The incidence of spina bifida in Sweden 1973–2003: the effect of prenatal diagnosis

Abstract

Background: Many studies have been conducted on the accuracy of prenatal ultrasound diagnosis of foetal CNS-malformations. These studies were mostly hospital-based or, sometimes, multicentre studies. We present here a population-based study of the prenatal diagnosis of spina bifida in Sweden over a period of 31 years. Methods: We compared the number of newborns with spina bifida and the elective terminations because of the prenatal diagnosis of spina bifida for different periods. Results: The rate of spina bifida among newborns diminished gradually from 0.55 per 1000 to 0.29 per 1000 during the study period. In M county the rate of spina bifida at birth decreased very rapidly and from 1993 onwards was about half of that in the rest of the country. Conclusion: There has been a decline in the rate of spina bifida at birth. This decline can be seen earlier in the southern part of the country, M county. The decline is probably, to a great extent, a consequence of prenatal ultrasound screening.

Several studies have reported a declining frequency of neural tube defects (NTDs) at birth.¹ The decline has been strongest in areas with a high previous prevalence of NTD. There are probably different reasons for this decline, one of them is elective abortion after prenatal diagnosis. However, before the introduction of prenatal diagnosis, a decline in spina bifida rate was noted in many countries including Sweden, apparently not due to under-reporting.² The efficiency of ultrasound screening in the prenatal diagnosis of spina bifida has been demonstrated but most of these studies are either hospital-based or multicentre studies. There are few population-based studies of this problem.³ In recent years, a further cause of the decline in NTD rate has been suggested: the use of periconceptional folic acid supplementation and/or food fortification with folic acid [e.g. Ref. (4)].

In this study, we present data on spina bifida over a 31 year period in Sweden, which traditionally has low NTD incidence and so far no food fortification with folic acid. We specifically study the performance and impact of prenatal diagnosis of spina bifida.

Methods

Newborns with spina bifida were identified from various Swedish health registers for the period 1973–2003. The main source was the Medical Birth Registry^5,6 from which infants with spina bifida were identified from ICD (International Classification of Diseases) codes, given at the paediatric examination of the newborn. Although all births in Sweden have to be reported to this register, a small percentage is missed each year. Recording of congenital malformations is incomplete in this register so data were supplemented from the Swedish Register for Congenital Malformations, a surveillance register of major congenital malformations.⁷ From 1987 onwards, data from the Hospital Discharge Register were also used (before 1987, this register did not cover the whole country) and the identification of the infants with spina bifida was based on ICD codes.

Selectively aborted foetuses after a prenatal diagnosis of spina bifida should be reported to the Register of Congenital Malformations, but this reporting was very incomplete until 1999 when a modification of the reporting system increased ascertainment dramatically.

Further data on selectively aborted foetuses with spina bifida were obtained from a local register in one area of the country (Malmöhus county, called M county). This register was started in 1984 but data from it were used for the period 1988–2003. The register has a good coverage and is based on direct reporting of findings from the units performing prenatal diagnosis.⁸

Infants and foetuses with a known chromosome anomaly were excluded from the analysis. Foetuses with anencephaly were also excluded from the analysis.

Prenatal diagnosis of spina bifida is based on ultrasound examination since maternal serum AFP screening has not been used in Sweden other than sporadically. Approximately 97% of pregnant women⁹ attend at least one ultrasound screening during their pregnancy. Most of these screening examinations are performed by specially trained midwives. In Sweden, elective abortion because of foetal malformations is allowed up to 23 gestational weeks, according to present practice. Thus, only the diagnoses made before 23 completed weeks give the parents-to-be a possibility to choose an abortion.

We compared the prevalence at birth of spina bifida for groups of years (in order to increase numbers) for M county and for the rest of the country. Up to 2003, 5 year periods were chosen.

Using data from the M county, estimates were made of how many foetuses with spina bifida have been aborted in the rest of the country, assuming that the total rate of spina bifida is similar in the M county and the rest of the country (as it was before the introduction of prenatal diagnosis). The estimated number of abortions in the rest of the country was compared with the number reported to the Register of Congenital Malformations for the period 1999–2003.

Comparisons of frequencies made with chi-square statistics and confidence intervals for specificity estimates were based on exact binomial distributions using SABER software. Trend analyses were made using StatExact software.

Results

Table 1 shows the number of infants and aborted foetuses with spina bifida and the total number of births in M county and the rest of the country for the studied periods.

Table 2 compares the rates of newborns in M county and the rest of Sweden. Up to 1992, little difference in rates existed, but after that the M county rate decreased more rapidly than the rate in the rest of the country. The rate of newborns with spina bifida in M county was by that time about half of that in the rest of the country.

In the M county, the ‘total’ rate of spina bifida (newborns and aborted foetuses) can be calculated from 1988 onwards. The rate for the years 1988–92 was 0.92 per 1000, for years 1993–98 it was 0.58 per 1000, and for years 1999–2003 it was 0.57 per 1000. The decline from years 1988–92 to 1993–2003 is statistically significant (P < 0.05).

In our material, all the foetuses with a diagnosis of spina bifida before 23 completed gestational weeks were aborted. There were seven cases of late diagnosis (after 23 gestational weeks) and these were the only ones born with a prenatal diagnosis. Thus, the efficiency of prenatal diagnosis in M county could be estimated as the percentage of all spina bifida cases aborted. Table 3 gives the results. There was increased efficiency during the observation period. A test for trend gives a P-value of 0.01.

When the total M county rate for 1999–2003 was applied to the rest of the country, 230 total cases should be expected. This agrees well with the numbers observed: 124 infants born, 112 aborted foetuses, and an efficiency of prenatal diagnosis of 47% [95% confidence interval (95% CI) 41–54%].

Applying the rates for 1988–98 in a similar way, the total rate in the rest of the country was estimated to be 816. As 552 newborns were found, this indicates that 264 foetuses with spina bifida were aborted and the efficiency of prenatal diagnosis was only 32% (95% CI 29–36).

Discussion

There has been a marked decline in the prevalence at birth of spina bifida in Sweden. Such a decline is difficult to interpret for various reasons but the contribution of prenatal diagnosis followed by elective abortion is undoubtedly strong. Our data showed that the efficiency of prenatal diagnosis of spina bifida increased in the M county during the observation period and estimates made for the rest of the country indicated a similar but weaker trend.

The effect of prenatal diagnosis was studied from 1988 onwards. In 1986 the intracranial signs for spina bifida were first reported.¹⁰ These are the ‘banana sign’ due to Arnold Chiari malformation (displacement of the cerebral vermis) and ‘lemon sign’ due to frontal bossing. A dilatation of the lateral ventricles is present in up to 70% of open spina bifida in the second trimester.³ These cranial signs have substantially contributed to the better detection rate of spina bifida.¹¹

Maternal serum AFP screening is used in many countries, notably in those with a high incidence of NTDs. The sensitivity of serum screening was estimated to be 84–92%.¹² Patients with a positive serum screening are usually referred to expert sonography and in these cases the accuracy of diagnosis is close to 100%.¹³ The diagnostic accuracy in routine non-targeted examinations is uncertain.³ There are several prospective studies on low-risk populations. The RADIUS study¹⁴ in which ultrasound was used in combination with maternal AFP screening reported a sensitivity of 80%. In two Belgian studies,^15,16 the sensitivity was much lower, 30 and 40% when maternal serum screening was presumably not used. These sensitivity figures are close to those estimated for the rest of Sweden in the period 1988–98; however, during the more recent years the sensitivity has increased. In M county, the detection rate during 1999–2003 was 73% but the upper confidence limit is close to 90% and the lower limit (52%) is about that estimated for the rest of the country (47%).

In Sweden, no major change in the prevalence at birth of spina bifida could be seen until the early 1990s, in spite of the introduction of widespread prenatal diagnosis, which apparently identified a substantial proportion of cases. This is difficult to explain if a real increase in the total rate of spina bifida did not occur, which was also seen in the high total rate in the M county for the 1988–92 period (0.92 per 1000), and then declined markedly.

Use of folic acid before and during early pregnancy has been thought to reduce the rate of spina bifida. Various estimates have been made of the efficiency but use of folic acid supplementation is believed to cause a reduction of ∼50%. No food fortification with folic acid has occurred in Sweden during the observation period but folic acid supplements (0.4 mg) have been available.

Differences in the rates of infants born with spina bifida between M-county and the rest of the country could have been explained by a lower use of folic acid in the rest of the country than in M-county. If so, the estimated rate of aborted foetuses with spina bifida would be too low, but for the last 5 year period, the estimated and observed numbers are very close (106 and 112, respectively).

Two estimates of the use of folic acid made in two different parts of Sweden^17,18 indicate that ∼5–6% of pregnant women in Sweden around 1997 had used folic acid supplementation before conception. Even if the use of folic acid supplementation has increased since then, it cannot explain the one-third decrease in rate observed in M county. Similar increases preceding decreases have been described previously in Sweden² and in other countries.¹ The explanation for these fluctuations of spina bifida rate is to a large extent unclear.

The clearly better detection rate during the last part of the long observation period may be explained not only by the use of cranial signs but also by the improvement in ultrasound technology.

References