Increased Risk of Velopharyngeal Insufficiency in Patients Undergoing Staged Palate Repair

Abstract

Objective:

To evaluate the association of two-stage cleft palate (CP) surgery on velopharyngeal insufficiency (VPI) incidence, speech surgeries, and cleft-related surgical burden.

Design:

Retrospective cohort with follow-up of 4 to 19 years.

Setting:

Academic, tertiary children’s hospital.

Patients:

Patients who underwent CP surgery between 2000-2017. Exclusions included submucous CP, or age at last contact under 3.9.

Interventions:

CP surgery, completed in either a single- versus two-stage repair.

Main Outcome Measure(s):

Rates of VPI diagnosis and speech surgery, total cleft surgeries; T-tests, tests of proportion, linear and logistic regression were performed. Total cleft-related surgeries were examined in a sub-set (n=418) of patients with chart reviews.

Results:

1,047 patients were included; 59.6% had two-stage CP repair, 40.4% had single-stage. Approximately 32% of children with two-stage CP repair were diagnosed with VPI, as opposed to 22% of one-stage patients (p<.001). Children with two-stage CP repair were 1.8 times as likely to be diagnosed with VPI (p<0.001). Speech surgery rates were similar across groups. Patients who had two-stage repair received an average of 2.3 more cleft-related procedures, when excluding prosthesis management procedures.

Conclusion:

Our data shows an increased risk of a VPI diagnosis and increased surgical burden among patients receiving two-stage CP repair.

Introduction:

Approaches to repair of cleft palate (CP) can vary; current approaches include both single-stage repair and two-stage repair, which can take various forms but includes delayed hard palate repair. The classic approach for a staged CP repair, the Schweckendiek procedure, closes the soft palate but delays hard palate closure to avoid the early subperiosteal dissection and associated scarring that could inhibit maxillary growth (Schweckendiek and Doz, 1978). The techniques commonly utilized during single-stage CP repair include a two-flap approach with intravelar veloplasty (Bardach style) and the Furlow palatoplasty (Katzel et al., 2009). A national movement toward single-stage CP repair followed the release of a 1982 study reporting poor articulation skills among children who had delayed hard palate repair (Noordhoff et al., 1982). This approach has gained favor and a 2005 survey evaluating international trends in cleft care demonstrated that 97% of cleft surgeons currently use single-stage CP repair (Weinfeld et al., 2005). Reports from 2009 noted that a majority (88%) of craniofacial surgeons in the United States perform single-stage CP repair, with many surgeons electing to perform surgery between 6 and 12 months of age (Katzel et al., 2009). These results demonstrate an overall shift in the practice patterns of CP surgeons.

Effects on maxillary growth, VPI diagnoses, the need for future surgical intervention, and concerns regarding total surgical burden (number of cleft related procedures) often determine surgeon preferences in approach. Those advocating for delayed hard palate repair believe this approach improves postoperative maxillary growth, avoiding early subperiosteal dissection and palatal scarring (Holland et al., 2007). This has led to some centers continuing to favor a two-stage approach.

Few studies have quantified the effects of two-stage CP repair on speech or surgical burden. Some prior research has shown that single-stage CP repair resulted in lower rates of fistulas (Landheer et al., 2010), improved speech scores, and decreased rates of velopharyngeal incompetency, as compared with two-stage repair (Holland et al., 2007; Stein et al., 2019). Past studies have often been limited by small sample sizes and the inability to address possible confounding factors. Researchers have also evaluated rates of pharyngoplasty after single-stage CP repair, with prior literature reporting a rate of approximately 25% (Bicknell et al., 2002). A recent large database study found 44% of included cleft patients had secondary CP surgery, but this rate ranged from 9-77% among the included hospitals; researchers did not assess cleft-related surgical burden or examine whether hospitals used two-stage protocols (Sitzman et al., 2019). There is a current lack of significant data comparing the rate of VPI or its surgical repair (speech surgery) among patients undergoing a single-stage versus two-stage CP repair, or examining the degree to which these approaches impact patients’ cleft-related surgical burden.

The gold standard for reporting VPI remains severity, with blinded perceptual speech assessment (Sell, 2005). Another way of understanding the clinical implications of VPI is through quality of life (QOL). While VPI severity from blinded assessment is important, studies have found that those with any severity of VPI have QOL that is significantly worse than either cleft palate patients without VPI or normal controls (Denadai et al., 2019). Additionally, previous work identifying the association of between VPI severity and QOL has shown a correlation, but has also shown that those with any severity of VPI can have severely impaired QoL (Skirko et al., 2013; Bhuskute et al., 2017). Because VPI of any severity can have an important impact on children’s lives, we elected to look at the incidence of VPI and incidence of speech surgery as clinically important outcomes.

Our research questions for this study were as follows: 1) Are there differences in the incidence of VPI between single-stage or two-stage CP repair patients; 2) Are there differences in the rate of speech surgery between groups, and 3) How does the CP repair type these patients receive contribute to their cleft-related surgical burden? The aim of our study was to evaluate the differences in the incidence of VPI, rate of speech surgery (including sphincter pharyngoplasty, pharyngeal flap, or revision z-plasty), and total cleft related surgical burden among a large cohort of patients who have undergone either single-stage or two-stage CP repair, as well as other predictors of these outcomes. The study hospital is uniquely situated to address these questions, as surgical providers at this center represent a mix of those utilizing single and two-stage CP repairs. This allowed for a natural retrospective experiment.

Methods:

Population

A retrospective cohort study was completed. Patients who had CP surgery and were followed by an academic pediatric hospital between 2000-2017 were examined. Patients with CP, with or without cleft lip (CL), were identified using ICD-9 and ICD-10 diagnostic and procedure codes (See Supplemental Table 1). Exclusions included loss to follow-up or current age below 3.9, to give sufficient time for the development and diagnosis of VPI, and those with a submucous CP. A total of 1,242 patients with CP repairs were identified in the electronic medical records (EMR) and were included. Of these, 85 were excluded due to age (<3.9), 102 did not have sufficient follow-up, and 8 had an isolated submucosal cleft. Demographic data was collected from EMR. Veau classification was imputed through diagnostic codes. Loss to follow-up was defined as lack of medical records within the Intermountain Medical System. This system is the largest healthcare system in the region and captures over half of all regional healthcare visits, annually (Intermountain Healthcare, 2018).

Care Protocol and Provider Details

Surgeons operating at the study hospital utilize several protocols for treating CP; some use a single surgery to repair the hard and soft palate, while others use delayed hard palate repair (two-stage CP repair). Approximately 60% of CP repairs at our institution are staged palate repair, which is largely attributed to surgeon preference. Patients receiving two-stage repair may have a soft palate adhesion, in which the mucosal surface of the soft palate is repaired without realignment of the levator muscles. A pin retained palatal prosthesis is affixed over the cleft of the hard palate at the time of the adhesion (if performed) or during CL repair. Many patients go on to have definitive soft palate repair with a double opposing z-plasty, prior to the time of hard palate repair. For our analysis, definitive palate closure in children with one-stage surgery was the date of hard and soft palate repair. For patients with two-stage surgery, it was the date of soft palate repair.

In terms of the surgical approach to soft palate repair, a majority of patients underwent a double opposing z-plasty repair. However, chart reviews revealed that approximately 30% underwent a straight-line repair. A majority of hard palate repairs are completed using bilateral mucoperiosteal flaps, with the Veau-Wardill-Kilner approach being the next most common approach.

During the study period, most palate repairs were performed by 5 surgeons; 4 plastic surgeons (80%) and one ENT (10%); the remaining 10% of surgeries were performed by 15 other surgeons. Plastic surgeons performed a majority of their repairs using two-stage approach (55%), while the ENTs favored one-stage repair (80%, p<0.001). Surgeons had caseloads ranging from an average of 1 repair a year to 23; most had between 1-4 a year, with 4 surgeons performing more than 10 repairs a year, on average.

Based on EMR data, at least 68% of the total sample and 86% of those with a diagnosis of VPI saw a hospital-based speech language pathologist (SLP) with craniofacial training and over 30 years of experience. The median age at first visit was 13 months, with an interquartile range (IQR) of 3-29 months. 90% of patients with evidence of an SLP visit had the first visit by age 5. Hospital SLPs use perceptual speech assessment, augmented by nasometry with MacKay-Kummer Simplified Nasometric Assessment Procedures Revised (SNAP-R) (MacKay, 1994; Kummer, 2008).

Outcomes & Metrics

Patients were diagnosed with VPI through perceptual speech analysis by a member of the craniofacial team. This assessment was augmented when possible with the use of the SNAP-R instrument (MacKay, 1994; Kummer, 2008) and/or nasal endoscopy. Because the ultimate goal of primary palate repair is normal resonance and speech (Kummer et al., 2012), we elected to analyze the incidence of VPI, rather than severity of VPI, as our primary outcome.

This incidence, as well as other outcomes and predictors, were captured in three ways, iteratively developed over the course of the project. These were text search of the EMR, ICD 9/10 and CPT code search of the EMR, and limited manual chart review. VPI diagnosis was identified using text search for the words “velopharyngeal,” “VPI,” and “hypernasality,” in multidisciplinary clinic notes, operative notes, ENT clinic notes, or plastic surgery clinic notes. Diagnostic codes for VPI and hypernasality were also searched (Supplemental Table 1). Chart review (n=459) revealed that the word “velopharyngeal,” used after the age of two in the above note types, gave the most accurate indication of a VPI diagnosis; this variable is referred to hereafter as VPI-text. Patients who received a pharyngeal flap or sphincter pharyngoplasty – or had these surgeries recommended in clinic visits – were also considered positive for VPI. Within this variable, 26% of VPI diagnoses were assigned through speech surgery receipt or recommendation; the remaining 74% were assigned through clinical notes from a member of the craniofacial team.

Speech surgeries were defined as sphincter pharyngoplasty, pharyngeal flap, or revision z-plasty. To identify these, text search for the words “sphincter,” “pharyngeal flap,” and “pharyngoplasty,” were run on operative notes, and medical records and billing data were searched for ICD and CPT codes for pharynx surgeries (see Supplemental Table 1). Text searches for the terms “z-plasty” and “Furlow,” with all possible spellings, were run on operative notes. Dates with z-plasty terms were compared with dates of CP procedures to exclude non-cleft z-plasties. All sphincter and pharynx procedures were considered speech, as were cleft-related z-plasties performed after the age of 3, or after date of hard palate closure. This flag is referred to as Surgery-text+codes.

To identify the definitive repair date necessary for the Surgery-text+codes variable, the first cleft-related Z-plasty between age 6 and 36 months was identified. For patients lacking this, the first palate procedure in that age range was used. The minimum age range was lowered to 3 months for patients still missing a procedure. For two-stage surgery patients, hard palate closure date was the final occurrence of the word “prosthesis” in an operative report, or the final date before the age of 6 years.

An additional set of outcome variables was constructed using only ICD and CPT codes, without the text search flags. This was undertaken with the goal of assessing methodology used in database studies without the option of text searches. To capture patients with VPI, codes for hypernasality and velopharyngeal insufficiency were used; this variable is called VPI-codes. All codes for sphincter and pharynx procedures were considered speech surgeries, as were codes for secondary CP repairs, if performed after 52 months old, or after hard palate closure. This age cut-off was developed through iterative chart review (n=85). This variable was called Surgery-codes. Patients receiving speech surgery were considered positive for VPI. After calculation of accuracy of all outcome flags via sensitivity and specificity against the gold standard of chart review, the VPI-text and Surgery-text+codes variables were found to be the most accurate and were used in all analyses. See Supplemental Table 1 for ICD and CPT diagnosis and procedure codes used.

Attention was then turned to creation of other variables. Text search for the word “prosthesis” in operative notes was used to differentiate two-stage from single-stage repairs, excluding those for whom the first use of the term occurred after age 2. Surgeon volume at each repair was assessed by constructing a count of the definitive repairs completed by each surgeon during the study period, as of the date of a patient’s definitive repair. Surgeon experience at each repair was evaluated by retrieving the date on which each surgeon finished fellowship and subtracting that from the date of repair. Veau classification was imputed through diagnostic codes (Supplementary Table 1).

Iterative chart review was used to assess and improve accuracy of coded variables in capturing outcomes and characteristics: 459 charts were reviewed for VPI diagnosis and 293 were reviewed for speech surgery. Charts review also fine-tuned variables capturing two-stage repair (n=156), definitive repair date (n=200), and Veau classification (n=418). Charts were reviewed using the EMR and compared with coded data; differences were used to refine coding until a high degree of accuracy was achieved.

In creating the surgical burden variable, it became clear that only CP surgeries were included in our data. To allow inclusion of cleft-related procedures not contained in our EMR search, a sub-analysis was conducted using only patients with full chart review data from the coding validation. The selection of these patients for chart review was done at random. The total cleft-related surgery variable was created as a count of unique days on which includable procedures took place (see Supplemental Table 2 for included procedures). Analyses were conducted both including and excluding dates when only a prosthesis insertion, exchange, or removal (hereafter called prosthesis management) occurred.

Statistical Analysis

Flags for staged repair, VPI, and speech surgery – both the set constructed with codes and text searches (VPI-text and Surgery-text+codes), and the set using only coding (VPI-codes and surgery-codes) – were analyzed for sensitivity and specificity against the gold standard of manual chart review. Using sensitivity and specificity, receiver operator curves were created and the area under the curve (AUC) was calculated for each variable, to understand which functioned best in our cohort. Descriptive analysis was performed comparing demographic characteristics of children who had single-stage versus two-stage CP repair using chi-square tests for categorical variables, and Wilcoxon rank-sum tests for non-parametrically distributed continuous variables. After assessing for normality, t-tests were used to examine differences in age at VPI diagnosis and speech surgery: differences in rates of VPI diagnosis and speech surgery were assessed through tests of proportion.

Multivariate linear and logistic regression models included a priori confounders of gender, race (dichotomously represented), use of public versus private insurance, surgeon volume, and Veau classification (Kosowski et al., 2012; Sitzman et al., 2019). Each Veau classification (1-4) was included as an independent variable. Logistic regressions also adjusted for year of patient birth. Variables not significant at the 0.05 level were removed in a step-wise fashion until only significant variables remained in multivariate analyses.

The number of cleft-related procedures performed was evaluated as an outcome, with staged surgery included as a predictor variable. This included CL and CP repair, cleft lip and palate (CLP) revision, bone grafting, rhinoplasty, LeFort, as well as myringotomy, tonsillectomy, and tympanoplasty (see Supplemental Table 2 for complete list). For all analyses, STATA 14.2 (College Station, TX) was used. Statistical significance was designated at alpha = 0.05. This study was approved by the University of Utah and Intermountain Healthcare Institutional Review Board. All data was stored on a secured server.

Results:

A total of 1,047 patients were included in the study and patients were followed up to age 19. The current median age of participants is 12.4, with an IQR of 9-16. 59.6% of included patients were two-stage and 40.4% were single-stage CP repair. Single-stage and two-stage patients had similar distributions in terms of race and insurance status (p=0.47 and 0.45), but two-stage patients were younger and more likely to be male (Table 1). A higher Veau classification was seen among those receiving two-stage CP repair; 87% of two-stage CP repair patients were Class 3 or 4, as compared to 24% of single-stage patients. Patients in the two groups were also different in terms of their surgical treatment; two-stage patients had surgeons who had performed a greater volume of cleft repairs during the study period (p<0.001). Additionally, two-stage patients received their first CP surgery younger than single-stage patients, and were older at time of both definitive repair and hard palate closure (p<0.001). All statistically different characteristics had a significance of <0.001. Surgeons had a median of 14 years of post-fellowship experience at time of repair (IQR 7.8-21 years); this was similar between surgeons performing one-stage and two-stage repairs (p=0.249).

Table 1:

Demographic Characteristics of Study Population

	Single-Stage (n=423)	Two-Stage (n=624)	Test Statistic	p
Veau Classification n (%)a			chi2=436.67	<0.001
1	46 (11%)	2 (0.3%)
2	262 (65%)	81 (13%)
3	73 (16%)	293 (47%)
4	31 (8%)	247 (40%)
Race n (%)a			chi2=0.52	0.47
Non-White	25 (7%)	32 (5%)
White	362 (93%)	548 (95%)
Gender n (%)a			chi2=23.08	<0.001
Female	215 (51%)	224 (36%)
Male	208 (49%)	400 (64%)
Insurance n (%)a			chi2=0.02	0.45
Private	290 (69%)	414 (66%)
Public	133 (31%)	210 (34%)
Median Current Age in Years (IQR)b	13.23 (9.6, 16.5)	11.8 (8.5, 15.3)	z=3.77	<0.001
Median Follow-Up in Years (IQR)b	11.3 (7.7, 14)	10.1 (7.1, 13.6)	z=2.14	0.032
Median Surgeon Volumec (IQR)b	86 (27, 186)	129.5 (57, 244.5)	z=−7.76	<0.001
Median Years of Surgeon Experience (IQR)b	17.2 (7.4, 21.9)	17.4 (8.0, 22.8)	z=−1.15	0.249
Median Age in Months at First CP Surgery (IQR)b	9.7 (8, 11.3)	3.2 (2.2, 4.64)	z=21.87	<0.001
Median Age in Months at Definitive CP Closure (SD)b	9.8 (8.1, 11.4)	11.7 (9.8, 17.4)	z=−11.54	<0.001
Median Age in months at Hard Palate Closure (IQR)b	9.8 (8.1, 11.4)	36.9 (20.9, 48.2)	z=−21.92	<0.001
Median Total Cleft-Related Surgeriesd (IQR)b	2 (1,4)	6 (5, 8)	z=−20.37	<0.001
Median Total Cleft-Related Surgeriesd (IQR), excluding prosthesis-onlyb	2 (1,4)	5 (4, 8)	z=−17.93	<0.001

Abbreviations: CP,Cleft Palate; SD, Standard Deviation; IQR, Interquartile Range

^aCompared with Pearson’s chi²

^bCompared with Wilcoxon rank-sum

^cVolume of definitive repairs performed during study period on day of definitive repair

^dRepresents a sub-set (418) of total study population

Patients with two-stage CP repair had a higher rate of VPI diagnosis (32%) than patients who underwent single-stage CP repair (22%, p<0.001). After adjusting for significant confounders, children with two-stage CP repair were 1.8 times as likely to be diagnosed with VPI (p<0.001). Of potential covariates when assessing the association with VPI and two-staged surgery, only surgeon volume was significantly associated with VPI (Table 2). VPI incidence varied by Veau classification: 15% of those with Veau 1 were diagnosed with VPI; 24% of those with Veau 2, 28% of Veau 3, and 35% of those with Veau 4 received this diagnosis (p=0.003). In spite of this, Veau classification did not retain significance in multivariate regression. Age of VPI diagnosis was similar for patients receiving single stage (Mean 5.5 years, SD 2.9) and two-staged surgery (Mean 5.6, SD 2.8, p=0.86), in both univariate and multivariate analysis (data not shown). Age at diagnosis and age at surgery were normally distributed.

Table 2:

Regression Analysis Results

Outcome: VPI Diagnosis – Logistic Regression
Exposure	Univariate Analysis			Multivariate Analysis a
	OR	95% CI	p-value	OR	95% CI	p-value
Two-Stage Repair	1.67	(1.3,2.2)	<0.001	1.80	(1.4,2.4)	<0.001
Surgeon Volumeb				0.98	(0.97,.99)	0.006
Outcome: Speech Surgery – Logistic Regression
Two-Stage Repair	1.16	(0.8,1.6)	0.38	1.22	(0.9, 1.7)	0.23
Surgeon Volumeb				0.98	(0.96,0.99)	0.001
Outcome: Total Palate Proceduresd –Linear Regression
	Number of Proceduresd	95% CI	p-value	Number of Procedures	95% CI	p-value
Two-Stage Repair	3.45	(3.3,3.6)	<0.001	2.30	(1.7,3.0)	<0.001
Surgeon Volumeb				−0.03	(−0.06,−0.01)	0.004
Veau
I				−2.47	(−4.0,−0.9)	0.002
II				−2.13	(−2.9,−1.4)	0.002
III				−1.03	(−1.6,0.4)	<0.001
IV				--	--	--

Abbreviations: OR – Odds Ratio; CI – Confidence Interval

^aLogistic Regression – All multivariate models began by adjusting for Veau classification, surgeon volume, race, gender, and insurance status. Logistic regressions also adjusted for year of birth. Significant co-variates were retained in the final multivariate analysis.

^bPer 10 repairs done.

^cLinear Regression—Multivariate models adjusted as above, but excluding year of birth.

^dExcludes surgical days on which only prosthesis management occurred

Rates of speech surgery did not differ significantly between those with single stage surgery (16%) and two-stage surgery (18%, p=0.38). After adjusting for confounders in multivariate regression, two-stage surgery was not associated with an increased risk of VPI surgery (OR 1.2, p=0.23, See Table 2). Those with two-staged surgery had an age at speech surgery that was younger by 0.99 years, but this did not reach statistical significance (p=0.06, data not shown).

Analyses of the number of cleft-related surgeries (surgical burden) used data from chart reviews (n=418). Single-stage patients had fewer total cleft-related surgeries (median 2, IQR 1,4) than two-stage patients (median 6, IQR 5,8, p<0.001). Cohen’s d for the effect size was −1.44 (95% CI: −1.58, −1.30). When brief procedures for prosthesis management were excluded from the analysis, two-stage patients still experienced 3.5 more procedures (p<0.001); Cohens d was −1.19 (95% CI: −1.32, −1.05). These associations are considered very large (Cohen, 1992). This association continued when patients were stratified by Veau: excluding prosthesis management procedures, single-stage patients with Veau 4 had a median of 4 surgeries (IQR 3,5), while two-stage patients with similar clefts had a median of 6 (IQR 5,8; see Figure 1). In multivariate regression, Veau classification and surgeon volume were significantly associated with total surgeries, though the effect size of surgeon volume was small (Cohen’s d=−0.49). When controlling for these variables, two-stage patients had 3.1 more total cleft-related surgeries than single-stage patients (p<0.001). This effect dropped to 2.3 additional surgeries when prosthesis management procedures were excluded (p<0.001; see Table 2).

Figure 1:

Total cleft-related surgical days by surgical type and Veau classification. Excludes prosthesis management procedures

The reliability of our exposure and outcome ascertainment variables was found to be quite high. Chart reviews (n=156) for surgical type revealed that the staged surgery flag had a sensitivity of 98% and specificity of 93%. The VPI-text flag (459 charts reviewed) had a sensitivity of 82% and specificity of 90%, with an AUC of 0.86. The Surgery-text+codes flag was compared to 293 chart reviews and found to be 96% sensitive and 95% specific (AUC .96; See Figure 2). The outcome variables constructed using only coding differed slightly: the VPI-codes flag had a sensitivity of 64% and specificity of 90% (AUC 0.77); the Surgery-codes flag was 92% sensitive and 88.5% specific (AUC 0.90). 200 chart reviews specific to date of staged hard palate closure and definitive repair showed that variables designed to capture these succeeded to within 2 months in 90.1% and 90.8% of cases, respectively. The Veau classification variable, derived through ICD diagnosis coding, was found to be quite accurate when compared with 418 chart reviews. When examined as CL/CP (class 3 and 4) versus isolated CP (class 1 and 2), the sensitivity was 99% and the specificity was 87%.

Figure 2: Receiver Operator Curves (ROC) with plot of specificity versus sensitivity for outcome VPI-text and Surgery-text+codes flags.

Abbreviations: VPI – Velopharyngeal Insufficiency

Discussion:

This study of a large cohort of CP patients found an increased risk of VPI diagnosis after two-stage CP repair. This is the first study, to our knowledge, which quantifies the degree of risk of VPI among patients undergoing more than two-staged CP repair. It also examined several other questions relating to two-stage repair, including rates of speech surgery in both surgical groups, age at diagnosis, age at speech surgery, and the impact of two-stage repair on surgical burden. The sample size of over 1000 patients and long follow-up of up to 19 years allowed investigation of these questions.

In spite of the nearly double (OR 1.8) VPI risk identified by this study in two-stage patients, no difference in speech surgery rates was seen between the groups. Targeted chart reviews suggested that many patients with a diagnosis of VPI had their speech surgery delayed due to development of malocclusion and a plan of a Le Fort advancement prior to VPI surgery. This may help explain the seeming contradiction in findings between VPI diagnosis and speech surgery. However, these patients may be at risk of surgical burnout given the number of procedures they undergo during childhood.

Surgeons in this study favored a two-stage repair protocol over a single-stage, particularly for patients with higher Veau classification. Differences in surgeon preferences regarding CP repair often reflect relative weight given to maxillary facial growth and possible VPI risk. Those who perform a two-stage CP repair may rely on evidence that maxillary growth is inhibited by early hard palatal surgery. Past evidence on this question has been mixed. Comparing 48 adult CL/CP patients who had received only lip repair with 58 who had lip and palate repaired in childhood, one study found significant differences in several cephalometric measurements (Liao and Mars, 2005). Friede and Enemark conducted a comparison of cephalometric measurements between patients at two Scandinavian centers; both groups had separate hard and soft palate repairs, but one group was significantly older at hard palate repair (Friede and Enemark, 2001). In that study, researchers found more generally convex facial development in patients treated with the delayed hard palate repair. A final example of the negative impact early hard palate manipulation may have on maxillofacial growth is provided by Liao et al, who compared 31 patients who had single-stage surgery at approximately 12 months old, with 41 whose hard palate closure was delayed until around age 6 (Liao et al., 2010). Researchers reported that two-stage patients had significantly longer maxilla, as measured at age 20. This evidence is not uncontested, with a number of studies finding little or no impact of single-stage surgery on maxillary growth (Stein et al., 2007; Bardach et al., 1984). Some have even shown improved maxillary outcomes with single-stage repair (Holland et al., 2007). Ultimately, choosing to perform two-stage palate repair is often related to surgeon preference and goal of preserving maxillary growth.

Our study found a strong association between two-stage repair and VPI. This is consistent with previous research, which has identified an association with two-stage CP repair and VPI diagnoses (Holland et al., 2007; Bardach et al., 1984). A recent systematic review and meta-analysis evaluating outcomes related to CP repair in 8 studies (combined n=840) reported 45% less risk of VPI among patients undergoing single-stage CP repair compared to two-stage CP repair, p=0.03 (Stein et al., 2019). The previously-discussed study by Liao et al., while finding restricted maxillary growth in single-stage patients, also reported a velopharyngeal function level of “adequate” in 48% of single-stage patients, as compared to 22% of two-stage (Liao et al., 2010). Several small studies reported increased levels of hypernasality and nasal emission among two-stage CP patients (Pradel et al., 2009; Kappen et al., 2017). One reported the two-stage group experiencing a pharyngoplasty rate of 42% (Kappen et al., 2017). Finally, a randomized trial (n=448) comparing four CP repair protocols, including two different timings of delayed hard palate repair, found no statically significant difference among study arms when they evaluated both hypernasality and velopharyngeal competence at age 5 (Lohmander et al., 2017). Overall, there is significant evidence that patients undergoing two-stage palate repair may have higher risk for VPI, which may then require further surgeries to address.

In addition to unclear benefits for maxillary growth and a possible increase in risk of VPI, two-stage protocols may increase the inherent risks accompanying any surgical interventions. Repeated exposure to general anesthetic by young children has caused concern among both researchers and regulatory agencies due to possible neurotoxicities (Andropoulos and Greene, 2017; Vutskits and Davidson, 2017). Additionally, the disruptions of child development and family life represented by repeated surgeries, hospitalizations, and recoveries, to say nothing of the economic impact, should play a role in a surgeon’s decisions on the best approach for CP repair (Burke and Alverson, 2010; Thompson et al., 2017). Our study found that two-stage patients had 3.4 additional cleft-related procedures; when the brief surgeries associated with prosthesis management were excluded, these patients still received 2.3 more procedures than single-stage patients. This shows an increased surgical burden in two-stage patients with a large effect size.

This study provides valuable information regarding the risks of speech impairment and increased surgical burden among patients undergoing two-stage CP repair, but has several limitations. Among these is that, while the retrospective design allowed us to obtain a large study population, it relied on accurate coding. Though our team evaluated a large number of patient charts to ensure correct exposure and outcome ascertainment, there is potential for coding errors that could ultimately affect data outcomes. The high sensitivity and specificity of our diagnosis flags were reassuring. A coding issue not detectable through chart review is that of missing notes. Private practice notes were not available for search or review; these patients’ outcomes would only appear in our analysis through any surgeries they received. It is possible this introduced bias. An important limitation is this study’s inability to differentiate severity of VPI cases based on our methods. The gold standard for VPI evaluation is a blinded assessment by several trained evaluators, with inter- and intra-judge reliability reported (Sell, 2005). While this assessment protocol was not possible in a retrospective study, we recognize its desirability; future studies examining differences in severity would be helpful. Because it has been shown that any degree of VPI has important QOL implications (Skirko et al., 2013; Bhuskute et al., 2017), we felt investigating difference in incidence of VPI diagnosis among those with one and two-stage repair was important. A final important limitation of our study is that more severe CP cases may be more likely to receive two-stage CP repair based on surgeon preference; these patients may also have an increased risk of VPI. We did see a larger proportion of patients with Veau classification 3 and 4 undergoing two-stage CP repair, and so included Veau classification in multivariate analyses. It was not a significant predictor of VPI when staged surgery was also included. Future efforts should focus on how evolving CP care can continue to minimize adverse effects on both maxillary growth and VPI diagnoses. Ultimately, this study defines the risk of VPI and subsequent surgical burden among CP patients undergoing two-stage CP repair.

Conclusion:

Our study demonstrates that two-stage CP repair is associated with an increased risk of VPI diagnosis and increased surgical burden. A single-stage approach for CP may help decrease rate of VPI but should be balanced with other potential benefits of staged palate repair. This recommendation should be weighed against the lack of difference seen in the rates of speech surgery among this cohort of patients.