According to Little and Bryan [1], Congenital Malformations (CMs) occurred more frequently in twins than in singletons in most studies. In Japan, the prevalence of CMs was higher in twins (1.2%) than in singletons (1.07%) [2]. Several studies have reported that the prevalence of birth defects in multiple births increased following Assisted Reproductive Technology (ART) [3-6]. In Japan, the twinning rate has increased since 1987 and is attributed to ovulation stimulation treatments and ART [7]. It is of interest to know the yearly change in the Infant Mortality Rate (IMR) associated with birth defects for both singletons and twins.

According to Broussard et al. [8], birth defects were a major contributor to Infant Mortality (IM) in the US. In Japan, the proportion of CMs among all infant deaths was 1.9% in 1947 and increased to 10.7% in 1965 and 26.9% in 1981 [9] and to 35.3% in 1995 [10].

The purpose of this study was to compare all-cause IMRs associated with birth defects in singletons and twins during the period 1995-2008 and to identify risk factors for IMR associated with birth defects.

Data on Live Births (LBs) and infant deaths were obtained from the vital statistics of Japan for the years 1995-2008 (Health and Welfare Statistics and Information Department, Ministry of Health, Labour and Welfare, Japan). These data cover the entire population of Japan. Death certificates provide information regarding nationality, sex, dates, Birth Weight (BW), Gestational Age (GA), parental age, single or multiple births, birth order of multiple births, cause of death, and other details. LB certificate records contain the same information, except for data related to cause of death. In Japan, Neonatal Mortality (NM) refers to the death of a live-born infant occurring within 28 days from the time of birth. The NM Rate (NMR) is defined as the number of NM cases per 10000 LBs. IM refers to the death of a live-born infant occurring within 1 year from the time of birth. IMR is defined as the number of IMs per 10000 LBs.

All birth defects were classified into 11 categories according to the ICD 10^th revision [11]: Q00-Q07 (CMs of the nervous system), Q10-Q18 (CMs of the eye, ear, face, and neck), Q20-Q28 (CMs of the circulatory system), Q30-Q34 (CMs of the respiratory system), Q35-Q37 (the cleft lip and cleft palate), Q38-Q45 (other CMs of the digestive system), Q50-Q56 (CMs of genital organs), Q60-Q64 (CMs of the urinary system), Q65-Q79 (CMs and deformations of the musculoskeletal system), Q80-Q89 (other CMs), and Q90-Q99 (chromosomal abnormalities, not elsewhere classified). Of the 19 specific causes of deaths in Table 1, 18 causes of deaths were chosen in accordance with the large number of infant deaths resulting from each cause of deaths in singletons and twins as well as conjoined twins. In Table 2, the linear regression coefficients of IMR on the year was used to test the change in IMR from 1995-1996 to 2007-2008. The odds ratio (OR) and 95% Confidence Intervals (CIs) were used to test IMRs between the two categories. The correlation coefficient was also used to test the relationship between IMR and MA in Table 3. The average of MA in each group was computed in each CM category for both singletons and twins. The following numbers of infant deaths did not differ between singletons and multiple births: 366 (7.2%) in 1995, 313 (6.9%) in 1996, 337 (7.7%) in 1997, 330 (7.5%) in 1998, 271 (6.8%) in 1999, 268 (7.0%) in 2000, 259 (7.2%) in 2001, 220 (6.3%) in 2002, 206 (6.1%) in 2003, 199 (6.4%) in 2004, 209 (7.1%) in 2005, 185 (6.5%) in 2006, 169 (6.0%) in 2007, and 192 (6.9%) in 2008.

Table 1: Number of Infant Deaths due to 11 Categories of All CMs and 19Specific Cause of Deaths and Infant MortalityRatein Singletons and Twins, 1995-2008.

Yearly change in IMRs associated with birth defects

Table 2 shows IMRs in singletons and twins according to 11 categories of birth defects from 1995-1996 to 2007-2008. The percentage of infant deaths associated with birth defects among the total infant deaths was nearly constant (39%-42%) in singletons during the period, whereas it was 21% in 1995-1996 and gradually increased to 27% in 2007–2008 in twins. Excluding the small number of infant deaths in three CM categories (Q10-Q18, Q35-Q37, and Q50-Q56) among 11 categories, IMR decreased significantly for Q00-Q07, Q20-Q28, Q30-Q34, Q38-Q34, Q65-Q79, and Q80-Q89 in singletons from 1995-1996 to 2007-2008 and decreased significantly for Q20-Q28 and Q30-Q34 in twins. For Q00-Q07 in twins, IMR increased during the period; however, the increase was not statistically significant. IMR associated with all birth defects decreased significantly in both singletons and twins. The Relative Risk (RR) in twins vs. singletons was nearly constant between 1995-1996 and 2007-2008 (2.7-3.4-fold) for all birth defects. In contrast, the ratio for Q00-Q07 was 5.5-fold in 1997- 1998, increased to 19-fold in 2005-2006, and decreased to 13.2-fold in 2007–2008. Table 2 also shows IMR associated with all birth defects in both sexes. With one exception, IMRs between males and females were similar in each period in both singletons and twins. The exception was IMR for the period 2005–2006 in twins where the rate was significantly higher in males than in females.

Table 2: Number of Congenital Malformations, Deformations and Chromosomal Abnormalities (CM) of Infant Deaths and Death Rates in Singleton and Twins, 1995-2008.

Table 2 also shows the proportion of neonatal deaths among infant deaths in each CM category. Excluding the small number of infant deaths in three categories of CMs, the proportions ranged between 36% and 92% in singletons and 37% and 100% in twins. For all birth defects, these values were 57% in singletons and 75% in twins. The highest difference between percentages of singletons and twins was 29% (Q00-Q07), followed by 18% (Q20-Q28 and the total birth defects). NMR for all birth defects was significantly higher in twins (23.80 per 10000 LBs) than in singletons (6.26) (OR, 3.81; 95% CI, 3.54-4.11).

Comparison of IM between singletons and twins

Table 1 shows IMRs for 11 CM categories and 19 specific causes of deaths during the period 1995-2008. Excluding the small number of deaths in three categories (Q10-Q18, Q35-Q37, and Q50-Q56), IMRs were significantly higher in twins than in singletons for seven CM categories, where the highest RR in twins vs. singletons was 10.6- fold for Q00-Q07 and the lowest was 2-fold for Q65-Q79. In contrast, RR in twins vs. singletons was 0.87-fold for Q90–Q99. Of IMRs for 18 specific causes of deaths, IMRs for 10 specific causes of deaths were higher in twins than in singletons. In singletons, the highest IMR was 1.32 for Edward’s syndrome, unspecified (Q91.3), followed by 1.00 for hypoplasia and dysplasia of the lung (Q33.6) and 0.69 for hypoplastic left heart syndrome (Q23.4). In twins, the highest IMR was 4.70 for patent ducts arteriosus (Q25.0), followed by 3.58 for anencephaly (Q00.0) and 3.48 for Q33.6. RR in twins vs. singletons was 51.6-fold for Q00.0, followed by 13.7-fold for Q25.0. IMR in conjoined twins (Q89.4) was 0.64; in this case, the cause of death is peculiar to twins.

IMRs by maternal age (MA)

Table 3 shows IMRs for Q00-Q07, Q20-Q28, Q90-Q99, and the total birth defects according to MA during 1999-2008. IMRs for Q00-Q07 in singletons were 0.3-0.4 for MA <25 to 35-39 groups and 1.0 for the oldest MA group, where the differences between the lowest IMR and those for the oldest MA group were statistically significant at the 5% level. In contrast, IMRs in twins were also similar (4.5-5.6) for MA <25 to 35-39 years and the lowest (1.8) for MA ≥ 40 years, where the number of infant deaths was only one. IMRs in twins were independent of MA. Excluding the small number of infant deaths, RRs in twins vs. singletons were between 11- and 17-fold. For Q20– Q28, IMR in singletons was 4.2 for MA <25 years to 25-29 years and increased thereafter (11.1) up to MA ≥ 40 years. IMR was significantly lower for MA <25 years than for MA 35-39 years and ≥ 40 years. In contrast, IMR in twins decreased from 17.4 for MA <25 years to 11 at 30–34 years and increased thereafter to 22 up to MA ≥ 40 years. IMR for MA 30–34 years was significantly lower than that for MA <25 years or ≥ 40 years. IMR was significantly higher in twins than in singletons for each MA group. RR in twins vs. singletons was the highest for the youngest MA group (4.1-fold) and decreased (2-fold) for MA ≥ 40 years. For Q90–Q99, IMR in singletons was the lowest (1.1) for MA <25 years and increased (13.5) with an increase in MA up to ≥ 40 years. IMR was significantly lower for MA <25 years than for MA of 30-34, 35-39, and ≥ 40 years. In contrast, IMR in twins was 0.5 for MA <25 years and increased to 9.2 with an increase in MA up to ≥ 40 years. IMR for MA 25-29 years was significantly lower than those for MA of 35-39 years and ≥ 40 years. IMR was similar in both singletons and twins in each MA group. With regard to the birth defects, IMR in singletons was the lowest (8.5) for MA of 25-29 years, which was significantly lower than those for MA of 30-34 (9.3), 35-39 (13.6), and ≥ 40 years (30.8). Similarly, IMR in twins was the lowest (28) for MA 25-29 years, whereas IMRs in the oldest MA groups were the highest (46). The former IMR was significantly lower than the latter. IMR was significantly higher in twins than in singletons for each MA group. RR in twins vs. singletons was the highest in the youngest MA group (3.7- fold) and decreased (1.5-fold) with MA up to ≥ 40 years.

Table 3: Infant Mortality Rate (IMR) according to Maternal Age (MA), and Correlation Coefficients between IMR and MA, 1999-2008.

Correlation coefficients between MA groups and IMRs are shown in Table 3 for Q00-Q07, Q20-Q28, Q90-Q99, and the total birth defects. For singletons, correlation coefficients indicated higher values (0.73- 0.84); however, these were not significant at the 5% level. In contrast, these coefficients in twins were -0.60 for Q00-Q07, 0.31 for Q20-Q28, 0.82 for Q90-Q99, and 0.67 for the total birth defects, which were not significant at the 5% level.

IMRs by GA

Table 4 shows IMR for three categories of CMs and the total birth defects according to GA in singletons and twins during the period 1999- 2008. With one exception, IMRs in singletons and twins decreased with an increase in GA from the shortest GA to the longest for three CM categories and the total birth defects, with the exception of Q00-Q07 at GA of 37-38 weeks in twins. IMR for Q00-Q07 was significantly higher in twins than in singletons for four GA groups ≥ 33-34 weeks. In contrast, IMRs for Q20-Q28 (33-34 and 35-36 weeks), Q90-Q99 (five GA groups <39 weeks), and the total birth defects (three GA groups ranging between 31-32 and 35-36 weeks) were significantly higher in singletons than in twins. IMRs for the total birth defects at GA of 37-38 and ≥ 39 weeks were significantly higher in twins than in singletons, where the corresponding RRs in twins vs. singletons were 1.4-fold and 2.9-fold, respectively.

Table 4: Infant Mortality Rate according to Gestational Age, 1999-200.

IMRs by BW

Table 5 shows IMR for the three categories of CMs and the total birth defects according to BW in singletons and twins during 1995- 2008. With two exceptions, BW-specific IMRs in singletons and twins decreased with an increase in BW from the lightest to the heaviest weight in both singletons and twins. The exceptions were for the total birth defects at BW 1000-1499 g in singletons and ≥ 3000 g in twins. For Q00-Q07, IMRs were significantly higher in twins than in singletons for BW <1000g and 1000–1499 g. For Q20-Q28, IMR was significantly higher in twins than in singletons for BW <1000 g and ≥ 3000 g. In contrast, IMRs were higher in singletons than in twins from 1000-1499 g to 2000-2499 g. For Q90-Q99, IMRs were significantly higher in singletons than in twins from BW <1000 g to 2000-2499 g. With regard to the total birth defects, IMR was significantly higher in singletons than in twins from BW 1000-1499 g to 2000-2499 g, whereas the opposite result was obtained at the heaviest weight.

Table 5: Infant Mortality Rate (IMR) due to Congenital Malformations according to Birth Weight, 1995-2008.

IMRs according to birth order of twins

Table 6 shows IMR for categories of CMs according to the birth order of twins during 1995-2008. With respect to the total birth defects, IMR in the second birth order of twins was significantly higher than that in the first order of twins (1.3-fold). A similar tendency was observed for Q00-Q07 (1.5-fold), Q38-Q45 (2.2-fold), Q65-Q79 (2.0- fold), and Q80-Q89 (1.6-fold).

Table 6: Infant Mortality Rate (IMR) due to Congenital Malformations according to Birth Order of Twins, 1995-2008.

In the present study, IMRs for CMs of the nervous system in twins increased from 1995–1996 to 2005–2006. In contrast, the rate in singletons decreased during this period. The corresponding RRs in twins vs. singletons were 6-fold and 19-fold, respectively. Similarly, RR in twins vs. singletons was 51.6-fold for anencephaly during 1995– 2008. Lancaster [3] reported that the prevalence of neural tube defects increased following in vitro fertilization and stimulation of ovulation. A significant correlation was observed only between anencephaly and twin births [12]. In Japan, the twinning rate was 8.9 per 1000 births in 1995–1998 and increased to 10.2 in 1999–2003 and 11.1 in 2004–2008 [13]. According to Hansen et al. [4], children born following ART were at a high risk of birth defects. Ooki [5] reported that the early NMR was slightly higher in the ART group (5.09 per 10000 LBs) than in the general population (3.86) in Japan during the period 2004–2008. In Japan, the higher RR of anencephaly is related to the effect of in vitro fertilization and stimulation of ovulation.

RR in twins vs. singletons was 2.9–fold for IMR and 3.8–fold for NMR during the period 1995–2008. Therefore, intensive care of twins in the neonatal period is important to reduce the difference in IMRs between singletons and twins. IMR for patent ductus arteriosus (PDA; Q25.0) was 13.6–fold higher in twins than in singletons. Layde et al. [14] reported that incidence rates of PDA were 17.43 per 10000 LBs in white singletons and 86.07 in white twins in the US during 1969–1976, where RR in twins vs. singletons was 4.9–fold. The corresponding values in the black population were 24.38–, 184.44–, and 7.6–fold, respectively. They also reported that a high concordance rate was observed in samesex twin pairs, which suggested a strong genetic component to the etiology of PDA. In the present study, excluding infant deaths due to anencephaly, PDA, and conjoined twins from the total infant deaths associated with birth defects, IMR was recomputed using Table 1 and IMR of 10.67 was obtained in singletons and 23.06 in twins. RR in twins vs. singletons was 2.2–fold. The difference between IMRs in twins and singletons can be reduced by 25% compared with IMRs associated with all birth defects. It should be possible to reduce IMR associated with birth defects in twins through management for the reduction of premature and low BW and care for the second order of twins.

We are grateful to the staff of Statistics and Information Department, Ministry of Health, Labour and Welfare in Japan.