seasonality of congenital anomalies in europe

Seasonality of Congenital Anomalies in EuropeJohannes Michiel Luteijn1, Helen Dolk*1, Marie-Claude Addor2, Larraitz Arriola3,Ingeborg Barisic4, Fabrizio Bianchi5, Elisa Calzolari6, Elizabeth Draper7, Ester Garne8,Miriam Gatt9, Martin Haeusler10, Babak Khoshnood11, Bob McDonnell12, Vera Nelen13,Mary O’Mahony14, Carmel Mullaney15, Annette Queisser-Luft16, Judith Rankin17,David Tucker18, Christine Verellen-Dumoulin19, Hermien de Walle20, and Lyubov Yevtushok21

Background: This study describes seasonality of congenital anomalies inEurope to provide a baseline against which to assess the impact of specifictime varying exposures such as the H1N1 pandemic influenza, and to providea comprehensive and recent picture of seasonality and its possible relation toetiologic factors. Methods: Data on births conceived in 2000 to 2008 wereextracted from 20 European Surveillance for Congenital Anomalies population-based congenital anomaly registries in 14 European countries. We performedPoisson regression analysis encompassing sine and cosine terms toinvestigate seasonality of 65,764 nonchromosomal and 12,682 chromosomalcongenital anomalies covering 3.3 million births. Analysis was performed byestimated month of conception. Analyses were performed for 86 congenitalanomaly subgroups, including a combined subgroup of congenital anomaliespreviously associated with influenza. Results: We detected statisticallysignificant seasonality in prevalence of anomalies previously associated withinfluenza, but the conception peak was in June (2.4% excess). We alsodetected seasonality in congenital cataract (April conceptions, 27%), hipdislocation and/or dysplasia (April, 12%), congenital hydronephrosis (July,

12%), urinary defects (July, 5%), and situs inversus (December, 36%), butnot for nonchromosomal anomalies combined, chromosomal anomaliescombined, or other anomalies analyzed. Conclusion: We have confirmedpreviously described seasonality for congenital cataract and hip dislocationand/or dysplasia, and found seasonality for congenital hydronephrosis andsitus inversus which have not previously been studied. We did not findevidence of seasonality for several anomalies which had previously beenfound to be seasonal. Influenza does not appear to be an important factor inthe seasonality of congenital anomalies.

Birth Defects Research (Part A) 100:260–269, 2014.VC 2014 Wiley Periodicals, Inc.

Key words: abnormalities, congenital; Seasonal Variation; epidemiology; publichealth; Influenza, Human

IntroductionStudying the effects of unique events such as the 2009influenza pandemic on congenital anomaly rates can beconfounded by underlying seasonal factors. Unlike most

time-series analyses, where the population at risk remainsrelatively stable, the population of women currently preg-nant, or in a specific stage of pregnancy, is dynamic. Forthis reason, a clear understanding of seasonal variation in

1Institute of Nursing Research/School of Nursing, University of Ulster, Jordanstown Campus, Newtownabbey, United Kingdom2Division of Medical Genetics, Lausanne, Switzerland3Public Health Division of Gipuzkoa, Instituto BIO-Donostia, Basque Government, CIBER Epidemiolog�ıa y Salud P�ublica - CIBERESP, Spain4Children’s Hospital Zagreb, Medical School University of Zagreb, Croatia5Unit of Epidemiology, Institute of Clinical Physiology, Pisa, Italy6IMER Registry - Azienda Ospedaliero-Universitaria di Ferrara, Ferrara, Italy7Department of Health Sciences, University of Leicester, Leicester, United Kingdom8Department of Paediatrics, Hospital Lillebaelt, Kolding, Denmark9Malta Congenital Anomalies Registry, Department of Health Information, Guardamangia, Malta10Medical University of Graz, Austria11INSERM U953, Paris, France12Health Service Executive, Dublin, Ireland13Provinciaal Instituut voor Hygiene, Antwerp, Belgium14Health Service Executive, Cork, Ireland15Health Service Executive, Kilkenny, Ireland16University Medical Center of the Johannes Gutenberg University, Mainz, Germany17Institute of Health & Society, Newcastle-upon-Tyne, United Kingdom18Public Health Wales, United Kingdom19Institut de Pathologie et de G�en�etique, Charleroi, Belgium20University Medical Center Groningen, The Netherlands21OMNI-Net for Children and Rivne Medical Diagnostic Center, Rivne, Ukraine

The study was co-funded by the EC, under the framework of the EU Health Programme, Grant Agreement 2006103 (Executive Agency for Health & Consumers).

*Correspondence to: Helen Dolk, University of Ulster, Jordanstown Campus, Shore Road, Newtownabbey, Co. Antrim, BT37 0QB. E-mail: [email protected]

Published online 17 March 2014 in Wiley Online Library (wileyonlinelibrary.com). Doi: 10.1002/bdra.23231

VC 2014 Wiley Periodicals, Inc.

an ordinary situation is fundamental for interpreting theresults of a study conducted during unique events such aspandemics, or environmental disasters. Studying the sea-sonality of congenital anomalies can also help to furtherunravel their etiology.

One of the earliest analyses of seasonality in congenitalanomalies was carried out by McKeown and Record in1951 and described a seasonal pattern in the prevalenceof anencephaly, with more cases being born in the autumnand winter (McKeown and Record, 1951). No formal con-clusion was made in this study. Many potential etiologicfactors are seasonal including air pollution, influenza,other infectious diseases, dietary intake, and pesticides.Previous epidemiologic studies have investigated seasonal-ity in a range of different types of congenital anomaly(Record and Edwards, 1958; Edwards, 1961; Slater et al.,1964; Chen et al., 1970; Heikkila, 1984; Bound et al.,1989; Anand et al., 1992; Haargaard et al., 2005; Caton,2012). Results differ in different populations, and with theuse of different statistical methods (Caton, 2012) and sam-ple sizes. Because seasonal variation in environmental fac-tors changes over time, an assessment in recentpopulations is needed. Seasonality has also been associ-ated with spontaneous abortions (McDonald, 1971; San-dahl, 1974; Kallan and Enneking, 1992), which also raisesthe possibility that it is in utero survival of the fetus withcongenital anomaly that varies seasonally.

In the context of the surveillance response of the Euro-pean Surveillance for Congenital Anomalies (EUROCAT) tothe 2009 H1N1 pandemic influenza, we undertook ananalysis of seasonality of congenital anomalies for preg-nancies conceived in 2000 to 2008 to determine theunderlying seasonality of congenital anomalies in the pre-pandemic situation. Many of the registries participating inEUROCAT collect monthly birth data for their catchmentpopulation, allowing for analysis of congenital anomalyprevalence, rather than congenital anomaly counts. This,as well as the large coverage of 31% of European births,makes EUROCAT well suited for analyzing the seasonalityof congenital anomalies. The aim of this study is to analyzethe effect of season on the prevalence of all congenitalanomaly subgroups routinely surveyed by EUROCAT, manyof which have never been analyzed before in the contextof seasonality on the scale of the dataset in this study. Wefurther hypothesize that congenital anomalies previouslyassociated with influenza will show seasonal prevalencepeaks corresponding with the influenza season.

Materials and MethodsEUROCAT is a European network of population-basedregistries for the surveillance of congenital anomalies.EUROCAT collects information on congenital anomalies inlive births, stillbirths, fetal death from 20 weeks gestationand terminations of pregnancy for fetal anomaly at any

gestation. EUROCAT registries, which are described indetail by Greenlees et al (Greenlees et al., 2011), transmita standard anonymized variable set to a central database.EUROCAT Registries were included if they provided indi-vidual data (full member registries), provided gestationalage for >75% of the congenital anomalies cases, reporteda total congenital anomaly prevalence of over 200 per10,000 births in the catchment population, were located inthe Northern hemisphere (leading to exclusion of theReunion and French West Indies registers) and could pro-vide monthly birth data. Twenty EUROCAT registries from14 European countries met the inclusion criteria.

BIRTHS IN GENERAL POPULATION DATA

Total births per month for the populations of the includedEUROCAT registries were obtained for 2000 to 2009. Forthe Danish, English, and Welsh registries, only total annualbirths of the catchment population were available. Forregistries from these countries, total annual birthsreported by the registries were distributed over themonths using the monthly distribution of national birthdata per individual year. National monthly birth data (live-births) were extracted from the Office for National Statis-tics (ONS) for England and Wales (Office for NationalStatistics licensed under the Open Government Licensev.1.0, 2011) and from Statistics Denmark (Statistics Den-mark, 2012) for Denmark. Birth data for Northern Nether-lands, Styria, and Vaud are based on live births only, whilebirth data for the remaining registries included live andstillbirths.

For each month, births were divided evenly over thecorresponding number of days for each month, after whicheach daily total of births was calculated back to estimatedday of conception by subtracting 39 weeks (average gesta-tional length, to obtain estimated day of last menstrualperiod [LMP]) and adding 2 weeks. Next, daily estimatednumbers of conceptions were aggregated to monthly esti-mated numbers of conceptions. Only years fully coveredby month of estimated conception were retained for eachregistry. For example, if delivery data covers 2000 to2009, conception covers 2000 to 2008 fully while both1999 and 2009 are covered partially and thereforeremoved to avoid bias.

From a population of 3,687,000 births in 2000 to2009, a total of 3,295,000 births conceived during 2000 to2008 and were included in the study after calculatingback to month of conception and excluding years not fullycovered.

CONGENITAL ANOMALY DATA

Up to eight malformations are coded with the Interna-tional Classification of Diseases (ICD) version 10 for eachEUROCAT case, and according to these codes, cases areallocated to 89 standard subgroups (EUROCAT CentralRegistry, 2009). Congenital anomaly subgroups were

BIRTH DEFECTS RESEARCH (PART A) 100:260–269 (2014) 261

divided into congenital anomaly subgroups previouslyassociated with seasonality and/or influenza (priorhypothesis, for hypothesis testing) and congenital anomalysubgroups not previously associated with seasonality and/or influenza (no prior hypothesis, for hypothesis genera-tion) as shown in Table 2.

Congenital anomalies were classified as previouslyassociated with influenza based on the results of a system-atic review and meta-analysis (Luteijn et al., 2013).Although the etiology of chromosomal CA is unlikely to berelated to seasonal factors, seasonality can play a role inchromosomal CA as for example the seasonal reproductivepatterns of parents at increased risk can vary from thegeneral population. For this reason, it was decided to ana-lyze chromosomal congenital anomaly cases and nonchro-mosomal congenital anomaly cases separately. To identifycongenital anomaly subgroups previously associated withseasonality, we performed literature searches of EmbaseVR

and PubMedVR using the following MeSHTM terms andBoolean operators: “season” and “congenital abnormality”without year restrictions.

Registrations of congenital anomalies delivered in theperiod 2000 to 2009 for eligible EUROCAT registries wereextracted from the EUROCAT central database for all 89EUROCAT congenital anomaly subgroups (EUROCAT Cen-tral Registry, 2009), together with gestational age and dateof birth for each case. Not all included registries were ableto supply data for this entire period (Table 1). By subtract-ing gestational age from day of birth and adding 2 weeks,estimated month of conception was obtained for each con-genital anomaly registration and years not fully coveredfor conception were removed for each registry matchingthe births data. Congenital anomaly subgroups not of priorhypothesis with <100 cases (n5 3) were not analyzed,but did count toward “all anomalies” and cases wereincluded in other subgroups in case of multiple anomaly.Teratogenic syndrome subgroups (n5 3) were not ana-lyzed and registrations under teratogenic syndrome sub-groups were excluded from all analyses. Three newcongenital anomaly subgroups were generated: congenitalanomalies previously associated with influenza (with andwithout congenital heart defects) and congenital anomaliesnot previously associated with influenza. Including thesethree new groups, a total of 86 congenital anomaly sub-groups were analyzed.

Because aortic valve atresia/stenosis, cystic adenoma-tous malformation of the lung, and skeletal dysplasia wereaffected by moving from ICD-9 to ICD-10, years coveredby ICD-9 were removed from analysis for these congenitalanomalies. This applies to Antwerp 2000, Paris 2000,Mainz 2000, Dublin 2000, SE Ireland 2000, Emilia Roma-gna 2000 to 2002, Tuscany 2000 to 2001, North Nether-lands 2000 to 2001, and Vaud 2000.

A total of 3,295,000 births including 65,764 nonchro-mosomal congenital anomaly registrations and 12,682

chromosomal congenital anomaly registrations of pregnan-cies conceived during 2000 to 2008 were included in ouranalysis. The number of included cases ranged from 60 foranophthalmos to 20,201 for congenital heart defects. Asingle congenital anomaly registration can be in multiplesubgroups; for example a ventricular septal defect alsofalls under congenital heart defects.

STATISTICAL ANALYSIS

Seasonality of congenital anomaly prevalence was analyzedby estimated month of conception. Analysis of congenitalanomaly prevalence rather than counts eliminates severaldisturbing effects from the model, including varyinglengths of months and seasonal conception patterns (Fell-man and Eriksson, 2000). Monthly congenital anomalyprevalence was calculated by dividing congenital anomalyregistrations by total pregnancies conceived that month,for each estimated month of conception.

TABLE 1. Average Number of Annual Births, and Study Period Covered, byEach of the Participating EUROCAT Registries.

Country and registry Study periodAverage

annual births

Austria, Styriaa 2000–2008 10,304

Belgium, Antwerp 2000–2008 19,011

Belgium, Hainaut 2004–2007 11,805

Croatia, Zagreb 2000–2008 6,495

Denmark, Odensea 2000–2008 5,329

France, Paris 2000–2008 27,035

Germany, Mainz 2000–2006 3,109

Ireland, Cork & Kerry 2000–2004 8,346

Ireland, Dublin 2000–2008 24,015

Ireland, SE Ireland 2000–2008 7,022

Italy, E. Romagna 2000–2008 34,839

Italy, Tuscany 2000–2008 29,019

Malta, Malta 2000–2008 3,929

Netherlands, N. Netherlandsa 2000–2008 18,928

Spain, Basque Country 2000–2007 19,491

Switzerland, Vauda 2000–2008 7,275

UK, EMSYCARa 2000–2008 65,814

UK, N. Englanda 2000–2008 31,471

UK, Walesa 2000–2008 32,866

Ukraine, Ukraine 2005–2008 29,659

TOTALb 2000–2008 366,159

aFor Austrian, British, Danish, Dutch, and Swiss data, deliverieswere limited to live births only.bNote TOTAL average annual births is lower than the sum of averageannual births for all EUROCAT registries combined since someEUROCAT registries did not cover the entire study period.

262 SEASONALITY OF CONGENITAL ANOMALIES IN EUROPE

Analysis was performed using Poisson regression anal-ysis in Stata 9.2 encompassing sine and cosine terms asdescribed by Stolwijk et al. (1999). Seasonal peaks and

troughs can be calculated by solving t5 arctan b1b2

� �x T2p. T

is the number of time periods described by one cosinefunction. The Stolwijk-method allows for fitting multiplesine and cosine functions into a single model and both 12-and 6-month periods were explored (T5 12 and T5 6,respectively). b1 is the coefficient of the sine function andb2 is the coefficient of the cosine function in the Poissonregression. The amplitude a, the difference between meanand peak prevalence, can be calculated by solving

a5

ffiffiffiffiffiffiffiffiffiffiffiffiffiffib211b2

2

q. For studying seasonal variation in demo-

graphic data, trigonometric regression methods have beendemonstrated to yield more detailed descriptions and superiorresults in some cases compared with alternative methods (Fell-man and Eriksson, 2000). For all datasets of congenital anoma-lies, a likelihood-ratio chi-square test was performed to testwhether dispersion parameter a equals zero (Lawless, 1987).If overdispersion was detected, a negative binomial regressionmodel, rather than Poisson regression model was used.

All analyses were corrected for registry and year ascategorical variables.

ResultsCONGENITAL ANOMALIES

According to the systematic literature review, a total of 26subgroups of congenital anomaly were previously associ-ated with seasonality, of which 14 were also associatedwith influenza (Table 2). An additional eight subgroups ofcongenital anomalies were previously associated withinfluenza, but not with seasonality (Table 3).

Prevalence did not show seasonal fluctuation for allnonchromosomal anomalies combined (peak excess 1.6%,95% confidence interval [CI], 20.3–3.5) or chromosomalanomalies combined (peak excess 1.5%, 95% CI, 22.0–5.1). For anomalies previously associated with influenza, asignificant peak was found for June conceptions (peak2.4%, 95% CI, 0.1–4.7), but not for conceptions with thefirst trimester during the influenza season.

Of the 26 congenital anomaly subgroups previouslyassociated with seasonality (Table 2), we detected season-ality for congenital cataract (127.1%, 95% CI, 7.4–46.8compared with average prevalence during conception peakApril) and hip dislocation and/or dysplasia (112.4%, 95%CI, 3.8–21.0, April). Of the congenital anomalies not previ-ously associated with seasonality (Table 3), we detectedseasonality for all congenital anomalies previously associ-ated with influenza combined (12.4%, 95% CI, 0.1–4.7,June), congenital hydronephrosis (111.8%, 95% CI, 4.9–18.6, July), urinary defects (15.2%, 95% CI, 0.9–9.4, July)and situs inversus (Table 4, 135.6%, 95% CI, 7.2–63.9,December). We did not detect seasonality for other

congenital anomaly subgroups, nor for chromosomal con-genital anomalies. Prevalences of congenital anomaliesassociated with seasonality by month of conception areprovided in Figures 1–6. A large number of congenitalanomalies were not associated with seasonality in Europeduring 2000–2008 including all nonchromosomal congeni-tal anomalies (11.6%, 95% CI, 20.3–3.5), neural tubedefects (15.0%, 95% CI, 22.1–12.1), congenital heartdefects (12.2%, 95% CI, 20.9–5.3) and ventricular septaldefects (13.9%, 95% CI, 20.6–8.4).

Fitting 6-month sine and cosine periods instead of 12-month periods did not improve the fit of any of the mod-els for seasonality.

DiscussionThis study provides an overview of the effect of seasonal-ity on the prevalence of congenital anomalies in Europe.Generally, we found little evidence of seasonality suggest-ing that the main etiological factors for congenital anoma-lies in Europe today are not strongly seasonal in nature.

In this study, we also investigated whether congenitalanomalies previously associated with influenza showedseasonal prevalence peaks corresponding with the influ-enza season. During the 2000 to 2009 influenza seasons,European peak influenza activity generally ranged fromweek 51 in December to week 14 in April (except for2003–2004 season when it ranged from weeks 44 to week11) (EuroFlu, 2013). For most congenital anomalies, thevulnerable period involved the second and third month ofpregnancy and for this reason, we would expect concep-tions of congenital anomalies caused by influenza to peakaround January. Instead, we found a shallow peak in June.Our findings suggest that influenza is not an importantfactor in seasonality of congenital anomalies in Europe.

SEASONALITY OF CONGENITAL ANOMALIES OF PRIOR HYPOTHESIS

We detected previously described seasonality in the preva-lence of congenital cataract in April conceptions. This find-ing confirms a British study (1954–1960 births) (Slateret al., 1964) and an American study (1992–2006 births)(Caton, 2012) and seems to coincide with rubella infec-tions in spring and early summer. This is an interestingfinding because rubella is routinely vaccinated againstexcept in the Ukraine, and diagnosed cases of congenitalrubella syndrome were not included in the study. The pos-sibility that there may have been some undiagnosed con-genital rubella among women who had not beenvaccinated or where vaccination had failed, or that otherinfections causes are implicated, should be furtherfollowed-up. A Danish study (1977–2001 births), did notdetect seasonality in congenital cataract prevalence whenanalyzed by date of birth (Haargaard et al., 2005).

Seasonality in congenital dislocation of the hip with apeak in (late) winter births has been described before inEngland in the 1950s and 1980s and in Israel in the


TABLE 2. Seasonality Analysis Results for Congenital Anomalies Previously Associated with Seasonality.

EUROCAT subgroupCases

(N)

Peak month ofconceptions infitted function

Amplitude fittedsine and cosine

curve in % (95% CI)

Previously associatedwith seasonality

(references)

Previously associatedwith influenza

(Luteijn et al., 2013)

Neural tube defects 3,145 1.8 (Jan) 5.0 (22.1, 12.1) (Bound et al., 1989) Yes

- Anencephalus and similar 1,251 4.5 (Apr) 2.7 (28.7, 14.0) (Bound et at., 1989;

Leck and Record, 1966;

Slater et al., 1964)

Yes

- Spina bifida 1,559 1.3 (Jan) 6.9 (23.2, 17.1) (Bound et at., 1989;


Yes

Hydrocephaly 1,722 8.6 (Aug) 4.7 (24.9, 14.3) (Caton, 2012) No

Anophthalmos/microphthalmos 282 2.5 (Feb) 5.3 (218.5, 29.2) (Caton, 2012) Yes

- Anophthalmos 60 2.9 (Feb) 14.8 (236.9, 66.6) (Caton, 2012) No

Congenital cataract 418 4.0 (Apr) 27.1 (7.4, 46.8) (Caton, 2012;


No

Ventricular septal defecta 9,334 6.9 (Jun) 3.9 (20.6, 8.4) (Bound et at., 1989;

Caton, 2012)

Yes

Atrial septal defect 4,165 9.8 (Sep) 2.4 (24.6, 9.4) (Bound et at., 1989;

Caton, 2012)

No

Ebstein’s anomaly 161 3.3 (Mar) 25.4 (26.4, 57.1) (Caton, 2012) No

Pulmonary valve stenosis 1,460 4.1 (Apr) 4.1 (26.4, 14.6) (Slater et al., 1964) No

Pulmonary valve atresia 316 2.4 (Feb) 9.6 (213.0, 32.1) (Caton, 2012) Yes

Aortic valve atresia/stenosis 401 7.1 (Jul) 17.8 (23.2, 38.8) (Caton, 2012) Yes

Hypoplastic left heart 848 10.8 (Oct) 3.6 (210.1, 17.4) (Eghtesady et al., 2011) No

Coarctation of aorta 1,144 5.2 (May) 5.0 (26.8, 16.9) (Caton, 2012) Yes

Orofacial cleftsa 4,351 4.5 (Apr) 4.2 (21.8, 10.3) (Sandahl, 1977b) Yes

- Cleft lip with or without palate 2,626 5.4 (May) 3.8 (24.0, 11.7) (Sandahl, 1977b) Yes

- Cleft palate 1,725 3.7 (Mar) 6.0 (23.7, 15.6) (Sandahl, 1977b) Yes

Bilateral renal agenesis

including Potter syndrome

416 10.7 (Oct) 12.7 (27.0, 32.3) (Bound et at., 1989;

Caton, 2012)

No

Oesophageal atresia with or without

Trachea-oesophageal fistula

708 6.2 (Jun) 5.5 (29.5, 20.6) (Slater et al., 1964) Yes

Atresia or stenosis of other parts

of the small intestine

295 7.5 (Jul) 2.6 (220.7, 25.9) (Slater et al., 1964) No

Ano-rectal atresia and stenosis 946 7.4 (Jul) 5.4 (27.7, 18.4) (Slater et al., 1964) Yes

Atresia of bile ducts 103 2.4 (Feb) 6.0 (233.4, 45.4) (Caton, 2012) No

Abdominal wall defects 1,793 8.9 (Aug) 1.4 (28.0, 10.9) (Waller et al., 2010) Yes

Lower limb reduction 622 1.3 (Jan) 2.3 (213.7, 18.3) (Caton, 2012;


No

Hip dislocation and/or dysplasiaa 2,635 4.2 (Apr) 12.4 (3.8, 21.0) (Anand et al., 1992;

Chen et al., 1970;

Edwards, 1961; Heikkila, 1984;

Record and Edwards, 1958;


No

aAnalyzed by negative binomial regression model due to overdispersion.


1960s (Record and Edwards, 1958; Edwards, 1961; Slateret al., 1964; Chen et al., 1970; Anand et al., 1992) and thiscorresponds with April conceptions as found in our study.A Finnish study detected a seasonal peak in June and July,in female births (Heikkila, 1984). Seasonality in congenitaldislocation of the hip was not detected in 1956 to 1964British births (Bound et al., 1989) nor in 1978 to 2001Atlanta births (Siffel et al., 2005). Overall, the evidence fora seasonal peak in congenital dislocation of the hip in latewinter births seems strong due to the number of studiesreporting similar findings (Record and Edwards, 1958;Edwards, 1961; Slater et al., 1964; Chen et al., 1970;Anand et al., 1992). A recent meta-analysis reported sev-eral risk factors for congenital dislocation of the hip, noneof them being seasonal. Because congenital dislocation ofthe hip is associated with high birth weight (Chan et al.,1997), some seasonality in preterm birth or low birthweight could influence it. However, there is little literatureon seasonality of birth weight. One study from London

suggests a winter peak in preterm births (Lee et al.,2006), coinciding with the winter peak in congenital dislo-cation of the hip births and therefore not supporting sucha hypothesis.

The seasonal peaks we detected for congenital anoma-lies previously associated with influenza (congenitalhydronephrosis, all influenza-associated anomalies com-bined) did not correspond with the influenza season as wewould have expected them to peak around January con-ceptions. Of all the congenital anomalies previously associ-ated with influenza, we only detected seasonal peakscorresponding with influenza season for neural tubedefects (January), spina bifida (January), encephalocele(January), and diaphragmatic hernia (January), but none ofthese associations reached statistical significance. We con-clude influenza is not an important factor in seasonality ofcongenital anomalies in Europe.

A British study (1957–1981 births) detected a seasonalconception peak in June for ventricular septal defects

TABLE 3. Seasonality Analysis Results for Congenital Anomalies Previously Associated with Influenza (Luteijn et al., 2013), but not Seasonality.

EUROCAT group Cases (N)Peak conceptions in fitted function (month)

Amplitude fitted sine andcosine curve in % (95% CI)

All congenital anomalies previously associated with influenzaa 36,563 6.5 (Jun) 2.4 (0.1, 4.7)

All congenital anomalies previously associated

with influenza w/o congenital heart defectsa

17,736 6.5 (Jun) 2.4 (20.7, 5.6)

Encephalocele 335 1.7 (Jan) 14.4 (27.5, 36.3)

Congenital heart defectsa 20,201 6.2 (Jun) 2.2 (20.9, 5.3)

- Tricuspid atresia and stenosis 208 4.1 (Apr) 17.4 (210.4, 45.2)

Diaphragmatic hernia 865 1.1 (Jan) 6.4 (27.2, 20.0)

Omphalocele 683 8.1 (Aug) 6.3 (29.0, 21.6)

Congenital hydronephrosisa 4,041 7.4 (Jul) 11.8 (4.9, 18.6)

Upper limb reduction 1,173 5.5 (May) 0.4 (211.3, 12.1)

Syndactyly 1,438 8.3 (Aug) 3.8 (26.8, 14.3)

aAnalyzed by negative binomial regression model due to overdispersion.

TABLE 4. Seasonality Analysis Results for Congenital Anomalies not of Prior Hypothesis.

EUROCAT group Cases (N)Peak conceptions in fitted

function (month)Amplitude fitted sine and cosine

curve in % (95% CI)

All nonchromosomal anomaliesa 65,764 5.6 (May) 1.6 (20.3, 3.5)

All nonchromosomal anomalies

(excluding prior hypothesis for influenza)

29,201 3.8 (Mar) 1.5 (21.3, 4.3)

All chromosomal anomalies 12,682 10.3 (Oct) 1.5 (22.0, 5.1)

Urinary 10,314 7.8 (Jul) 5.2 (0.9-9.4)

Situs inversus 206 12.5 (Dec) 35.6 (7.2, 63.9)

aAnalysed by negative binomial regression model due to overdispersion.


(Bound et al., 1989). An American study (1992–2006births) detected a conceptional peak for ventricular septaldefects in March (Caton, 2012). We failed to detect season-ality in the prevalence of ventricular septal defects anddid not find seasonality in congenital heart defect preva-lence or any congenital heart defects subgroups.

We did not detect seasonality in neural tube defects.Nevertheless, it is interesting to note the nonsignificantseasonal peak for conceptions resulting in neural tubedefects of late January means these pregnancies were inthe third and fourth week of pregnancy, which is the criti-cal period for neural tube defects, during peak influenzaactivity in Europe (EuroFlu, 2013). Neural tube defects

have commonly been associated with seasonality (Leckand Record, 1966; Maclean and MacLeod, 1984; Ncayiyana,1986; Bound et al., 1989; de la Vega and Lopez-Cepero,2009) in the past, hypothesized to be due to dietary andother factors, although some studies did not find seasonal-ity (Naggan, 1969; Elwood and Nevin, 1973; Jorde et al.,1984; Castilla et al., 1990). Some studies have detectedseasonality in anencephaly but no seasonality in spinabifida (Fraser et al., 1986; Castilla et al., 1990). Neuraltube defects also have been associated with influenza(Luteijn et al., 2013), but no observations of excesses ininfluenza season have been made. We suggest that for

FIGURE 1. Prevalence per 10,000 of any nonchromosomal congenital anoma-

lies per month of conception in 20 EUROCAT registry populations, 2000 to

2008.

FIGURE 2. Prevalence per 10,000 of congenital cataract per month of concep-

tion in 20 EUROCAT registry populations, 2000 to 2008.

FIGURE 3. Prevalence per 10,000 of hip dislocation and or dysplasia per

month of conception in 20 EUROCAT registry populations, 2000 to 2008.

FIGURE 4. Prevalence per 10,000 of congenital anomalies previously associ-

ated with influenza per month of conception in 20 EUROCAT registry popula-

tions, 2000 to 2008.


spina bifida, the high amplitude of the variation (6.9, 23.2,17.1), the previous literature finding a relationshipbetween first trimester influenza exposure and spinabifida (Luteijn et al., 2013) and the generally acceptedrelationship between hyperthermia and central nervoussystem defects (Moretti et al., 2005; Edwards, 2006), seemsuggestive of a possible association between in utero influ-enza exposure and spina bifida.

SEASONALITY OF CONGENITAL ANOMALIES NOT OF PRIOR HYPOTHESIS

After testing the remaining EUROCAT congenital anomalysubgroups with �100 registrations, we detected highlysignificant seasonality for situs inversus with a seasonalpeak in early December conceptions. This seasonal peakcoincides with the European 2000 to 2008 influenza sea-sons. Unfortunately, no literature exists on situs inversusand seasonality or influenza, or indeed its epidemiology ingeneral, probably due to the rare nature of this congenitalanomaly. We also detected seasonality for urinary defectsand this seasonality is driven by congenital hydronephro-sis which makes up roughly 40% of urinary defects.

STRENGTHS AND WEAKNESSES OF THE STUDY

The strengths of our study include analysis of seasonalityof congenital anomalies by date of conception, whichallows for more accurate analysis because gestational ageis heavily influenced by early termination of pregnancy(terminations of pregnancy for fetal anomaly) and averagegestational ages of some congenital anomalies have beendemonstrated to differ from unaffected pregnancies (San-dahl, 1977a). Other strengths are related to the datasource. Congenital anomaly registries have high case ascer-tainment from multiple sources, consistent coding, inclu-sion of all pregnancy outcomes (including stillbirths andterminations of pregnancy for fetal anomaly). Additionally,

EUROCAT registries are population-based, meaning thatcongenital anomalies enrolled are representative of theirpopulations. Our study population was very large, asneeded for rare anomalies.

Our study is limited by the fact that the statisticalmodel used assumes a perfect sine curve, while observedlocal peaks disturb the perfect sine curve and thereforecan distort the fitted seasonal function toward the localpeak. This might have led to overdispersion in some analy-ses. Another limitation is that, due to the large amount oftests performed, some associations could be due tochance, particularly those of borderline significance. It ispossible that by combining a large area of Europe, weobscured seasonality in smaller geographic areas. This isdifficult to investigate without specific hypotheses, becausedividing up Europe results in much more multiple testingof small datasets with a high chance of spuriously signifi-cant results. However, we are conducting a study of con-genital anomaly prevalence specifically in relation torecorded influenza attack rates across Europe and the lackof generalized seasonality we have found in this study willhelp in the interpretation of studies of specific time vary-ing risk factors.

AcknowledgmentWe thank Lolkje de Jong-van den Berg for her helpfulcomments.

References

Anand JK, Moden I, Myles JW. 1992. Incidence of neonatal hipinstability: are there seasonal variations? Acta Orthop Belg 58:205–208.

FIGURE 5. Prevalence per 10,000 of congenital hydronephrosis per month of

conception in 20 EUROCAT registry populations, 2000 to 2008.

FIGURE 6. Prevalence per 10,000 of situs inversus per month of conception in

20 EUROCAT registry populations, 2000 to 2008.


Bound JP, Harvey PW, Francis BJ. 1989. Seasonal prevalenceof major congenital malformations in the Fylde of Lanca-shire 1957–1981. J Epidemiol Community Health 43:330–342.

Castilla EE, Orioli IM, Lugarinho R, et al. 1990. Monthly and sea-sonal variations in the frequency of congenital anomalies. Int JEpidemiol 19:399–404.

Caton AR. 2012. Exploring the seasonality of birth defects in theNew York State Congenital Malformations Registry. Birth DefectsRes A Clin Mol Teratol 94:424–437.

Chan A, McCaul KA, Cundy PJ, et al. 1997. Perinatal risk factorsfor developmental dysplasia of the hip. Arch Dis Child Fetal Neo-natal Ed 76:F94–F100.

Chen R, Weissman SL, Salama R, Klingberg MA. 1970. Congenitaldislocation of the hip (CDH) and seasonality: the gestational age ofvulnerability to some seasonal factor. Am J Epidemiol 92:287–293.

de la Vega A, Lopez-Cepero R. 2009. Seasonal variations in theincidence of some congenital anomalies in Puerto Rico based onthe timing of conception. P R Health Sci J 28:121–125.

Edwards JH. 1961. Seasonal incidence of congenital disease inBirmingham. Ann Hum Genet 25:89–93.

Edwards MJ. 2006. Review: Hyperthermia and fever during preg-nancy. Birth Defects Res A Clin Mol Teratol 76:507–516.

Eghtesady P, Brar A, Hall M. 2011. Seasonality of hypoplastic leftheart syndrome in the United States: a 10-year time-series analy-sis. J Thorac Cardiovasc Surg. 141:432–438.

Elwood JH, Nevin NC. 1973. Factors associated with anencepha-lus and spina bifida in Belfast. Br J Prev Soc Med 27:73–80.

EUROCAT Central Registry. 2009. EUROCAT Guide 1.3 and refer-ence documents. Instructions for the Registration and Surveil-lance of Congenital Anomalies. Revised: August 2013. Availableat: http://www.eurocat-network.eu/content/EUROCAT-Guide-1.3.pdf. Accessed January 8, 2014.

EuroFlu. 2013. EuroFlu Electronic Bulletin. Available at: https://www.euroflu.org. Accessed on February 26, 2014.

Fellman J, Eriksson AW. 2000. Statistical analysis of the seasonalvariation in demographic data. Hum Biol 72:851–876.

Fraser FC, Frecker M, Allderdice P. 1986. Seasonal variation ofneural tube defects in Newfoundland and elsewhere. Teratology33:299–303.

Greenlees R, Neville A, Addor MC, et al. 2011. Paper 6: EUROCATmember registries: organization and activities. Birth Defects ResA Clin Mol Teratol 91(Suppl 1):S51–S100.

Haargaard B, Wohlfahrt J, Rosenberg T, et al. 2005. Risk factorsfor idiopathic congenital/infantile cataract. Invest Ophthalmol VisSci 46:3067–3073.

Heikkila E. 1984. Congenital dislocation of the hip in Finland. An epi-demiologic analysis of 1035 cases. Acta Orthop Scand 55:125–129.

Jorde LB, Fineman RM, Martin RA. 1984. Epidemiology of neuraltube defects in Utah, 1940–1979. Am J Epidemiol 119:487–495.

Kallan JE, Enneking EA. 1992. Seasonal patterns of spontaneousabortion. J Biosoc Sci 24:71–75.

Lawless JF. 1987. Negative binomial and mixed Poisson regres-sion. Can J Stat 15:209–225.

Leck I, Record RG. 1966. Seasonal incidence of anencephalus. BrJ Prev Soc Med 20:67–75.

Lee SJ, Steer PJ, Filippi V. 2006. Seasonal patterns and pretermbirth: a systematic review of the literature and an analysis in aLondon-based cohort. BJOG 113:1280–1288.

Luteijn JM, Brown MJ, Dolk H. 2013. Influenza and congenitalanomalies: a systematic review and meta-analysis. Hum Reprod[Epub ahead of print].

Maclean MH, MacLeod A. 1984. Seasonal variation in the fre-quency of anencephalus and spina bifida births in the UnitedKingdom. J Epidemiol Community Health 38:99–102.

McDonald AD. 1971. Seasonal distribution of abortions. Br J PrevSoc Med 25:222–224.

McKeown T, Record RG. 1951. Seasonal incidence of congenitalmalformations of the central nervous system. Lancet 1:192–196.

Moretti ME, Bar-Oz B, Fried S, Koren G. 2005. Maternal hyper-thermia and the risk for neural tube defects in offspring: system-atic review and meta-analysis. Epidemiology 16:216–219.

Naggan L. 1969. The recent decline in prevalence of anencephalyand spina bifida. Am J Epidemiol 89:154–160.

Ncayiyana DJ. 1986. Neural tube defects among rural blacks in aTranskei district. A preliminary report and analysis. S Afr Med J69:618–620.

Office for National Statistics licensed under the Open GovernmentLicence v.1.0. 2011. Characteristics of birth 2, England and Wales,2010. Available at: http://www.ons.gov.uk/ons/rel/vsob1/charac-teristics-of-birth-2–england-and-wales/2010/index.html. AccessedJanuary 8, 2014.

Record RG, Edwards JH. 1958. Environmental influences relatedto the aetiology of congenital dislocation of the hip. Br J PrevSoc Med 12:8–22.

Sandahl B. 1974. A study of seasonal and secular trends in inci-dence of stillbirths and spontaneous abortions in Sweden. ActaObstet Gynecol Scand 53:251–257.

Sandahl B. 1977a. Seasonal incidence of some congenital malfor-mations in the central nervous system in Sweden, 1965–1972.Acta Paediatr Scand 66:65–72.


Sandahl B. 1977b. Seasonal incidence of cleft lips and cleftpalates in Sweden, 1965–1974. Scand J Plast Reconstr Surg 11:39–43.

Siffel C, Alverson CJ, Correa A. 2005. Analysis of seasonal varia-tion of birth defects in Atlanta. Birth Defects Res A Clin Mol Ter-atol 73:655–662.

Slater BC, Watson GI, McDonald JC. 1964. Seasonal variation incongenital abnormalities. Preliminary report of a survey con-ducted by the Research Committee of Council of the College ofGeneral Practitioners. Br J Prev Soc Med 18:1–7.

Statistics Denmark. Statistics Denmark – StatBank DenmarkBEV3A: Live births and deaths by months. Available at: http://www.statistikbanken.dk/statbank5a/SelectVarVal/Define.asp?Maintable5BEV3A&PLanguage51. Accessed January 8, 2014.

Stolwijk AM, Straatman H, Zielhuis GA. 1999. Studying seasonal-ity by using sine and cosine functions in regression analysis. JEpidemiol Community Health 53:235–238.

Waller SA, Paul K, Peterson SE, Hitti JE. 2010. Agricultural-relatedchemical exposures, season of conception, and risk of gastroschi-sis in Washington State. Am J Obstet Gynecol 202:241.e1–241.e6.


seasonality of congenital anomalies in europe

Documents