In Japan hormonal stimulation of ovulation has been used since 1966 [1]. In July 1975, the “Survey on Socio-Economic Aspects of Vital Events–Plural Births in 1975” was conducted by the Health and Welfare Statistics and Information Department, Ministry of Health and Welfare [2]. Questionnaires were distributed directly to all mothers with multiple births through January 1 to June 30, 1974. This Survey examined the proportion of mothers treated with hormonal stimulation. Based on the data from this survey, Imaizumi and Inouye determined that the proportion of mothers treated with hormonal stimulation, was 4.5% (158/3485) for like-sexed pairs and 9.2% (72/786) for unlike-sexed pairs [3]. In February 1996, the Japan Society of Obstetrics and Gynecology recommended that only 3-4 eggs or embryos should be transferred per treatment cycle. In May 2008, the Japan Society of Obstetrics and Gynecology decided that a single embryo should be transferred per treatment cycle as a general rule.

Dizygotic (DZ) twinning rate has increased since the introduction of reproductive technology in many countries [4-9]. Japan witnessed an increase in the DZ twinning rate from 1987 to 1998 [6,10]. According to vital statistics in Japan, the overall twinning rates increased till 2005 and decreased thereafter (http://www.e–stat.go.jp/SG1/estat). In the US, the rate has stabilized in 2006 [11]. On the contrary, the monozygotic (MZ) twinning rate increased following the introduction of ovulation induction and assisted reproductive technology (ART) and increased the risk of monochorionic twinning associated with the ART [12-14]. However, the recent trend in the MZ twinning rates in Japan is unknown. The summary incidence (27 studies) of monozygotic twins after the ART was 0.9% (0.8–0.9%) [15]. In Japan, the estimated MZ twinning rate after single embryo transfer was around 2% from 2007 to 2009 [16].

This study therefore aimed to estimate the MZ and DZ twinning rates during the period 1999-2008. It also deals with comparison of the MZ and DZ twinning rates before and after the introduction of reproductive treatments.

Data on live births (LBs) and fetal deaths (FDs) were obtained from vital statistic records maintained by the Health and Welfare Statistics and Information Department, Ministry of Health, Labour and Welfare (Tokyo, Japan) for the years 1999-2008. These data cover the entire population of Japan. LB certificates contain information concerning the decedents’ nationality, sex, dates of birth, birth weight, gestational age, parental dates of birth and age, single or multiple births, birth order in multiple births, and other details. FD certificates contain the same information, including the date and cause of FD, but not the parental dates of birth.

Twin pairs were estimated using three record pairings: LB-LB (2LB), FD-FD (2FD), and LB-FD. 2LB and 2FD cases were obtained from the LB and FD certification records, respectively. The LB-FD cases were obtained from LB and FD certification records that excluded 2LB and 2FD twin pairs. We identified 122,758 pairs of twins (99.9% of the total twin births).

The numbers of MZ and DZ twins were estimated by Weinberg’s method [17]. The maternal age (MA) of twin is not always the same because twin pairs could be born on different dates. Then numbers of like- or unlike-sexed twin pairs were comprised of even numbers of twin pairs according to MA and odds numbers of twin pairs.

The parameters of the linear regression model (y=αx+β) were estimated by the least squares method. The linear regression coefficient (α) of the twinning rate (y) on the year (x) was used to determine the trend from 1999 to 2008. The linear regression coefficient for MA (x) was also assessed (19, 22.5, 27.5, 32.5, 37.5, 42.5, and, 47.5 years). The rates are described as per 1000 births.

In addition, we compared the MZ and DZ twinning rates before (1955-1967) and after (1975-1998) the introduction of reproductive treatments [3,7,10]. To enable comparison of the twinning rates, the rates determined in the present study were combined with the results of Imaizumi and Inouye and recomputed using the total number of births as denominators (Table 3 in their report) [3].

Annual change in the twinning rate by zygosity

Table 1 shows the annual change in MZ, DZ, and overall twinning rates from 1999 to 2008. The MZ twinning rates remained fairly constant (3.60-4.34) over this period, while the DZ twinning rate increased from 5.10 in 1999 to 7.66 in 2005, before decreasing to 5.98 in 2008. The overall twinning rates in 1999, 2005, and 2008 were 9.65, 11.51, and 10.37, respectively. Linear regression coefficients (p-values) of the twinning rates on the year were -0.015 (0.544) for MZ twins, 0.165 (0.046) for DZ twins, and 0.151 (0.037) for overall twins. The latter two coefficients were significant at the 5% level (Table 1).

Table 1: Monozygotic (MZ) and dizygotic (DZ) twin deliveries according to survival states of twins and twinning rates for zygotic and overall twins, 1999-2008.

Zygotic and overall twinning rates by MA

Table 2 shows the MZ, DZ, and overall twinning rates according to MA from 1999 to 2008. In general, the twinning rates slightly increased in each MA group (DZ and overall twins), but did not reach significance at the 5% level. The DZ twinning rate for MA 20-24 years and overall rate for MA ≥ 40 years between 1999 and 2008 were exceptions. Figure 1 shows the annual change in the MZ twinning rate according to MA. In the youngest MA group (<20 years), the MZ twinning rate was approximately 4 throughout the period. The rate for MA ≥ 40 years increased from 3.64 in 1999 to 5.40 in 2008, with a twinning rate linear regression coefficient (p–value) of 0.099 (0.189); this was not statistically significant at the 5% level. The MZ twinning rate significantly increased according to MA in 2002 (α=0.042, p=0.012), 2003 (0.032, 0.047), 2006 (0.072, 0.004), and 2008 (0.073, 0.003). The DZ and overall twinning rates significantly increased according to MA annually (Table 2 and Figure 1).

Table 2: Number of twin pairs and twinning rates by zygosity and overall twins by maternal age (MA), 1999-2008

Figure 1: Yearly change of MZ twinning rate per 1000 births by maternal age, 1999-2008.

Comparison of twinning rates before and after introduction of reproductive technology

Figure 2 shows the annual change in the MZ and DZ twinning rates during 1955-1967 and 1974-2008. The MZ twinning rates remained approximately constant (3.56-4.43) for the overall period, whereas the DZ twinning rates were consistent from 1955 to 1987 (1.86-2.47) and gradually increasing up to 2005 (7.57) and finally decreasing slightly in 2008 (5.92). The DZ twinning rate was 2.02-fold higher in 1997 (4.56) than that for the period 1955-1967 (2.26). In 1997 the twinning rate was similar for both MZ (4.40) and DZ twins. The DZ rate was 3.35-fold higher in 2005 than that during 1955-1967 (Figure 2).

Figure 2: Yearly change of MZ and DZ twinning rates, 1955-1967and 1974- 2008.

Figure 3 shows comparison of the MZ and DZ twinning rates according to MA for the period 1999-2008 vs. 1960-1967. The MZ and DZ twinning rates in Figure 3 are a composite of those obtained from Table 3 of the report by Imaizumi and Inouye and from the present study (Table 2) [3]. The MZ twinning rate was constant with MA between the two periods: 0.9-fold for MA <20 years; and 1.0-fold for MA 20-24 years to ≥ 40 years. Conversely, the DZ twinning rate increased with MA between the two periods: 1.2-fold for MA <20 years, gradually increasing to 3.7-fold for MA ≥ 40 years (Figure 3).

Figure 3: Comparison of MZ and DZ twinning rates according to maternal age (MA) for the period 1999-2008 vs. 1960-1967. RR: Comparison of twinning rates for the period 1999-2008 vs. 1960-1967.

The MZ twinning rate remained fairly constant from 1999 to 2008. The rate for MA ≥ 40 years slightly increased during the period where the regression coefficient was 0.099 (p=0.19). The fact may have reflected the increase in the number of older mothers (≥ 35 years) undergoing fertility treatment in recent years. The DZ twinning rate increased from 1999 to 2005 and decreased thereafter. The number of twin cases delivered by ART in Japan was 1975 cases in 1999, which increased to 3559 cases in 2005 and decreasing to 2099 cases in 2008 [18]. Saito reported that restriction of the number of embryos transferred certainly takes effect. This reflected on the decline in the DZ twinning rate after 2005 in the present study [19]. Thus, the DZ twinning rate in Japan may be slowly decreasing to near the levels achieved with natural fertility.

Data on multiple pregnancies from vital statistics are very useful information which covers the entire population of Japan. However, there was no information of pregnancies achieved by ART and pregnancy outcomes.

It is very important to reduce twinning rates regarding to stillbirth and infant mortality rates. In Japan, stillbirth rates were more than two times higher in MZ twins than in singletons and in DZ twins during 1986-1994 [20]. The infant mortality rate was also 4.6 times higher in twins than in singletons during 1999-2008 [21]. Then It is very good policy that a single embryo should be transferred per treatment cycle as a general rule by the Japan Society of Obstetrics and Gynecology.

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