The prevalence of birth defects can be influenced by many factors including case definition, TOPFA, the time of observation after birth, population study methods, case ascertainment methods and reporting and statistical procedures used –.
Termination of pregnancy is legal up to 22 weeks in Vietnam, but reporting of pregnancy termination is not required. However, prenatal diagnosis does not exist in our study's population. We therefore believe that the reason for termination of pregnancy in Binh Thuan is rarely an external birth defect. Consequently, TOPFA most likely does not have an influence the prevalence of the birth defects.
We found that the prevalence of EBDs across the age distribution tended to be a U-shaped curve; prevalence dropped substantially for women over 40 years of age and only marginally for other age groups. For non-chromosomal defects, the U-shaped pattern of prevalence across maternal age has been documented by many authors [7, 15, 17].
In this study, the relationship between maternal education and an EBD did not necessarily mean that maternal education itself was a risk factor for EBDs. Educational qualification most probably determines socio-economic level and/or occupation and prenatal care behavior. It is therefore conceivable that education might affect the occurrence of EBDs indirectly .
Most reported associations between occupational exposures and adverse reproductive outcomes in epidemiological studies are equivocal and often controversial . Significant association of occupational pesticide exposure and all birth defects were reported by Nurminen, et al. from a study in Finland , and by Restrepo et al. in Colombia . Our findings show that the prevalence of EBDs was not significantly different between women involved in agricultural activities and/or working as an agricultural chemical products seller and mothers involved in another occupation.
Our results show an increased prevalence of external birth defects occurring among mothers with either primigravida or gravida over 4.
According to Swain et al., infants born to gravida 4 or more mothers have higher rate of birth defects when compared to mothers of lower gravidity . Tan et al. reported that the prevalence of birth defect increased with birth order .
The relationship between the mother’s age at delivery and gravidity may be one possible explanation for the high rate of EBDs at both extremes of maternal gravidity in the present study.
This study demonstrates that birth defects are significantly associated with preterm birth and low birth weight. Although preterm and low birth weight infants are more likely to have birth defects, the effect of birth defects on preterm birth and low birth weight has been difficult to study because of multiple confounding risk factors [24, 25].
Many studies have documented male preponderance in birth defects [26, 27]. However, in the present study, a very slight female preponderance was found (42 females versus 40 males).
As expected, the overall prevalence of EBDs in our study (6.02 per 1000 live births) was lower than the EUROCAT (25.53/1000) and Belgian (23.11/1000) registries  because the present study reported only EBDs detected within 24 hours after birth. Our finding was similar to the prevalence rate in Taiwan, which is 7.3/1000 births. In Taiwan, EBDs were detected within a few days after birth .
When considering the type of external birth defect, limb defects, nervous system defects, orofacial clefts and external genital system defects are by far among the most common birth defects worldwide –. In the present study, limb defects, orofacial clefts and central nervous system defects were the three most common groups.
In our study, the most common limb defects were clubfoot, polydactyly and limb reduction, respectively.
Club foot is the common type of limb defect. Prevalence varies widely in among recent international reports. According to data from EUROCAT, the prevalence of clubfoot was reported to be 10.31/10,000 total births for all members, 11.21/10,000 in Belgium and varied from low (3.22 per 10,000) to high (18.00 per 10,000) in Ukraine and Saxony-Anhalt (Germany), respectively . In recent studies in the United States, Parker et al. reviewed data from the 10 population-based birth defect surveillance programs (6,139 cases of clubfoot) to better estimate the prevalence of clubfoot and found the overall prevalence of clubfoot to be 19.2 per 10,000 live births . Boo et al. reported an incidence of clubfoot in Malaysia at 45 per 10,000 live births . In our study's group, club foot was the second most common EBD and the prevalence of 12.18/10,000 live births fell within the range reported for other registries.
Polydactyly is a major group. It is a defect that is easily detectable after birth and is an isolated finding in 85% - 88% of cases . Our polydactyly prevalence of 6.45/10,000 live births was comparable to other European prevalence rates of 6.79/10,000 for Belgium, 6.80/ 10,000 for Paris, and 6.6/10,000 for Portugal respectively . The prevalence of this birth defect is much higher in China (22.4/10, 000)  and in Alberta, Canada (18.84/10,000) . Prevalence of polydactyly was reported to be lower in Barcelona, Spain (3.06/10,000)  and in Lombardy, Italy (5.82/10,000) .
Limb reduction is one of the most common types of limb defects and accounts for 3.2 to 7.06 per 10,000 births in the literature [10, 26, 27]. This very visible birth defect is symbolic because it launched the development of congenital anomalies surveillance activities worldwide after the thalidomide tragedy in the early 1960s. Limb reduction prevalence was found to be 4.3/10,000 live births among our newborns.
Orofacial clefts are among the most common of all major birth defects. Orofacial clefts are usually obviously visible immediately after birth.
Cleft lip with or without palate involved 20 out of 13,954 live births (14.33 per 10,000 live births), which is similar to the prevalence in Northern Ireland (14.70 per 10,000 live births) [10, 27, 35], lower than in Pakistan (19.10 per 10,000)  but higher than in full member EUROCAT registries (8.63 per 10, 000) , in Norway (10.9 per 10,000) , China (18.9 per 10,000)  and Korea (10.3 per 10,000) .
According the international perinatal database report on typical oral clefts, the prevalence of cleft lip with or without cleft palate from 54 registries in 30 countries over at least 1 complete year during the period 2000 to 2005 was 9.92 per 10,000 births, which was lower than our finding .
Isolated cleft palate is very difficult to detect prenatally due to shadowing artefacts from amniotic bands or other overlying structures.
The prevalence of 5.37 per 10,000 live births for cleft palate in this study was comparable to those observed in the full member EUROCAT registry (5.59 per 10,000)  and in Lombardy, Italy (5.82/10,000) . Our figure was slightly higher than those reported in Taiwan (4.67 per 10, 000) , and in Belgium (3.59 per 10,000) , but lower than those reported in Wessex, United Kingdom (10.0/10 000) and in Ireland (7.21/10,000) .
Neural tube defects can be categorized as either anencephalus or similar (lack of closure in the head region) or spinabifida (lack of closure below the head). The two major categories of neural tube defects occur in approximately equal frequencies at birth [13, 42].
Our data revealed that the prevalence for anencephalus or similar was 3.58 per 10,000 live births. This figure is comparable to that of the full member EUROCAT registry (3.50/10,000) and Belgium (3.16/10,000) . In contrast with the relatively high frequencies of anencephaly, we did not observe any spinabifida in the present study or in the pilot study in 2008 with 16,593 births. The explanation for the absence of spinabifida cases in our study is complex. It may be due in part to the small sample size, the diagnostic technique used, and/or genetic factors.
As expected, the prevalence of hydrocephaly in our study (1.43/10,000 live births) was low compared to the full member EUROCAT registry (5.31/10,000), Belgium (5.52/10,000) and other registries [10, 26, 27, 29, 39]. Hydrocephaly is a malformation that is easier to diagnose by prenatal ultrasound scanning. It is not often obvious at birth and is usually detected after birth by an increasing head circumference that crosses percentiles on the growth chart. We therefore believe that hydrocephaly was under-diagnosed in our study.
Hypospadias is considered the most common congenital malformation in the genitourinary system. Usually hypospadias is detected at birth by a detailed examination of the newborn or by abnormal flow of urine during urination. Experienced clinical personnel are required to detect hypospadias. In our study, newborns were examined within 24 hour after birth by a local health provider with limited expertise in hypospadias recognition. Thus the prevalence of hypospadias was low (0.72/10,000 live births) compared to other registries [10, 27, 29, 39].
The prevalence of external birth defects was not different between commune health stations and hospitals demonstrating health workers' abilities in detecting EBDs at commune health stations in Binh Thuan province.
Internal organ defects are not visible during a physical exam or they are often asymptomatic, particularly during the first 24 hours of life. In this study, since the examinations were executed by simple measurements and observations of the newborn, birth defects of internal organs (e.g. digestive system heart and circulatory system, internal urogenital system and certain domains of the central nervous system) were undetected.