European Journal of Epidemiology

, Volume 22, Issue 6, pp 389–395

Epidemiology of cleft palate alone and cleft palate with accompanying defects

Authors

    • Department of EpidemiologyUniversity of North Carolina-Chapel Hill
    • Department of EpidemiologyTulane University School of Public Health and Tropical Medicine
  • Allen J. Wilcox
    • Epidemiology BranchNational Institute of Environmental Health Sciences
  • Rolv Terje Lie
    • Department of Public Health and Primary Health CareUniversity of Bergen
    • Medical Birth Registry of Norway, Norwegian Institute of Public Health
  • Frank Åbyholm
    • Department of Plastic SurgeryThe National Hospital
  • Hallvard Vindenes
    • Department of Plastic SurgeryHaukeland University Hospital
Birth Defects

DOI: 10.1007/s10654-007-9129-y

Cite this article as:
Harville, E.W., Wilcox, A.J., Lie, R.T. et al. Eur J Epidemiol (2007) 22: 389. doi:10.1007/s10654-007-9129-y

Abstract

The epidemiology of cleft palate with multiple defects is often thought to be different from that of cleft palate alone, but there are few empirical data on this question. We explored this in a population-based data set created by combining data from two sources: the 1.8 million live births recorded from 1967 to 1998 in the Norwegian Birth Registry, and the two Norwegian surgical centers that repair cleft palate. Accompanying defects were identified from either source. Stratified analysis and logistic regression were used to assess relative risks by covariates. Of 1,431 babies with cleft palate, 31 % had another birth defect recorded by one or both sources. Prevalence of isolated cleft palate was steady over time, while cleft palate with other defects increased substantially. Girls had a higher risk of isolated cleft palate (relative risk 1.4; 95% confidence interval, 1.2–1.6) but not of cleft palate accompanied by other defects (1.1; 0.88–1.3). Older parents and parents who were first cousins had no increased risk of isolated cleft palate, but were twice as likely as others to have a baby with cleft palate accompanied by other defects. Risk factors differ between cases of cleft palate with and without accompanying defects.

Keywords

Birth defects; Cleft palate; Multiple abnormalities; Sex distribution

Abbreviations

CP

Cleft palate

ICD

International classification of diseases

RR

Relative risk

OR

Odds ratio

Introduction

A case of “cleft palate” can refer to one of a heterogeneous group of defects. In most studies, the estimated occurrence of cleft palate with accompanying defects is 25% or less; [16] but other studies have reported 45% up to 70% [710] A few specific combinations of defects are known, such as the Pierre Robin Sequence, but most combinations are not recognized as specific syndromes [11]. Still, cases with multiple defects are often called “syndromic” cases. Prevalence rates likely differ due to varying ascertainment methods, study populations, and case definitions [12].

Most studies of CP etiology limit their cases to isolated CP, while others do not distinguish between cases of CP that occurred in isolation and cases that occurred in conjunction with other defects [1316]. The rationale for grouping all cases is that the same etiological factors may affect the risk of CP regardless of whether other defects are present [17]. Also, the cases with multiple defects can comprise a substantial portion of all cleft palate cases, and (all else being equal) the statistical power is increased by including all cases. There is some evidence that similar genes contribute to both syndromic and non-syndromic cases, perhaps modified or with variable penetrance [18].

Vieira and Orioli, in their meta-analysis, present data for syndromic and non-syndromic clefts cases both pooled and separately [19]. Along the same lines, other studies exclude recognized syndromes or genetic disorders, but group isolated CP cases with clefts having accompanying defects of no clear origin. Sayetta recommended classifying cases as familial or sporadic, then syndromic and non-syndromic within these categories [20]. Tolarova et al. suggest dividing isolated cases from those with accompanying defects, and further dividing the accompanying defects into monogenic, chromosomal, known environmental, associations, and defects of unknown etiology [10]. Christensen simply recommends performing analyses of clefts with and without accompanying defects [17].

It has been argued that clues to etiology are most likely to be found among “sporadic, non-syndromic cases that probably best represent the gene/environment interaction type” [21]. The purpose of this study is to describe and compare the epidemiology of CP with or without accompanying defects.

Methods

Norwegians are assigned a unique identifying number at birth. Since 1967, all births in Norway have been required to be reported in the Medical Birth Registry, which includes all live and stillbirths starting at 16 weeks of gestation. Care for oral clefts is centralized at two hospitals, Haukeland in Bergen and Rikshospitalet in Oslo; virtually all cases in the country over the last 50 years have been sent to one of these two hospitals for treatment.

The medical records and birth registry were linked by identifying numbers, and the analyses were restricted to livebirths. Cases of CP in the medical birth registry were identified as those with any of four possible blanks for birth defects containing International Classification of Disease (ICD)-8 code 7592, excluding those with cleft lip or cleft lip and palate (codes 7590 and 7591). In 65 instances, the birth registry and the hospital records agreed that a cleft was present but disagreed on the type. In these cases, the hospital’s classification was used. The cases where the registry reported CP and the hospital reported no cleft (N = 240) are a mix of infant deaths (N = 76), clefts too minor to require surgery, clefts operated at another hospital in Norway, children who left the country, and false positives. Since we have no way of distinguishing false positives from the other categories, all these children are included in the dataset as cases.

Concurrent birth defects for the birth registry were defined as a birth defect of ICD-8 codes 7400–7489 and 7493–7599. Until 1998, birth defects ascertained in the first 7 days were recorded in the registry. Cases of cleft lip with CP were not considered to have a concurrent defect. Concurrent birth defects from the hospital were defined as a birth defect noted in the field “Other defects or syndrome diagnosis”. This was examined by a Norwegian speaker to determine which were defects and which were other notations (for instance, “no defect” or “twin”). A birth defect recorded at any point during treatment, at any age, would be included in this record. No information was available specifically about genetic diagnoses. For the purposes of this analysis, a defect recorded by either source is considered an accompanying defect. We did not exclude any defects on the grounds that they were minor; the vast majority of the defects were at least potentially serious. Possible exceptions are undescended testes (n = 4), hip dislocation (n = 3), and patent ductus arteriosus (n = 1). (Other infants had these problems as well, but they occurred in conjunction with other defects.)

Ascertainment was assessed using capture-recapture methods. These methods use the amount of overlap between cases from the two data sources to estimate the proportion of cases missed by both sources [22]. This method assumes that the sources are independent, which is not the case in this analysis: a case noted at birth would be more likely to be referred for treatment. We carried out stratified capture-recapture analyses to determine whether ascertainment varied by characteristics of the mother, child, or birthplace.

For the remaining analyses, we combined the two sources of information on clefts to define the case group. We assessed time trends, and associations with birthweight, and gestational age. The time trend data were smoothed using proc loess in SAS with a window of 0.4. All infants had a record in the medical birth registry and so had data on covariates, even if a cleft had not been recorded at birth. Prevalence at birth was stratified by available covariates, including age of the mother and father, sex of the baby, region, and mother’s marital status. Data were examined by stratified analysis using chi-square tests to test for statistical significance of differences in pattern. Logistic regression was used to adjust mutually for the effects of several variables.

This study was approved by the Norwegian Data Inspectorate.

Results

Approximately half the cases were recorded by both sources (Table 1). More cases were reported by the hospital than by the registry. Many of the cases missed by the hospital died before they could get treatment; of the 939 cases reported by the birth registry, 8% died in the first year, while only one of the 1,227 cases reported by the hospital died in the first year.
Table 1

Capture-recapture estimates of the prevalence of cleft palate in Norway, 1967–1998

 

Registry

No cleft palate

Cleft palatea

Hospital:

No cleft palate

b

240

Cleft palate

528

699

Capture-recapture estimate, missed cases (95% confidence interval)

181 (141, 222)

Capture-recapture estimate, total cases

1,648

Completeness of registry

57%

Completeness of hospital records

74%

a Includes 36 cases defined as cleft lip by hospital (excluded from the remainder of this analysis)

b 1,846,613 cases were negative for CP by both measures

In general, no clear patterns emerged when ascertainment was examined by region, hospital size, hospital type, or mother’s marital status. The only consistent association with under-ascertainment was with the highest maternal age (>40 years), among whom ascertainment was estimated to be 50% for the registry and 56% for the clinics. However, only a small number of births fit in this category.

Eighteen percent of CP cases had another defect reported in the Medical Birth Registry, while 23% had a defect reported by the hospitals (Table 2). Only 10% had a defect reported in both sources, meaning that, overall, 31% had an accompanying defect. Among the 699 with a recorded CP in both sources, 96 had a birth defect recorded in the registry and 173 in the hospital, and 72 (58%) of these overlapped.
Table 2

Accompanying defects in cases of cleft palate, Norway, 1967–1998. 36 registry cleft palate cases that the hospital recorded as cleft lip are excluded

  

Registry

  

No cleft palate

Cleft palate, no concurrent birth defect

Concurrent birth defect

Hospital

No cleft palate

124

80

Cleft palate, no concurrent defect

364

502

24

Cleft palate, concurrent defect

164

101

72

The prevalence of CP with no other defect among livebirths appeared to stay steady over time (OR 1.0, 95% CI 0.99, 1.01), while the prevalence of CP and accompanying defects rose (OR for each year: 1.03, 95% CI 1.02, 1.04) (Fig. 1). Overall prevalence for the 30-year period was 7.7/10,000 livebirths: 5.3/10,000 for CP and no other defect, and 2.4/10,000 for CP with other defects. Eighteen of the 21,305 stillbirths had a recorded CP (8.5/1000); 10 of those (56%) had an accompanying defect. These cases were excluded from our analysis because they could only be captured by one data source.
https://static-content.springer.com/image/art%3A10.1007%2Fs10654-007-9129-y/MediaObjects/10654_2007_9129_Fig1_HTML.gif
Fig. 1

Prevalence of cleft palate with and without accompanying defects, 1967–1998. Dashed lines are absolute and heavier lines are smoothed prevalence

Mortality was high among CP babies, with overall mortality around 5%; mortality for infants without clefts was 0.1%. However, mortality among babies with CP alone was 1.5%, versus 13.6% for infants with CP and other defects. Mortality declined in all groups over the years of the study.

Among children without a cleft, 1.7% were born very preterm (<34 weeks), 4% were born moderately preterm (34–37 weeks), and mean birthweight among fullterm babies was 3,567 g. Children with a CP and no other defects were less likely to be born preterm (1.4% and 3.7%, respectively) but had a smaller mean birthweight among fullterms (3,458 g). Children with CP and another defect were much more likely to be born preterm (4.7% very preterm and 8.3% moderately preterm) and small (mean birthweight among fullterm: 3,331 g).

Girl infants were more likely to have CP alone, but the sex ratio was almost even for CP with other defects. For other covariates, CP with no accompanying defect was higher in children born to single mothers and those with higher parity (Table 3). CP with an accompanying defect was higher among children whose parents were cousins. A U-shape was seen with both maternal and paternal age and CP alone (Table 4), with the highest risk occurring in the youngest and oldest parents. Risk rose with maternal and paternal age for CP and accompanying defect. In order to limit potential confounding from the strong correlation of maternal and paternal age, we also examined the risk associated with increasing age among those infants whose other parent was in a low-risk group, ages 20–29. Among those with mothers aged 20–29, CP with no other defect was most prevalent among the youngest fathers, while CP with other defects was more prevalent among older fathers (those over age 25 compared with those 20–24).
Table 3

Cleft palate in Norway, 1967–1998, with and without accompanying defects, by covariates

 

No accompanying defect

Accompanying defect

N

cases per 10,000

RRa

95% CI

N

cases per 10,000

RR

95% CI

Sex of child

Male

422

4.4

1

 

224

2.4

1

 

Female

553

6.2

1.39

(1.22, 1.57)

232

2.6

1.10

(0.91, 1.32)

Mother’s marital status

Married

704

5.0

1

 

337

2.4

1

 

Cohabiting

146

5.9

1.17

(0.98, 1.40)

74

3.0

1.24

(0.96, 1.59)

Single

108

6.2

1.24

(1.01, 1.52)

38

2.2

0.91

(0.65, 1.28)

Widowed/Divorced

13

6.9

1.37

(0.79, 2.37)

4

2.1

0.88

(0.33, 2.36)

Relationship between parents

None

960

5.3

1

 

441

2.4

1

 

First cousins

3

4.8

0.90

(0.29, 2.81)

3

4.8

1.97

(0.63, 6.12)

Second cousins

3

4.7

0.88

(0.28, 2.74)

3

4.7

1.92

(0.62, 5.98)

Plurality

Singleton

955

5.3

1

 

443

2.5

1

 

Twin

20

5.0

0.94

(0.60, 1.47)

12

3.0

1.22

(0.69, 2.16)

Region

East

489

5.2

1

 

238

2.5

1

 

Middle/West

350

5.1

0.98

(0.86, 1.13)

166

2.4

0.96

(0.79, 1.17)

North

136

6.1

1.17

(0.96, 1.41)

52

2.3

0.92

(0.68, 1.24)

Parity

0

398

5.2

1

 

192

2.5

1

 

1

310

4.9

0.93

(0.80, 1.08)

136

2.1

0.85

(0.68, 1.06)

2

167

5.6

1.07

(0.90, 1.29)

89

3.0

1.19

(0.92, 1.53)

3

64

6.9

1.32

(1.01, 1.71)

24

2.6

1.02

(0.67, 1.56)

4+

34

7.9

1.51

(1.06, 2.14)

13

3.0

1.20

(0.68, 2.10)

a RR, relative risk; CI, confidence interval

Table 4

Parental age and cleft palate with or without accompanying defect, Norway, 1967–1998

 

Cleft palate, no accompanying defect

Cleft palate, accompanying defect

Mother’s age

 

N

cases per 10,000

RRa

95% CI

N

cases per 10,000

RR

95% CI

<20

 

74

5.8

1.25

(0.97, 1.62)

24

1.9

0.93

(0.60, 1.44)

20–24

 

256

4.6

1

 

112

2.0

1

 

25–29

 

341

5.3

1.15

(0.97, 1.35)

154

2.4

1.18

(0.93, 1.51)

30–34

 

202

5.5

1.18

(0.98, 1.42)

109

2.9

1.45

(1.12, 1.89)

35–39

 

84

6.4

1.37

(1.07, 1.76)

44

3.3

1.64

(1.16, 2.33)

≥40

 

18

6.9

1.49

(0.92, 2.40)

13

5.0

2.46

(1.38, 4.36)

Father’s age

<20

 

11

10.7

2.21

(1.19, 4.09)

2

1.9

1.37

(0.33, 5.70)

20–24

 

123

4.8

1

 

36

1.4

1

 

25–29

 

271

4.7

0.96

(0.78, 1.19)

149

2.6

1.80

(1.25, 2.60)

30–34

 

264

5.3

1.10

(0.89, 1.36)

128

2.6

1.82

(1.26, 2.64)

35–39

 

133

5.4

1.11

(0.87, 1.41)

77

3.1

2.19

(1.47, 3.25)

≥40

 

91

6.8

1.40

(1.07, 1.83)

43

3.2

2.26

(1.45, 3.51)

Father’s age

Mother’s age

        

20–29

<20

42

5.3

1.21

(0.86, 1.69)

18

2.3

1.10

(0.67, 1.83)

20–24

188

4.4

1

 

88

2.1

1

 

25–29

140

4.8

1.08

(0.87, 1.35)

64

2.2

1.06

(0.77, 1.46)

30–34

22

6.9

1.57

(1.01, 2.44)

12

3.8

1.83

(1.00, 3.34)

≥35

2

6.0

1.37

(0.34, 5.50)

3

9.0

4.38

(1.39, 13.82)

Mother’s age

Father’s age

        

20–29

<20

3

13.8

2.88

(0.91, 9.09)

0

0.0

0.00

 

20–24

91

4.8

1

 

22

1.2

1

 

25–29

237

4.5

0.93

(0.73, 1.18)

130

2.4

2.11

(1.34, 3.31)

30–34

163

5.3

1.10

(0.85, 1.42)

80

2.6

2.23

(1.39, 3.58)

≥35

49

5.5

1.15

(0.81, 1.63)

23

2.6

2.24

(1.25, 4.01)

a RR, relative risk; CI, confidence interval

Logistic regression generally showed similar results to stratified analysis (Table 5). Even when adjusted for maternal and paternal age, increased parity was associated with increased odds of CP and no other defect. Odds of CP with other defects was reduced in women of higher parity.
Table 5

Logistic models of cleft palate with and without accompanying defects, Norway, 1967–1998

 

Cleft palate and no accompanying defect

Cleft palate with accompanying defect

ORa

95% CI

OR

95% CI

Mother’s age

<20

1.21

(0.88, 1.69)

1.16

(0.68, 2.00)

20–24

1

 

1

 

25–29

1.13

(0.93, 1.37)

1.09

(0.82, 1.43)

30–34

1.13

(0.88, 1.46)

1.30

(0.92 , 1.84)

35–39

1.21

(0.86, 1.70)

1.51

(0.95, 2.41)

≥40

1.40

(0.81, 2.40)

2.03

(0.95, 4.31)

Father’s age

<20

1.93

(1.02, 3.68)

1.18

(0.27, 5.05)

20–24

1

 

1

 

25–29

0.93

(0.73, 1.18)

2.05

(1.36, 3.08)

30–34

1.00

(0.76, 1.31)

1.96

(1.24, 3.08)

35–39

0.94

(0.68, 1.30)

2.12

(1.28, 3.50)

≥40

1.10

(0.76, 1.60)

1.99

(1.12, 3.53)

Female vs. male

1.42

(1.25, 1.63)

1.09

(0.90, 1.32)

Marital status

Cohabiting

1.20

(0.99, 1.44)

1.22

(0.93, 1.58)

Single

1.15

(0.83, 1.59)

1.45

(0.93, 2.26)

Widowed/Divorced

1.27

(0.60, 2.67)

0.38

(0.05, 2.72)

Married

1

 

1

 

Parents first cousins

0.93

(0.30, 2.91)

2.16

(0.69, 6.76)

Twins

0.89

(0.56, 1.41)

1.19

(0.67, 2.11)

Region

Central/West

0.88

(0.72, 1.09)

1.14

(0.82, 1.59)

North

0.87

(0.70, 1.07)

1.10

(0.78, 1.54)

East

1

 

1

 

Parity

0

1

 

1

 

1

0.97

(0.82, 1.15)

0.74

(0.59, 0.94)

2

1.07

(0.86, 1.33)

0.94

(0.70, 1.26)

3

1.28

(0.95, 1.74)

0.68

(0.42, 1.10)

4

1.45

(0.91, 2.29)

1.09

(0.57, 2.08)

5+

1.20

(0.65, 2.21)

0.15

(0.02, 1.06)

a OR, odds ratio; CI, confidence interval

Discussion

Our analysis of a large, population-based registry demonstrates that CP with other defects and CP alone have different epidemiologic features. For instance, the sex ratio was skewed towards girls among cases of CP alone, but not among cases of CP with other defects. This finding confirms the work of Czeizel and Tusnadi [23], but contrasts with some other studies, which found a female preponderance for both isolated and non-isolated CP [7, 24]. We found a higher risk for CP alone among higher-parity women, even when adjusted for maternal and paternal age, but this pattern was absent among cases of CP with accompanying defects. The patterns of association were different for parental age as well: the risk of CP and accompanying defects rose more steeply with maternal age than for CP alone, confirming the results of Forrester and Merz [7]. Unlike Czeizel and Tusnadi, who found increased risk of CP alone with increased paternal age [23], we found that risk of CP with other defects rose with paternal age, but risk of CP alone was highest in the youngest fathers. Similar to Wyszynski et al. and Czeizel and Tusnadi, but unlike Forrester and Merz, we found a higher risk of LBW among isolated CP infants, but no higher risk of preterm birth [7, 23, 25].

There was no evidence that related parents had an increased risk of CP alone. Related parents had a higher risk of CP with accompanying defects, although the confidence intervals overlapped substantially. This could be due to recessive genetic factors, or may be a marker for membership in ethnic groups that practice cousin marriage. CP alone was most common in the northern region, while CP with other defects was least common there, though again, the confidence intervals overlap.

A Danish study concluded that the female preponderance among CP only was probably due to differences in ascertainment [26]. Our data do not support that idea. In our data, girls were at higher risk for CP and no accompanying defect in both the birth registry and the treatment record. However, the registry reported more boys with CP and another defect, while in the hospital, the pattern was reversed. Male children may be more likely to have more serious defects; infants who died in the first year were more likely to be male. The rise in CP with an accompanying defect may be due better ascertainment of CP in cases of other birth defects, as well as better survival among infants with defects. Similar time trends have been noted in Denmark [17].

Under-ascertainment of birth defects is common in population-based birth registries. CP cases noticed at birth are more likely to be referred for treatment; more severe cases are more likely to be noticed at birth. Therefore, positive dependence is present, in violation of the basic capture-recapture assumptions, and under-ascertainment is probably greater than our calculations would indicate. However, the 74% ascertainment of CP estimated for the hospital in this study was higher than for the similar Danish treatment centers [17]. The fact that ascertainment did not seem to vary substantially by covariates indicates that it is probably not a cause of the associations that we saw. Misclassification is also a pitfall of registry studies, although having clinical data to supplement the registry reduces the problem.

Use of the two sources gave an estimate of 31% of cases having accompanying defects, higher than most studies, but lower than others [9, 10]. There was very little overlap in the two sources, indicating that accompanying defects were probably under-ascertained. There are several reasons for the lack of overlap. One possibility is that that defects non-specifically recorded as related defects in the birth registry would be later correctly classified as CP by the hospital. This appears not to have been the case. Overall, 49 infants in the total dataset had code 7459 (unspecified anomalies of face and neck). One of these had a CP recorded by both the hospital and the registry, and the hospital recorded an accompanying ear defect. Of the infants classified as having CP by the hospital but not the registry, 12 had ICD code 7458 (other specified anomalies of face and neck); 10 of these were classified by the hospital as Pierre Robin Sequence (a sequence that includes facial anomalies beyond CP). Twenty-two infants among the CP group had code 7560 (anomalies of skull and face bones); 14 of these were classified by the hospital as Pierre Robin Sequence, 2 as multiple defects, 1 as a facial defect, and 5 as no other defect. A complete clinical and genetic workup would of course provide much more detail than is available in a population-based registry, including characterization of genetic syndromes, familial cases, and phenotypic subtypes.

Another possibility is that severe defects were recorded by both sources, but minor defects were not. This may be true to a point, but it operates within a limited range. In the majority of cases where both the CP and the accompanying defect were recorded only in the birth registry, the infant died in the first year. This group likely includes infants with the most serious defects. Beyond this, the cell sizes become small and conclusions are difficult to draw, though severe and discrete defects may be reported more accurately. To illustrate, of the five reported cases of co-occurring anal atresia, two died in the first year, and the other three were reported in both sources. There were 24 cases where CP was recorded by both the registry and the hospital and a co-occurring case of clubfoot was recorded. Of those, 12 reported clubfoot in sources, 5 only in the registry, 2 only in the hospital, 4 had a defect reported differently by the two sources, and in 1 case the hospital recorded “multiple malformations” but no specific defect. There were 17 cases where CP was recorded by both the registry and the hospital and a co-occurring heart/circulatory system defect was also recorded. Of those, 5 had a heart defect reported by both sources, 2 had the defect reported only in the registry, 8 only in the hospital, 1 had different defects reported by the two sources, and in 1 case the hospital recorded “multiple malformations” but no specific defect. Overall, there was more variation in recording minor or hidden defects, but these differences were not systematic in the direction of the hospital or the registry capturing more cases.

The most important limitations in our data on accompanying defects––specifically under-ascertainment and misclassification––would be expected to dilute any observed differences between the two case groups. Thus, there may be differences between the two CP groups that we were unable to detect. However, this limitation strengthens the case that real differences exist. A focusing on isolated CP (as many studies already do) may increase the power to detect causes of clefting by excluding cases with diverse etiologies. The higher risk in females, for instance, is a feature unusual in birth defect epidemiology, and one seemingly limited to isolated CP.

Acknowledgments

At the time of this research, Emily Harville was a Howard Hughes Medical Institute Predoctoral Fellow.

Copyright information

© Springer Science+Business Media B.V. 2007