Assessment of PresuPredicted Surgical Outcome

and Without Naso

Marcie S. Rubin, DrPH, MPH,* Sean CloustonKristen M. Lowe, DDS, MS,* Pradip R. Shetye,

Stephen M. Warren, MD,* and

Abstract: Obtaining an esthetic and functional primary surgical

before primary repair, (2) having better anticipated surgical out-

ORIGINAL ARTICLEcomes, and (3) being less likely to require early revision surgery.

MATERIALS AND METHODSAfter obtaining institutional review board approval (IRB#13-

676), an initial invitation and up to 2 follow-up invitations to non-respondents were sent to all of the plastic surgeons and oral surgeons(hereafter cleft surgeons) on the American Cleft Palate-Craniofacial

Accepted for publication July 23, 2014.Address correspondence and reprint requests to Barry H. Grayson, DDS,

Institute of Reconstructive Plastic Surgery, New York UniversityMedical Center, 307 E 33rd Street, New York, NY 10016;E-mail: barry.grayson@nyumc.org

Presented at the 12th International Congress on Cleft Lip/Palate and RelatedCraniofacial Anomalies; Lake Buena Vista, FL; May 9, 2013.

Supported by NIDCR-R21 DE021853 to Broder (PI).The authors report no conflicts of interest.Received January 23, 2014.cal outcomes, including the need for secondary cheilorhinoplasty.If NAM therapy influences esthetic and surgical outcomes, thenthe following hypotheses should be upheld: compared with patientswithout NAM preparation, surgeons will, on average, rate patientswho received NAM preparation as (1) having less severe clefts

From the *Department of Plastic Surgery, New York University LangoneMedical Center, New York, NY; Program in Public Health and Depart-ment of Preventive Medicine, Stony Brook University, Stony Brook, NY;Cariology and Comprehensive Care, New York University College ofDentistry, New York, NY.repair in patients with complete cleft lip and palate (CLP) canbe challenging because of tissue deficiencies and alveolar ridge dis-placement. This study aimed to describe surgeons assessments ofpresurgical deformity and predicted surgical outcomes in patientswith complete unilateral and bilateral CLP (UCLP and BCLP, re-spectively) treated with and without nasoalveolar molding (NAM).Cleft surgeon members of the American Cleft Palate-CraniofacialAssociation completed online surveys to evaluate 20 presurgicalphotograph sets (frontal and basal views) of patients with UCLP(n = 10) and BCLP (n = 10) for severity of cleft deformity, qualityof predicted surgical outcome, and likelihood of early surgicalrevision. Five patients in each group (UCLP and BCLP) receivedNAM, and 5 patients did not receive NAM. Surgeons were maskedto patient group. Twenty-four percent (176/731) of surgeons withvalid e-mail addresses responded to the survey. For patients withUCLP, surgeons reported that, for NAM-prepared patients, 53.3%had minimum severity clefts, 58.9% were anticipated to be amongtheir best surgical outcomes, and 82.9% were unlikely to need revi-sion surgery. For patients with BCLP, these percentages were 29.8%,38.6%, and 59.9%, respectively. Comparing NAM-prepared withnonNAM-prepared patients showed statistically significant differ-ences (P < 0.001), favoring NAM-prepared patients. This studysuggests that cleft surgeons assess NAM-prepared patients as morelikely to have less severe clefts, to be among the best of their surgicaloutcomes, and to be less likely to need revision surgery when com-pared with patients not prepared with NAM.Copyright 2014 by Mutaz B. Habal, MDISSN: 1049-2275DOI: 10.1097/SCS.0000000000001233

The Journal of Craniofacial Surgery Volume 26, Number 1, Janu

Copyright 2014 Mutaz B. Habal, MD. Unauthorgical Clefts andin Patients Treated With

alveolar Molding

, PhD, Mohammad M. Ahmed, DDS, MS,*DDS, MDS,* Hillary L. Broder, PhD, MEd,Barry H. Grayson, DDS*

Key Words: Nasoalveolar molding, cleft lip and palate

(J Craniofac Surg 2015;26: 7175)

The tissue deficiencies and alveolar displacement characteris-tic of complete cleft lip and palate can make achieving an es-thetic and functional primary repair challenging for the surgeon.1

Patients with wider and more severe clefts generally have more pro-nounced nasolabial stigmata that tend to worsen with growth.2 Ifthe residual deformity is severe, early secondary cheilorhinoplastymay be indicated to improve nasolabial esthetics, function, and psy-chosocial development. Salyer et al3 and Henkel et al4 report theneed for early revisions in as many as 35% of their patients. How-ever, the incidence of secondary deformities varies by cleft typeand is most likely in patients with complete bilateral cleft lip andpalate (BCLP).

Nasoalveolar molding (NAM) is a technique developed byGrayson et al5 intended to improve long-term outcomes in patientswith cleft lip and palate.6,7 The goal of NAM is to reduce thecleft defect and mold it into a more mild form to improve primarysurgical outcomes. This therapy uses a removable appliance care-fully adjusted to gradually restore normal form and symmetry tothe nasolabial region, precisely approximate the alveolar segments,increase patency of the cleft-sided nostril(s), and nonsurgicallyelongate the columella.6 Benefits of NAM include improved long-term nasal symmetry, reduction in the number of nasal surgical re-visions, and reduced need for secondary alveolar bone graftingwhen gingivoperiosteoplasty is performed.813 As of 2012, 37%of cleft centers in the United States reported offering NAM.14

The aim of this rater-masked quasiexperimental study wasto compare cleft surgeons assessments of NAM-prepared andnonNAM-prepared patients with complete cleft lip and palatein the presurgical severity of cleft deformities and anticipated surgi-Association's member e-mail list asking them to review and answerquestions based on preoperative photographs of 20 patients (here-after patients). Between April 11 and May 5, 2013, we received

ary 2015 71

rized reproduction of this article is prohibited.

responses from 176 of the 731 surgeons with valid e-mail addresses(24% response rate).

The surgeons were asked to (1) rate the severity of the cleftdeformity in this patient compared with other patients they hadseen before primary surgical repair as minimum, moderate, or max-imum severity; (2) take into consideration the width of the cleft, dis-placement of premaxilla, position of lip segments, asymmetry ofalar bases, nasal cartilage form, amount of columella, and the shapeof the nose and indicate whether they expected the patient wouldbe among the best, average, or least good of their usual surgicaloutcomes; and (3) indicate, based upon the expected outcome ofthe primary surgical repair and degree of severity that present pre-surgically, whether the patient would be likely or not likely to needrevision surgery before entering school (age, 56 y).

All surgeons received the patients preoperative photographswith standardized pictures of 2 views: frontal and basal (Figs. 1, 2).Presentation was consistent for all surgeons. Of the 20 patients pre-sented in the survey, 10 had received presurgical preparation withNAM (all performed by the senior author, B.H.G.), and 10 hadnot; respondents were masked to the NAM treatment status. Withineach treatment group, 10 of the patients had complete unilateralcleft lip and palates (UCLPs), and 10 had complete BCLPs.

Responses on the first 2 questions were dichotomized (min-imum [1] versus moderate-to-maximum severity clefts [0] and bestsurgical outcome [1] versus average-to-least good surgical out-

ing that the new model is the better fitting model. Beta coefficientsand 95% confidence intervals (95% CIs) are provided. For exposi-tory purposes, these models are further used to estimate the meanpercent of surgeons that describe NAM-prepared and nonNAM-prepared patients as having a minimum severity cleft, having the bestanticipated surgical outcome, and being unlikely to need revision;these percentages and their corresponding 95% CIs are presented inthe figures.

RESULTS

Descriptive AnalysesThe survey provided information on surgeons ratings of pa-

tients' cleft severity, best anticipated surgical outcome, and whetherearly revision surgery was likely to be necessary (Table 1). Of the20 patient photograph sets, by design, 10 (50%) were patients withBCLP. Masked to the surgeons, 50% of the photographs were ofNAM-treated patients. Considering the entire sample, surgeonsrated 21.3% of patients as having minimal severity clefts, 26.6%as having the best anticipated surgical outcome, and 53.0% asunlikely to require early revision surgery. Within both the UCLPand BCLP groups, crude differences suggest that NAM-preparedpatients were more positively rated compared with nonNAM-pre-pared patients on each of the 3 outcomes of interest. For example,among patients with UCLP, whereas more than half of thoseNAM-prepared were considered to have a minimal severity cleft,

Rubin et al The Journal of Craniofacial Surgery Volume 26, Number 1, January 2015come [0]). The responses to the survey resulted in 2774 observationsincluding 1496 observations on patients with BCLP and 1278 onthose with UCLP.

To provide summary statistics and compare treatment groups,the likelihood that a particular surgeon would positively rate a NAM-prepared patient (eg, as having a minimum severity cleft) was cal-culated and compared with the likelihood that a nonNAM-preparedpatient was described in the same way. Cross-tabulations of thesedata are presented, which provide bivariate results. As a starting esti-mate, crude differences between surgeons ratings of patients withand without NAM are also provided. However, these crude estimatesmay be biased by repetition in the data: each surgeon rated morethan 1 patient, and more than 1 surgeon rated each patient. Such

FIGURE 1. Representative sample of UCLP patient photographs.NAM-prepared patient (top). NonNAM-prepared patient (bottom).

Of note, respondents were masked to the NAM treatment status.

72

Copyright 2014 Mutaz B. Habal, MD. Unauthorrepetition can lead to substantial biases and poor model fit, likely in-creasing the chance of type I error.15 Thus, a cross-classified multi-level logistic modeling (CCMLM) approach was used to ensurerobust estimates.16,17 The CCMLM approaches are particularly usefulbecause they provide estimates that are robust to data that are miss-ing at random, for instance, due to surgeons' attrition because of timeconstraints. The importance of using multilevel logistic over logisticregression was tested to ensure that the CCMLM models providedthe best fittingmodels. To compare types of statistical models, AkaikeInformation Criteria (AIC) and the relative likelihood (RL) wereused.18 The RL estimates the probability that a new model improveson the older model; an RL of less than 0.05 is interpreted as suggest-

FIGURE 2. Representative sample of BCLP patient photographs.NAM-prepared patient (top). NonNAM-prepared patient (bottom).Of note, respondents were masked to the NAM treatment status.have the best anticipated surgical outcome, and be unlikely to need

2014 Mutaz B. Habal, MD

ized reproduction of this article is prohibited.

revision, fewer than a third of those without NAM preparationwere rated as such. Among patients with BCLP, results were similaras patients with NAM preparation who were more favorably ratedthan patients without NAM preparation.

being among the best anticipated surgical outcomes (B = 3.84,P < 0.001), and not needing early revision surgery (B = 4.68,P < 0.001) compared with the nonNAM-prepared patients.

As depicted in Figure 3, which illustrates estimated popula-tion means, more than half of the respondents reported that NAM-

TABLE 1. Sample Characteristics and Bivariate Analyses Examining the Relationship Between NAM Preparation Status and Ratings of Cleft Severity andExpected Outcomes

Outcome Minimum Severity Clefts Best Anticipated Surgical Outcome Unlikely to Need Revision

Overall, % 21.3 26.6 53.0

Cleft type Unilateral Bilateral Unilateral Bilateral Unilateral Bilateral

% by cleft type 27.8 15.7 33.7 20.6 64.6 43

NAM status No NAM NAM No NAM NAM No NAM NAM No NAM NAM No NAM NAM No NAM NAM

% by cleft type and NAM status 2.5 53.1 1.7 29.8 8.8 58.6 2.7 38.6 26.4 59.9 46.8 82.6

Difference between patientswith and without NAM

50.6 *** 28.1 *** 49.8 *** 36 *** 33.5 *** 35.9 ***

***P < 0.001.

are

ulti

The Journal of Craniofacial Surgery Volume 26, Number 1, January 2015 Surgeon Survey on NAM Cleft TreatmentSupport for Model SelectionTo address potential repetition bias by the respondents or by

repeated analysis of the same patients' photographs, we assesseddifferent model assumptions using tests of model fit (Table 2).These results show that logistic regression provides a good modelfitthe AIC is significant, and the coefficients (not shown: B =3.2 and 3.8 for UCLP and BCLP, respectively) are also highly sig-nificant. Incorporating the first level of model clustering, aroundthe surgeons' responses, provides a better fitting model: the AICis smaller, and the RL is less than 0.05. Furthermore, using cross-classified models provides an even better model fit: the AIC issmaller again, and the RL, comparing cross-classified models tomultilevel models, shows improved fit as well (RL < 0.001). Thisanalysis suggests that using the cross-classified models providesthe best-fitting results; however, in this case, the conclusions de-rived from each of these models would be similar.

Unilateral Cleft Lip and PalateAssociations between NAM treatment among patients with

UCLP and the likelihood of surgeons positively rating patients onthe variables of interest are provided (Table 3). Significantly differentresponses from surgeons regarding NAM-prepared and nonNAM-prepared patients were observed; NAM-prepared patients were morelikely to be rated as having minimal severity clefts (B = 5.45, P < 0.001),

TABLE 2. Assessing Model Fit Using the AIC and the RL That Progressive ModelsStatus to Surgical Outcomes

Logistic Regression MEst. P

UCLP Minimum severity cleft AIC 1034

P < 0.001). Similarly, whereas 38.6% of the surgeons anticipatedthat NAM-prepared patients would be among the best of theirusual surgical outcomes, 2.5% suggested as such for nonNAM-prepared patients (38.6% 2.5%, P < 0.001). Finally, whereas59.9% of the surgeons rated NAM-prepared patients as unlikely toneed revision surgery, 26.2% anticipated the same for nonNAM-prepared patients (59.9 26.2, P < 0.001).

DISCUSSIONThe objective of this study was to test the impact of NAM

Estimating the correct results can be difficult with repeat-ing measurement: individuals often respond to questions based ontheir own degree of optimism and experience. Although a numberof potential models were examined, CCMLM was chosen becauseit provided the best-fitting and most robust results. In this study,whereas we are most confident in the cross-classified results be-cause the model accounts for sample characteristics, we are partic-ularly confident in our findings because all models showed thatNAM treatment significantly improves the surgeons ratings of pa-tients for the outcomes measured. Surgeons rated NAM-preparedUCLP as minimal severity 53.3% of the time, whereas only rat-

TABLE 3. CCMLM Examining the Relationship Between NAM PreparationStatus and Ratings of Cleft Severity and Expected Outcomes AmongPatients With UCLP

MinimumSeverity Cleft

Best AnticipatedSurgical Outcome

Unlikely toNeed Revision

B 95% CI B 95% CI B 95% CI

NAM 5.45 4.46, 6.44 3.84 3.14, 4.54 4.68 3.65, 5.71

Random Intercepts, SD

Surgeons 0.50 0.28, 0.92 0.41 0.22, 0.76 0.59 0.33, 1.07

Patients 2.01 1.61, 2.5 1.69 1.37, 2.08 4.34 3.52, 5.34

AIC 875.5 1089.6 879.5

TABLE 4. CCMLM Examining the Relationship Between NAM Preparation Statusand Ratings of Cleft Severity and Expected Outcomes Among Patients With BCLP

MinimumSeverity Cleft

Best AnticipatedSurgical Outcome

Unlikely toNeed Revision

B 95% CI B 95% CI B 95% CI

NAM 5.00 2.81, 7.19 4.62 2.94, 6.30 3.47 2.11, 4.83

Random intercepts, SD

Surgeons 1.50 0.88, 2.58 1.19 0.71, 2.00 1.03 0.63, 1.66

Patients 1.85 1.45, 2.36 1.66 1.32, 2.09 3.31 2.76, 3.96

AIC 807.8 972.5 1257.2

Rubin et al The Journal of Craniofacial Surgery Volume 26, Number 1, January 2015preparation using a rater-masked quasiexperimental design. Surgeonswere asked to rate patients presurgical cleft severity and anticipatedsurgical outcomes. Using cross-classified multilevel logistic regres-sion to account for repeat-testing bias, the results from the surgeonsratings reveal a consistent pattern: NAM-prepared patients as havingminimal severity clefts, having better anticipated surgical outcomes,and being less likely to need early revision surgery compared withthose without NAM preparation. This study lends support to the no-tion that practicing cleft surgeons view NAM-prepared patients ashaving less severe clefts preoperatively and better anticipated surgicaloutcomes than nonNAM-prepared patients.FIGURE 3. Surgeons ratings of cleft severity and anticipated surgical outcomesamong UCLP patients by NAM preparation group. Estimates and 95% CIswere generated from CCMLMs. Note: All differences between NAM-preparedand nonNAM-prepared patient groups are significant (P < 0.001).

74

Copyright 2014 Mutaz B. Habal, MD. Unauthoring nonNAM-prepared UCLP as minimal severity 2.0% of thetime, making them approximately (relative risk, 53.3/2.0) 27 timesas likely (P < 0.001) to rate NAM-prepared ULCP as minimal se-verity. Results were similarly positive and significant regard-less of the measure used; surgeons were 7 times as likely to rateNAM-prepared UCLP patients as among their best surgical out-comes and 1.8 times as likely to say UCLP patients were unlikelyto need revision compared with nonNAM-prepared UCLP pa-tients. Analogous trends were evident for patients with BCLP.

This study has several limitations. For example, the responserate to the e-mailed, Web-based survey was low (24%). Histor-ically, clinicians, especially physicians, have been known to be achallenging group to survey,19 and recent evidence suggests a de-cline in physician response rates to surveys.20 To increase sampleFIGURE 4. Surgeons ratings of cleft severity and anticipated surgical outcomesamong BCLP patients by NAM preparation group. Estimates and 95% CIswere generated from CCMLMs. Note: All differences between NAM-preparedand nonNAM-prepared patient groups are significant (P < 0.001).

2014 Mutaz B. Habal, MD

ized reproduction of this article is prohibited.

size, we surveyed the entire population of plastic and oral surgeonsbelonging to the American Cleft Palate-Craniofacial Association,which yielded a substantial sample size (N = 176), representing ap-proximately one quarter of the population.

In addition, baseline assessments of NAM-prepared andnonNAM-prepared patients were not available. Such assessmentsmight bias study results if patients who received NAM preparationhad less severe pretreatment clefts than those without NAM prepa-ration. Previous research suggests that patients who receive NAMtherapy, on average, tend to have wider pretreatment clefts than

14

REFERENCES1. Dec W, Olivera O, Shetye P, et al. Cleft palate midface is both

hypoplastic and displaced. J Craniofac Surg 2013;24:89932. Latham RA. Development and structure of the premaxillary deformity

in bilateral cleft lip and palate. Br J Plast Surg 1973;26:1113. Salyer KE, Genecov ER, Genecov DG. Unilateral cleft lip-nose

repair: a 33-year experience. J Craniofac Surg 2003;14:5495584. Henkel KO, Gundlach K, Saka B. Incidence of secondary lip

surgeries as a function of cleft type and severity: one centersexperience. Cleft Palate Craniofac J 1998;35:310312

5. Grayson BH, Santiago PE, Brecht LE, et al. Presurgicalnasoalveolar molding in infants with cleft lip and palate.

The Journal of Craniofacial Surgery Volume 26, Number 1, January 2015 Surgeon Survey on NAM Cleft Treatmentnevertheless represents a major limitation. Future research shouldinclude such data.

Our analysis is also limited because the outcomes ana-lyzed here represent surgeons subjective ratings of anticipated out-comes and not actual outcomes. Surgical outcomes, including thelikelihood of revision, are influenced by multiple factors, some ofwhich are not visually evident (eg, myofibroblast contraction orwound healing characteristics). In general, it makes teleologic sensethat the degree of preoperative cleft severity would be associatedwith the long-term surgical outcome as well as the revision rate.Investigating this supposition, Henkel and colleagues4 found thatthe incidence of secondary cleft surgery was related to the initialcleft severity. Whether one agrees with Henkel and colleagues find-ings, our study suggests that surgeons prefer to operate on NAM-prepared patients.

Irrespective of the impact of NAM on predicted surgical out-comes, NAM therapy has potential limitations including a greaterinvestment of time and resources preoperatively than surgery aloneand the potential for caregiver burden. Our analyses suggest that,although NAM increases the preoperative investment, the dividendsmay pay off in a reduction in the need for secondary surgery. Futureanalyses should examine the extent to which NAM therapy im-pacts the overall cost of cleft care in relation both to the higher pre-operative investment and to the potential savings due to decreasedpotential need for secondary surgeries. Future research should alsoinvestigate caregiver responses to NAM and effect of NAM on fam-ily functioning.

Regarding facial appearance and surgical outcomes, thestudy findings are consistent with previous reports indicating a ben-eficial role for NAM. Other studies note that NAM may improvelong-term nasal symmetry and less need for secondary alveolarbone grafting when gingivoperiosteoplasty is performed.813 Ourstudy supports this research by showing that NAM treatment wasassociated with an improvement in surgeons ratings of patients.We found that, compared with nonNAM-prepared patients, sur-geons were more likely to rate NAM-prepared patients as havingminimal severity clefts and to anticipate that the patients wouldbe among their best surgical outcomes and less likely to need earlysurgical revision. Although further research is needed, insofar asresults represent a causal relationship, presurgical NAM prepara-tion may benefit patients, patient families, and surgeons. 2014 Mutaz B. Habal, MD

Copyright 2014 Mutaz B. Habal, MD. UnauthoCleft Palate Craniofac J 1999;36:4864986. Grayson BH, Cutting CB. Presurgical nasoalveolar orthopedic molding

in primary correction of the nose, lip, and alveolus of infants born withunilateral and bilateral clefts. Cleft Palate Craniofac J 2001;38:193198

7. Grayson BH, Maull D. Nasoalveolar molding for infants born withclefts of the lip, alveolus, and palate. Clin Plast Surg 2004;31:149158

8. Santiago PE, Grayson BH, Cutting CB, et al. Reduced needfor alveolar bone grafting by presurgical orthopedics and primarygingivoperiosteoplasty. Cleft Palate Craniofac J 1998;35:7780

9. Maull DJ, Grayson BH, Cutting CB, et al. Long-term effects ofnasoalveolar molding on three-dimensional nasal shape inunilateral clefts. Cleft Palate Craniofac J 1999;36: 391397

10. Barillas I, Dec W, Warren SM, et al. Nasoalveolar molding improveslong-term nasal symmetry in complete unilateral cleft lip-cleftpalate patients. Plast Reconstr Surg 2009;123:10021006

11. Clark SL, Teichgraeber JF, Fleshman RG, et al. Long-term treatmentoutcome of presurgical nasoalveolar molding in patients withunilateral cleft lip and palate. J Craniofac Surg 2011;22:333336

12. Lee CT, Grayson BH, Cutting CB, et al. Prepubertal midface growthin unilateral cleft lip and palate following alveolar molding andgingivoperiosteoplasty. Cleft Palate Craniofac J 2004;41:375380

13. Pfeifer TM, Grayson BH, Cutting CB. Nasoalveolar molding andgingivoperiosteoplasty versus alveolar bone graft: an outcomeanalysis of costs in the treatment of unilateral cleft alveolus.Cleft Palate Craniofac J 2002;39:2629

14. Sischo L, Chan JW, Stein M, et al. Nasoalveolar molding:prevalence of cleft centers offering NAM and who seeks it.Cleft Palate Craniofac J 2012;49:270275

15. Luo W, Kwok O-M. The impacts of ignoring a crossed factor inanalyzing cross-classified data. Multivar Behav Res 2009;44:182212

16. Goldstein H. Multilevel cross-classified models. Sociol Methods Res1994;22:364375

17. Duncan C, Jones K, Moon G. Context, composition andheterogeneity: using multilevel models in health research.Soc Sci Med 1998;46:97117

18. Wagenmakers EJ, Farrell S. AIC model selection using Akaikeweights. Psychon Bull Rev 2004;11:192196

19. Klabunde CN, Willis GB, McLeod CC, et al. Improving the qualityof surveys of physicians and medical groups a research agenda.Eval Health Prof 2012;35:477506

20. McLeod CC, Klabunde CN, Willis GB, et al. Health care providersurveys in the United States, 20002010: a review. Eval Health Prof2013;36:106126those who do not. The lack of data on pretreatment cleft severity75

rized reproduction of this article is prohibited.