likely to substantially influence findings for the present analyses. Surgical technique related to initial palatoplasty was only available for a small subset of the sample; however, of that sample, almost 90% reported using the same technique, intravelar veloplasty. Previous evidence has reported that 94% of palatoplasties within the United Kingdom were performed using an intravelar veloplasty (Fell et al. 2023), suggesting that there is little variation in the technique used across the United Kingdom. Surgical measurements within the Cleft Collective cohort have not been validated, and therefore, the rate of measurement error is unknown within this sample. It is worth noting that the type of palatoplasty and subsequent surgical outcomes have been shown to influence speech development and resonance outcomes, including hypernasality and nasal airflow disorders (Hofman et al. 2023; Van Lierde et al. 2004), and future studies should assess additional outcomes and predictive factors, if available. Additionally, data on velopharyngeal function, occlusion and positioning of teeth were not collected at the same timepoint as the other data used in this study, and therefore, these factors were not included in this analysis. Future studies incorporating data on velopharyngeal function, occlusion and dentition may be beneficial at earlier time points, specifically related to the influence of these additional anatomic factors on CSCs.

5. Conclusion

Findings from this study highlight the importance of early, individualized speech assessment and intervention, particularly for those with wider clefts and those with fistula. Both cleft width and fistula status contribute to CSCs. However, these factors alone may not fully account for the complexity of CSC aetiologies or their varied clinical implications. Notably, differences in both the type and frequency of CSCs across cleft types underscore the need for more targeted therapeutic and/or surgical approaches. These findings may improve clinical interventions and support tailored and timely management plans. Knowledge of differences in CSCs, as well as the anatomic predictors leading to their development, may aid in advancing future predictive models for early identification of children who may require additional supports to prevent the occurrence of these CSCs.

Funding

This study was funded by The Scar Free Foundation, the Underwood Trust, and the Vocational Training Charitable Trust (VTCT) (REC approval 13/SW/0064). Funding was also received from the National Center for Advancing Translational Sciences under award numbers UL1TR003015/KL2TR003016. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the named Foundations or the National Institutes of Health, the University of Bristol, University of Virginia or University of Wyoming.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supplementary Figure S1: A flow diagram illustrating the derivation of participants meeting inclusion criteria from the Cleft Collective.

Supplementary Table S1: Exploring the influence of Cleft Type on CSC Development using logistic regression. Supplementary Table S2: Exploring the influence of Cleft Width on CSC Development using logistic regression. Supplementary Table S3: Exploring the influence of Fistula on the Development of CSCs using logistic regression.

Supplementary Table S4: Confounders accounted for across differing models

Mason, K. , Kotlarek K., Davies A., and Wren Y.. 2025. “Beyond Prevalence: Understanding the Relationship Between Early Anatomic Factors and the Likelihood for Cleft Speech Characteristics.” International Journal of Language & Communication Disorders 60, no. 6: e70152. 10.1111/1460-6984.70152

Data Availability Statement

The datasets generated and/or analysed during the current study are available via a supported access resource through the Cleft Collective. The full data availability statement and data access policy for the Cleft Collective Resource can by located at https://www.bristol.ac.uk/cleft‐collective/professionals/access/.