IJE Advance Access published online on May 8, 2008
International Journal of Epidemiology, doi:10.1093/ije/dyn083
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Space-time analysis of Down syndrome: results consistent with transient pre-disposing contagious agent
1 School of Clinical Medical Sciences (Child Health), Newcastle University, Newcastle upon Tyne, UK.
2 Institute of Health and Society, Newcastle University, Newcastle upon Tyne, UK.
3 Regional Maternity Survey Office, Newcastle upon Tyne, UK.
4 Institute for Research on Environment and Sustainability, Newcastle University, Newcastle upon Tyne, UK.
* Corresponding author. Sir James Spence Institute, Newcastle University, Royal Victoria Infirmary, Queen Victoria Road, Newcastle upon Tyne NE1 4LP, UK. E-mail: Richard.McNally{at}ncl.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Background Whilst maternal age is an established risk factor for Patau syndrome (trisomy 13), Edwards syndrome (trisomy 18) and Down syndrome (trisomy 21), the aetiology and contribution of genetic and environmental factors remains unclear. We analysed for space-time clustering using high quality fully population-based data from a geographically defined region.
Methods The study included all cases of Patau, Edwards and Down syndrome, delivered during 1985–2003 and resident in the former Northern Region of England, including terminations of pregnancy for fetal anomaly. We applied the K-function test for space-time clustering with fixed thresholds of close in space and time using residential addresses at time of delivery. The Knox test was used to indicate the range over which the clustering effect occurred. Tests were repeated using nearest neighbour (NN) thresholds to adjust for variable population density.
Results The study analysed 116 cases of Patau syndrome, 240 cases of Edwards syndrome and 1084 cases of Down syndrome. There was evidence of space-time clustering for Down syndrome (fixed threshold of close in space: P = 0.01, NN threshold: P = 0.02), but little or no clustering for Patau (P = 0.57, P = 0.19) or Edwards (P = 0.37, P = 0.06) syndromes. Clustering of Down syndrome was associated with cases from more densely populated areas and evidence of clustering persisted when cases were restricted to maternal age <40 years.
Conclusions The highly novel space-time clustering for Down syndrome suggests an aetiological role for transient environmental factors, such as infections.
Keywords Congenital abnormality, Down syndrome, Edwards syndrome, Patau syndrome, aetiology, environment, space-time clustering
Accepted 7 April 2008
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
The exact aetiology of trisomy 13 (Patau syndrome), trisomy 18 (Edwards syndrome) and trisomy 21 (Down syndrome) is not known. Both genetic and environmental factors are likely to contribute.1,2 Down syndrome arises from non-disjunction, usually in formation of the eggs or sperm, where a gamete ends up with an extra chromosome 21. Non-disjunction may occur in the first or second stage of meiosis. Patau and Edwards syndromes also arise from non-disjunction during meiosis, resulting in an extra chromosome in one of the gametes (egg or sperm). The only established risk factor is advanced maternal age which is associated with increased risk of all three trisomies in the offspring.1–6 A number of other factors have been suggested to increase the risk of Down syndrome including higher socio-economic status, oral contraception use, maternal smoking, ionizing radiation, paternal occupation of janitor, mechanic and farm manager/worker and residential proximity to hazardous waste sites.7–11 However, consistency between studies has been poor. Studies that have investigated the possible association between lifestyle and environmental factors (such as living near incinerators or landfill sites, exposure to chlorination by-products, exposure to pesticides or solvents) in the risk of Patau and Edwards syndromes, have not led to definitive evidence.12–16
Investigations of anecdotal geographical excesses of Down syndrome have failed to provide further insight concerning aetiology.17,18 In the 1960s there was speculation that viral hepatitis and influenza may be involved either around the time of conception or during pregnancy.19–21 However, there have been no recent studies examining this hypothesis further. If infections or similar spatially and temporally varying environmental exposures are contributing to the aetiology of these conditions, then the distribution of cases can be expected to exhibit space-time clustering. More specifically, transient occurrences of a relevant aetiological exposure may be predicted at a number of different geographical locations. Two previous studies, using less advanced statistical methodology, did not report evidence for space-time clustering amongst cases of Down syndrome.22,23
Space-time clustering occurs when excess numbers of cases are observed within small geographical areas for short periods of time. The aims of our study were: (i) to test predictions of space-time clustering which could arise from an environmental exposure occurring in-utero; and (ii) to test for differences due to residential population density.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
All cases of Patau, Edwards and Down syndrome, delivered during the period January 1, 1985 to December 31, 2003 were abstracted from a high quality fully population-based registry, the Northern Congenital Abnormality Survey (NorCAS), which covers the former Northern National Health Service (NHS) Health Region of England (as defined by 1972 NHS boundaries). NorCAS collects data on all congenital anomalies resident in this region whether diagnosed ante-natally or not. Notification of cases is made using multiple sources to ensure high case ascertainment. All cases occurring in miscarriages
20 weeks, termination of pregnancy for fetal anomaly following pre-natal diagnosis, stillbirths and livebirths were included if delivered during the study years. Post-natal diagnostic confirmation by cytogenetic examination is recorded for all cases. The study area comprised a geographically defined region of Northern England with a population of 2.6 million and an average of 30 000 births per year, residing in two main urban conurbations and extended rural areas. Further details concerning the method for data collection have been described previously.24 There were low levels of migration into or out of the region.25,26 An analysis of migration showed that only 9% of mothers notified to NorCAS had changed residence between the time of booking and delivery. Furthermore, most of these moves were local (J. Rankin, personal communication). Markedly higher migration rates have been reported from Canada and the USA, where it has been estimated that around 12 and 25%, respectively of mothers moved during pregnancy.27,28
Ordnance Survey (OS) four-digit grid references (Easting and Northing) were allocated to each case with respect to the centroid of the postcode of the residential address at time of delivery, locating each address to within 0.1 km. In the UK there are 
1.7 million postcodes. These are unique alphanumeric geographical identifiers for delivery of mail that may cover a number of residential addresses (typically around 15–20 houses), a smaller number of multiple occupancy dwellings or a single commercial address.29 There are 112 999 postcodes in the study area.
The following aetiological hypotheses were tested: (i) a primary factor influencing geographical or temporal heterogeneity of prevalence of the anomaly is related to exposure to an infection or other similar occurring environmental agent either in-utero or pre-conception and (ii) geographical or temporal heterogeneity of prevalence of the anomaly is modulated by differences in patterns of exposure related to level of population density.
Statistical analyses
The analyses are based on K-functions,30 which is a generalization of the Knox test.31 In the Knox test, a pair of cases is regarded as being in ‘close proximity’ if their dates of delivery are close and their residential addresses at time of delivery are close. The number of pairs of cases observed to be in close proximity is obtained (O) and the number of pairs of cases expected to be in close proximity is calculated (E). If O exceeds E there is evidence of space-time clustering. The magnitude of the excess (or deficit) is estimated by calculating strength S = [(O–E)/E] x 100.
There are two problems with the Knox test. Boundary problems may occur since it can be less likely (or sometimes impossible) for some cases to be close in one dimension to other cases. Another problem is the arbitrary choice of thresholds. Use of many thresholds would result in multiple testing. A simplification of a second-order procedure based on K-functions was used to partially circumvent this arbitrary choice of thresholds which is implicit in the Knox test and also to avoid multiple testing.30 This procedure involved a set of 225 Knox-type calculations, where the boundaries changed over a pre-specified set of values (for close times, t = 0.1, 0.2, ..., 1.5 years and for close in space, s = 0.5, 1, 1.5, ..., 7.5 km). The observed value of the K-function was calculated using the splancs package in R. The unknown distribution of the K-function was simulated. At each simulation, the dates of delivery were randomly re-allocated to each of the cases in the analysis and a realization of the K-function was obtained from these simulated data. This was repeated for a total of 999 random permutations of time. Statistical significance was assessed by comparing the observed value with the simulated distribution. The K-function method does not give an estimate of the strength of the clustering effect. S, derived from the Knox test was calculated for the same ranges of critical values (for time 0.1, ..., 1.5 year and for space 0.5, ..., 7.5 km) and was used to indicate the close times and distances for which clustering was strongest.
To adjust for the effect of different population densities, the K-function analyses were repeated using a nearest neighbour (NN) approach. The delivery locations of all the cases in the complete data set (that comprised all cases of Patau, Edwards and Down syndrome from the defined geographical region of Northern England) were used to determine NNs. By inspection, the mean distance to the 21st NN was found to be 5 km. Fixed critical distances (0.5, ..., 7.5 km) were replaced by variable critical distances to the 14th, ..., 28th NNs. It should be noted that space-time clustering based on fixed distance thresholds provides support for the role of geostationary exposures in aetiology, whilst clustering based on variable NN thresholds is more indicative for the role of an agent that is transmissible between individuals.
The distribution of distances between the 21st NNs was highly skewed, with distances varying from 1.0 to 58.6 km. The median distance between the 21st NNs was 3.1 km. In order to test whether population density was associated with space-time clustering, cases were split into two groups: 50% were classified as belonging to a ‘more densely populated’ group and 50% were classified as belonging to a ‘less densely populated’ group on the basis of whether the 21st NN was nearer or further away than the median distance (3.1 km) of the 21st NN. It has been estimated that there are at least 23.5 deliveries per square kilometre per year in ‘more densely populated’ areas and at most 23.5 deliveries per square kilometre per year in ‘less densely populated’ areas. Analysis by population density was then done by considering clustering pairs that included at least one case from the ‘more densely populated’ category and clustering pairs that included at least one case from the ‘less densely populated’ category. It should be stressed that these analyses of population density (particularly analyses of ‘less densely populated: any clustering pairs’) may be subject to a diluting influence from edge effects because ‘less densely populated’ areas do not form a single contiguous zone.
Space-time interactions based on time and place of delivery were tested. Statistical significance was assessed using one-sided tests.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
The study analysed 116 cases of Patau syndrome, 240 cases of Edwards syndrome and 1084 cases of Down syndrome. Twins have been excluded from all analyses. Dates of delivery were 100% complete. However, 10 cases (0.7%) were excluded from the analyses because they had missing postcodes. Table 1 presents the total number of cases for each anomaly, the numbers of cases from ‘more’ and ‘less’ densely populated areas and the numbers of cases with ‘younger’ (aged <40 years) and ‘older’ (aged
40 years) mothers.
|
Table 2 shows that for Down syndrome there was evidence of space-time clustering (P = 0.01 and P = 0.02 using the geographical distance and the NN threshold versions of the K-function method). Additional analyses using the Knox test (Table 3) found that the greatest clustering effect generally occurred with thresholds of 0.5–1 km and 0.2–0.4 year (2–5 months). There were only 7 out of 225 space-time combinations with negative values of S. Furthermore, space-time clustering for Down syndrome was still present when maternal age at delivery of the index child was restricted to under 40 years of age (P = 0.04 and P = 0.02 using the geographical distance and NN threshold versions of the K-function method, Table 4). Additional analyses of Edwards and Patau syndromes were also carried out using the Knox test. There were 128 out of 225 space-time combinations with negative values of S for Edwards syndrome and 94 out of 225 for Patau syndrome. Thus, in spite of the smaller numbers of cases, there was little or no evidence for space-time clustering amongst cases of Patau and Edwards syndromes.
|
|
|
Analyses by population density (Table 5) showed that clustering was only present for pairs of cases of Down syndrome that included at least one from a ‘more densely populated area’ and was not seen for pairs of cases that included at least one from a ‘less densely populated area’ (‘more densely populated: any’ pairs: P = 0.02 and P = 0.01, using the geographical distance and NN threshold versions of the K-function method; ‘less densely populated: any’ pairs: P = 0.13, P = 0.09). There was limited evidence of space-time clustering for cases of Edwards syndrome involving cases from less densely populated areas (P = 0.07, P = 0.04 using the geographical distance and NN threshold versions of the K-function method), but not from more densely populated areas. Similarly, there was also very limited evidence of space-time clustering for cases of Patau syndrome from less densely populated areas (P = 0.37, P = 0.01 using the geographical distance and NN threshold versions of the K-function method).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
This study has found evidence of space-time clustering amongst cases of Down syndrome. The clustering was particularly marked for pairs of cases who were delivered within short time periods and was only found for cases from more densely populated areas. The analyses have been carried out using modern systematic statistical methods on high quality population-based data. Data collection methods have remained constant over the 19 years and include annual cross validation with cytogenetics laboratories in the region that has ensured a very high level of case ascertainment.
This new study has a number of strengths. The high quality, population-based data are on individual cases from a geographically well-defined area with very low migration rates. Multiple data sources were used to ascertain cases. Cases occurring in late miscarriages and termination of pregnancy for fetal anomaly are included in the analyses. The statistical methods allow the inclusion of different types of threshold to differentiate between aetiological hypotheses relating to fixed geographical exposures (e.g. agents originating from point sources) and much more mobile exposures (e.g. agents transmitted from person to person). The K-function method gives no measure of the clustering effect. Strength of clustering (S), derived from the Knox test, gives an estimate of the excess numbers of pairs of cases that cluster in space-time. However, some cases may have more than one close space-time partner. Therefore, the estimates of S should be regarded as upper limits on the proportion of cases attributable to the space-time interaction.
The finding of space-time clustering amongst cases of Down syndrome suggests that relevant aetiological agents may also cluster in space-time. The types of aetiological agent involved are likely to be infectious. It remains possible that they are non-infectious, but it would be unlikely for such agents to cluster in space-time. It would be less unusual for possible sources of bias to cluster in space-time. One potential bias may occur if changes in the ability to identify cases migrated throughout the study region unevenly in space and time. Since the NorCAS is population-based, with high levels of ascertainment and rigorous data collection procedures, space-time clustering cannot be explained by this type of spatio-temporal heterogeneity in detection rates. Another potential bias may result if high risk mothers migrated into different geographical areas unevenly in space and time. Since thresholds for both close in space and close in time are relatively small, it is possible, but highly unlikely that people at high risk have migrated into a number of different localized areas unevenly in space and time.
We must also acknowledge that the absence of a space-time clustering effect does not provide strong evidence against an aetiology that involves agents that cluster in space-time. Misclassification could dilute any clustering. A possible source of misclassification is geographical location. In these analyses we have used residential address at delivery as a proxy for the location of the putative exposure. The adequacy of this address as a proxy for the location of an aetiological agent (such as an infection) is not known.
Migration of cases may also weaken the space-time clustering effect. Additional sensitivity analyses were performed to investigate the potential influence of migration on results. The effect of migration of cases out of the study region was studied by deletion of samples from the complete data set (5 and 10% of the total number of cases were chosen using two different sampling schemes). Migration of 5% had little effect on the overall conclusions. However, migration of 10% tended to bias the results towards a less pronounced clustering effect. Thus, it is possible that results may have been diluted by 9% of mothers changing residence between exposure to an unknown environmental agent and delivery of the child.
There is another potential limitation of the methodology used here relating to bias that may be introduced by shifts in the underlying population.32,33 This can happen when the population grows or declines at different rates in different parts of the study region. Such population shifts may either strengthen or weaken any space-time clustering effect. We have evidence that there is little population movement across the region used in this study, and so the analyses will not have been biased by such perturbations.25,26
The methods used in our study are appropriate for examining overall space-time patterning in disease data. Space-time clustering differs from spatial clustering because it seeks patterns in time and space and as such any observed patterning points to aetiological factors that are highly temporally heterogeneous. Infections are an example of an agent that may exhibit such a pattern (conversely, spatial clustering is associated with static aetiological factors). The analyses cannot be used to identify isolated clusters; furthermore, studies of individual clusters are more limited in eliciting aetiological clues.34,35 Rothman34 has argued that space-time clustering analyses are not as useful for testing aetiological hypotheses as studies that incorporate individual level exposure assessment. We suggest that space-time clustering studies of pre-existing databases such as the one used here offer a resource-effective way of generating hypotheses which may be further tested by subsequent prospective studies. Similar analyses showed that cases of cervical cancer exhibited space-time clustering and subsequent research arising from this demonstrated that human papilloma virus was the causative agent.36,37
Despite the potential methodological limitation, space-time clustering has been specifically identified amongst cases of Down syndrome. The limited evidence of clustering for Patau and Edwards syndromes may well be spurious. It is possible that the lack of space-time clustering for Patau and Edwards syndromes may also be explained by low statistical power. However, in spite of the small numbers of cases, the analyses of strength of clustering indicated little or no overall positive evidence of space-time clustering. The space-time clustering found amongst cases of Down syndrome cannot be explained by variations in population density, since it was present using both the geographical distance and NN threshold methods.
There are a number of environmental agents that may cause localized variations in prevalence. These include airborne infections and pollutants. However, the finding of space-time clustering (for Down syndrome) with respect to time and maternal residence indicates a role for a transient aetiological agent and thus is not compatible with prolonged exposure to environmental contaminants either geographically or over time, which has been suggested as a possible mechanism for the failure of meiosis in oocytes that leads to the condition.2 Hence, our results do not suggest a role of long-term pollutant exposure. The results are however consistent with an environmental exposure occurring shortly before or after conception.
Several earlier studies hinted at the possible involvement of infectious agents in the aetiology of Down syndrome.19–21 Two older studies applied formal statistical methods to population-based data on the prevalence of Down syndrome. The first of the studies analysed 2529 cases of children with Down syndrome, born during 1950–64, from Michigan (US) and used the Ederer-Myers-Mantel method.38 Cases were considered to be close in space if their birth addresses were located in the same county. However, a number of different temporal thresholds were considered. Whilst some combinations showed nominal statistical significance, they concluded that overall there was no evidence of space-time clustering.22 The second study analysed around 8000 cases, occurring between 1989 and 1995, from the National Down Syndrome Cytogenetic Register (England and Wales) and used the Knox method. Cases were in close proximity if their birth addresses were located in the same Regional Health Authority and they occurred together in consecutive months.23 There was no evidence of space-time clustering. There are several explanations for the apparent differences between our current and these older studies. First, there are methodological differences and secondly there are differences in scale. The older studies relied on testing at specific thresholds whilst the use of the K-function method in the current study alleviates this problem. Also, the older studies used much larger spatial (areal) units to assess closeness in space that might not be sufficiently fine scaled to reflect an infectious aetiology. Both older studies used fixed geographical distance metrics and much larger geographical scales which could not uncover the effects of person-to-person transmission of any putative infection.
The proportion of pregnancies with congenital anomalies resulting in a termination has increased during recent years as a result of increased uptake of pre-natal screening corresponding with a decrease in the proportion of live births with congenital anomalies.4,39 Whilst this trend may show geographical variation, it would be unlikely to explain clustering in space-time. The only established risk factor for Down syndrome is maternal age.1–5 Fetal loss rates have been shown to increase with maternal age for Down syndrome pregnancies.40 It may be that the distribution of residences of older mothers is geographically patterned because of socio-economic patterning in the maternal age distribution. However, this would be predicted to lead to a geographical excess in clustering and not to clustering in space-time. Additionally, our results specifically demonstrate that space-time clustering was present for cases of younger mothers (aged <40 years, which was chosen as a cut-off for younger mothers because the probability of a Down syndrome birth is <0.01 in mothers aged under 40 years and is >0.01 in mothers aged over 40 years).41,42 Similarly, the other putative environmental risk factors, such as oral contraceptive use and maternal smoking are likely to display spatial heterogeneity, but not space-time clustering.
Our findings (for Down syndrome) are more consistent with an aetiological agent that exhibits a temporary presence at many points in time and space. An example of such an agent would be an airborne infection. It should also be noted that the significance of the NN threshold method provides more support for an aetiological agent that is spread by person-to-person transmission, rather than a geographically fixed source. Furthermore, the restriction of the finding to cases from ‘more densely populated areas’ is consistent with enhanced opportunity for close contact between individuals. An infectious aetiology is further supported by the fact that the strongest clustering mostly occurred at critical spatio-temporal thresholds of 0.5–1.0 km and 0.2–0.4 year, which is consistent with person-to-person transmission of an aetiological agent arising from a number of common types of social mixing. The meeting places could include various settings such as nurseries, schools and GP surgeries. The optimal close temporal periods (0.2–0.4 year) would be entirely consistent with an infectious aetiology. Indeed, one study showed that mothers of children with Down syndrome had higher herpes simplex virus type 2 antibody levels.43
Our rigorous analysis has shown evidence of space-time clustering using several approaches and is especially supportive of an aetiological agent associated with person-to-person transmission. It must be stressed that the putative agent (that may be an infection) would only result in Down syndrome in a small number of susceptible cases, who also have other (as yet unknown) contributing aetiological factors. The majority of other non-susceptible individuals would not develop the condition.
In conclusion, this study is the first to find marked evidence of space-time clustering amongst cases of Down syndrome suggestive of contributions of an infectious agent. The results are consistent with factors acting early in pre-natal development or pre-conception. The finding of space-time clustering of Down syndrome using high quality population-based registry data should be confirmed by using prospective designs involving individual exposure assessment methodologies. More specifically, hypotheses concerning maternal infections that occurred up to 1 year before conception and in the first trimester of the index pregnancy should be investigated.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
We thank the anonymous referees for their most helpful and constructive comments on an earlier version of this article. We are also grateful to all the Link Clinicians in the Northern Region for their continued support of the NorCAS and to Mary Bythell, NorCAS data manager. NorCAS has section 60 Class Support under the UK Health and Social Care Act (2001) for the collection of personal information without consent, and ethics approval (04/MRE04/25) to undertake studies involving the use of its data.
NorCAS is funded by the Department of Health Policy Research Programme (Disease Register) and Dr Judith Rankin is funded by a Personal Award Scheme Career Scientist Award from the National Institute of Health Research (UK Department of Health).
Conflict of interest: None declared.
KEY MESSAGES
|
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
1 Hassold TJ, Jacobs PA. Trisomy in man. Annu Rev Genet (1984) 18:69–97.[CrossRef][Web of Science][Medline]
2 Sherman SL, Freeman SB, Allen EG, Lamb NE. Risk factors for nondisjunction of trisomy 21. Cytogenetic Genome Res (2005) 111:273–80.[CrossRef]
3 Carothers AD, Hecht CA, Hook EB. International variation in reported livebirth prevalence rates of Down syndrome, adjusted for maternal age. J Med Genet (1999) 36:386–93.
4 Bell R, Rankin J, Donaldson LJ. Down's syndrome: occurrence and outcome in the north of England, 1985–99. Paediatr Perinat Epidemiol (2003) 17:33–39.[CrossRef][Web of Science][Medline]
5 Carothers AD, Boyd E, Lowther G, et al. Trends in prenatal diagnosis of Down syndrome and other autosomal trisomies in Scotland 1990 to 1994, with associated cytogenetic and epidemiological findings. Genet Epidemiol (1999) 16:179–90.[CrossRef][Web of Science][Medline]
6 Forrester MB, Merz RD. Trisomies 13 and 18: prenatal diagnosis and epidemiologic studies in Hawaii, 1986–1997. Genet Test (1999) 3:335–40.[Web of Science][Medline]
7 Martinez-Frias ML, Bermejo E, Rodriguez-Pinilla E, Prieto L. Periconceptional exposure to contraceptive pills and risk for Down syndrome. J Perinatol (2001) 21:288–92.[CrossRef][Medline]
8 Vrijheid M, Dolk H, Stone D, Abramsky L, Alberman E, Scott JE. Socioeconomic inequalities in risk of congenital anomaly. Arch Dis Child (2000) 82:349–52.
9 Murdoch JC, Ogston SA. Characteristics of parents of Down's children and control children with respect to factors present before conception. J Ment Defic Res (1984) 28:177–87.[Web of Science][Medline]
10 Hook EB. Human chromosome abnormalities. In: Perinatal Epidemiology—Bracken MB, ed. (1984) New York: Oxford University Press.
11 Bound JP, Francis BJ, Harvey PW. Down's syndrome: prevalence and ionising radiation in an area of north west England 1957–91. J Epidemiol Community Health (1995) 49:164–70.
12 Cordier S, Chevrier C, Robert-Gnansia E, Lorente C, Brula P, Hours M. Risk of congenital anomalies in the vicinity of municipal solid waste incinerators. Occup Environ Med (2004) 61:8–15.
13 Harrison RM. Hazardous waste landfill sites and congenital anomalies. Occup Environ Med (2003) 60:79–80.
14 Cedergren MI, Selbing AJ, Lofman O, Kallen BA. Chlorination byproducts and nitrate in drinking water and risk for congenital cardiac defects. Environ Res (2002) 89:124–30.[Medline]
15 Berkowitz GS, Obel J, Deych E, et al. Exposure to indoor pesticides during pregnancy in a multiethnic urban cohort. Environ Health Perspect (2003) 111:79–84.[Web of Science][Medline]
16 Wennborg H, Magnusson LL, Bonde JP, Olsen J. Congenital malformations related to maternal exposure to specific agents in biomedical research laboratories. J Occup Environ Med (2005) 47:11–19.[Web of Science][Medline]
17 Ramsay CN, Ellis PM, Zealley H. Down's syndrome in the Lothian region of Scotland 1978 to 1989. Biomed Pharmacother (1991) 45:267–72.[CrossRef][Medline]
18 Dean G, Nevin NC, Mikkelsen M, et al. Investigation of a cluster of children with Down's syndrome born to mothers who had attended a school in Dundalk, Ireland. Occup Environ Med (2000) 57:793–804.
19 Stoller A, Collmann RD. Incidence of infective hepatitis followed by Down's syndrome nine months later. Lancet (1965) 286:1221–23.[CrossRef]
20 Stark CR, Fraumeni JF. Viral hepatitis and Down's syndrome. Lancet (1966) 287:1036–37.
21 Stoller A, Collmann RD. Virus aetiology for Down's syndrome (mongolism). Nature (1965) 208:903–4.[Medline]
22 Stark CR, Mantel N. Lack of seasonal- or temporal-spatial clustering of Down's syndrome births in Michigan. Am J Epidemiol (1967) 86:199–213.
23 Morris JK, Alberman E, Mutton D. Is there evidence of clustering in Down's syndrome? Int J Epidemiol (1998) 27:495–98.
24 Richmond S, Atkins J. A population-based study of the prenatal diagnosis of congenital malformations over 16 years. BJOG (2005) 112:1349–57.[CrossRef][Web of Science][Medline]
25 Pless-Mulloli T, Howel D, King A, et al. Living near opencast mining sites and children's respiratory health. Occup Environ Med (2000) 57:145–51.
26 Pless-Mulloli T, Howel D, Prince H. Prevalence of asthma and other respiratory symptoms in children living near and away from opencast mining sites. Int J Epidemiol (2001) 30:556–63.
27 Fell DB, Dodds L, King WD. Residential mobility during pregnancy. Paediatr Perinat Epidemiol (2004) 18:408–14.[CrossRef][Web of Science][Medline]
28 Shaw GM, Malcoe LH. Residential mobility during pregnancy for mothers of infants with or without congenital anomalies: a reprint. Arch Environ Health (1992) 47:236–38.[Web of Science][Medline]
29 Bailey TC, Gatrell AC. Interactive Spatial Data Analysis. (1995) Harlow: Longman.
30 Diggle PJ, Chetwynd AG, Haggkvist R, Morris SE. Second-order analysis of space-time clustering. Stat Methods Med Res (1995) 4:124–36.
31 Knox EG. The detection of space-time interactions. Appl Stat (1964) 13:25–30.[CrossRef]
32 Mantel N. The detection of disease clustering and a generalized regression approach. Cancer Res (1967) 27:209–20.
33 Kulldorff M, Hjalmars U. The Knox method and other tests for space-time interactions. Biometrics (1999) 55:544–52.[CrossRef][Web of Science][Medline]
34 Rothman KJ. A sobering start for the cluster busters’ conference. Am J Epidemiol (1990) 132(Suppl 1):S6–13.[Web of Science][Medline]
35 Neutra RR. Counterpoint from a cluster buster. Am J Epidemiol (1990) 132:1–8.
36 Burbank F. A sequential space-time cluster analysis of cancer mortality in the United States: etiologic implications. Am J Epidemiol (1972) 95:393–417.
37 DiMaio D, Liao JB. Human papillomaviruses and cervical cancer. Adv Virus Res (2006) 66:125–59.[Web of Science][Medline]
38 Ederer F, Myers MH, Mantel N. A statistical problem in space and time: Do leukemia cases come in clusters? Biometrics (1964) 20:626–38.[CrossRef][Web of Science]
39 Savva GM, Morris JK, Mutton DE, Alberman E. Maternal age-specific fetal loss rates in Down syndrome pregnancies. Prenat Diagn (2006) 26:499–504.[CrossRef][Web of Science][Medline]
40 Rankin J, Pattenden S, Abramsky L, et al. Prevalence of congenital anomalies in five British regions, 1991–99. Arch Dis Child Fetal Neonatal Ed (2005) 90:F374–79.
41 Dolk H, Loane M, Garne E, et al. Trends and geographic inequalities in the prevalence of Down syndrome in Europe, 1980–1999. Rev Epidemiol Sante Publique (2005) 53(Spec No 2):2S87–95.[Web of Science][Medline]
42 Morris JK, Wald NJ. Estimating the risk of Down's syndrome in antenatal screening and the gestation at which this risk applies. J Med Screen (2007) 14:5–7.
43 Anneren G, Gronowitz JS, Kallander CF, Sundqvist VA. Mothers of children with Down syndrome have higher herpes simplex virus type 2 (HSV-2) antibody levels. Hum Genet (1986) 72:9–14.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  J. K Morris Commentary: Clustering in Down syndrome Int. J. Epidemiol., October 1, 2008; 37(5): 1179 - 1180. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||