Effect of ascertainment and data transfer by local registers on

The initial data transfer stages had implications for the study, since the ability and the ease with which data transfer was possible was important to identify whether results could be extended to other registers and could be sustained. The series of iterations working with the Wales register to transfer data in the correct format was important to resolve problems of providing data in the required format. Once these were resolved, data transfer with the subsequent registers was much faster, suggesting that the general problems in providing data in the required format had been resolved, thus facilitating data transfer from other local registers.

Electronic data transfer resulted in significant increases in total notification rates from all four local registers, by 67 per cent to 600 per cent, with the largest increase for Wales which had had the lowest notification rate before data transfer. Those local registers with higher notification rates prior to data transfer would be expected to increase less than those with lower rates. In addition, the Wales register had been transferring data for longer than the other registers at the time of the study reported here, so had had more time to accumulate late notifications for babies born in the first year of data transfer.

In the absence of any reason why true prevalence should be lower in Wales than in England, the lower notification rates for Wales in 1992-1997 could have been due to less complete notification from Wales. This might have been due to problems when transferring responsibility for notifying NCAS of congenital anomalies as a result of administrative changes in Welsh Health Authorities in the early and mid 1990s. For Wales and Trent there was a lower rate in the second year of data transfer compared with the first year, but this may be due to fewer late notifications

for that year having accumulated by the time of the analysis for this study, as shown for the Wales register in Figure 3.1.

Levels of ascertainment varied for different conditions in different registers. Neural tube defects, cleft lip and palate, and limb reduction defects are usually obvious at birth. All four local registers saw a significant increase (p<0.01) in notification rates for cleft lip and palate compared to expected rates. Heart and circulatory, respiratory, and abdominal defects are largely internal anomalies and may not be diagnosed quickly unless prenatally detected. All four local registers saw a significant increase (p<0.01) in notification rates for heart and circulatory anomalies and abdominal defects, compared to expected rates. For heart and circulatory anomalies, the rate was 23 times higher in Wales and 20 times higher in Mersey than expected. All registers except Mersey experienced a significant increase in notification rates for respiratory anomalies on beginning data transfer. The notification rate was double that expected for Trent and over five times that expected for Wales and North West Thames in these conditions. Eye anomalies and Down syndrome require a test for clinical diagnosis. Notification rates from Wales were the highest for both Down syndrome and eye anomalies. All registers except Mersey experienced a significant increase in notification rates for eye anomalies on beginning data transfer.

The various studies reported in Chapter 2 showed that underascertainment in NCAS was largely a result of NCAS only being notified of the most severe cases and those easily visible at birth, whilst multi-source ascertainment by the local registers picked up more minor anomalies and internal anomalies not detected at birth. The increases for the selected conditions reported in this Chapter varied by condition, with heart defects which are known to be least well notified to NCAS showing the largest increases compared with the expected rates.

experienced significant increases in notification rates for limb reduction defects on the change to data transfer, and only in Wales was the notification rate more than double that expected. Whilst all the local registers experienced a significant increase in notification rates for cleft lip and palate on beginning data transfer, the increases were much smaller than increases seen for other conditions. Both these conditions which were visible at birth and specifically checked at the baby’s delivery and therefore likely to have been well recorded on the birth notification form, which was the primary or sole source used for notification by the majority of

HAS prior to data transfer by local registers.

Down syndrome had already been the subject of an earlier follow-up exercise for notifications in 1997 and 1998 which had resulted in an increase in the number of notifications by 44 per cent (Sotting 2000a). Therefore, it was not surprising that data transfer had not lead to a significantly higher rate for Down syndrome than expected for two of the three local registers that began data transfer in 1999 and

2000.

In 2000, after the four local registers had begun data transfer, notification rates for the registers showed close agreement for cleft lip and palate, abdominal wall defects and Down syndrome. To test whether a maximum detection level had been reached, notification rates for these conditions were compared with international notification rates, on the assumption that there are no real differences between different countries for the prevalence of these conditions.

Table 3.8 compares a selection of these notification rates with the range of rates for participating programmes in the International Clearinghouse for Birth Defects Monitoring Systems (International Clearinghouse 2002). The chosen conditions were those available from the annual published reports.

Condition (based on ICD codes as shown in Table 3.5)

Range of rates from local registers, 2000

Range of rates from Clearinghouse Programmes, 1999

Rate for Rest of England, 2000

All Babies 114.6-266.1 68.0

Cleft lip and palate 10.4-13.7 8.24-23.62 7.0

Abdominal defects 54-7.3 0.18-7.54 2.0

Down syndrome 8.6-11.2 4.34-25.35 4.2

Table 3.8 Comparison of notification rates for 4 local registers and the remainder of England in 2000 with Clearinghouse Programmes in 1999.

The range of rates for Clearinghouse programmes was wide reflecting the diversity of programmes and their different prevalence and ascertainment rates. Compared with international data for 1999, which was the latest published data from the International Clearinghouse, the notification rates from the local registers for these conditions were consistent with rates from Clearinghouse programmes, showing that either their rates were closer to true prevalence or that some maximum ascertainment been reached. The rates for the rest of England were lower, and, for cleft lip and palate and for Down syndrome, were below the range of international data.

Although the notification rates from local registers were higher than those received by NCAS from HAs, the rates varied between the local registers. Another study in the USA found similar large differences in prevalence rates between different registries (Hobbs 2001). The authors concluded that factors which influence prevalence rates include case ascertainment sources, case inclusion criteria and inclusion of elective terminations and stillbirths. The impact of each of these factors was thought to be defect specific. Levels of notification for the specific conditions analysed here also varied between the four local registers. This reflects real differences in prevalence levels, the different ways the local registers operated,

differences needed to be explored further to explain the differences in terms of the strengths and weaknesses of the registers. This could help identify whether there were common factors and core requirements for an ideal local register to ensure the sustainability of local registers as the core data providers for NCAS. This work was outside the scope of the study reported here.

There were several differences in the ways registers operated. These included the location of the register (for example in a hospital or child health department), the range of sources notifying to the local register, and the support and interests of local clinicians. Other factors also included the local registers’ use of prenatal diagnosis information, and later notification of older children with anomalies. It also reflected different policies on which conditions should be recorded by the local register. Some registers excluded more minor cases of certain anomalies, for example. North Thames West does not collect isolated hypospadias.

The total notification rate for North Thames West after beginning data transfer was noticeably lower than the total rate for the other two local registers in England. This was likely to be due to a range of factors including being a hospital-based register, and from the use of exclusion criteria by the register. Notification rates for North Thames West suggest that hospital based registers find it harder to achieve good coverage, but there are a number of other local factors that need to be taken into account. There are inherent problems of hospital-based registers as cases are harder to follow up in the long term as they may have been referred from a wide geographical area. The differences reported here, however, could also be a function of this register being in London where other registers have experienced problems in achieving good coverage (Roderick 1992), or their different IT or data collection systems.