A Nationwide Registry-Based Study on Mortality Due to Rare Congenital Anomalies

This study aimed to analyse population-based mortality attributed to rare congenital anomalies (CAs) and assess the associated time trends and geographical differences in Spain. Data on CA-related deaths were sourced from annual mortality databases kept by the National Statistics Institute of Spain (1999–2013). Based on the ICD-10, only CAs corresponding to rare diseases definition were included in this study. Annual age-adjusted mortality rates were calculated and time trends were evaluated by joinpoint regression analysis. Geographical differences were assessed using standardised mortality ratios and cluster detection. A total of 13,660 rare-CA-related deaths (53.4% males) were identified in the study period. Annual age-adjusted mortality rates decreased by an average of −5.2% (−5.5% males, −4.8% females, p < 0.001). Geographical analysis showed a higher risk of rare-CA-related mortality in regions largely located in the south of the country. Despite their limitations, mortality statistics are essential and useful tools for enhancing knowledge of rare disease epidemiology and, by extension, for designing and targeting public health actions. Monitoring rare-CA-related mortality in Spain has shown a 15-year decline and geographical differences in the risk of death, all of which might well be taken into account by the health authorities in order to ensure equality and equity, and to adopt appropriate preventive measures.


Introduction
Congenital anomalies (CAs) comprise a large, highly heterogeneous group of birth outcomes, usually classified according to the specific organ or system affected. CAs are an important cause of premature death, chronic illness and lifelong disability worldwide. According to the 2015 Global Burden of Disease study, CAs led to 8.5% (7.7-9.5%) of deaths under the age of 5 years [1]. Liu et al., using vital registration data, also concluded that CAs were the most important cause of death in countries with very low (<10 per 1000 livebirths) and low (10-25 per 1000 livebirths) under-5 mortality rates [2]. It is also clear that CA-related mortality merits attention since, while nearly all leading causes of death registered some form of decrease from 2005 to 2015 [1], CAs, along with neonatal sepsis, were the exception without any significant change. Moreover, Oza et al. observed that the proportion of deaths from congenital disorders was relatively stable across the period (data for 2000-2013 in 194 countries), with the smallest relative decrease in risk being predicted for congenital disorders [3].  Spain (1999Spain ( -2013 by underlying cause of death, this being a rare CA. The following list of rare congenital anomalies is based on EUROCAT subgroups of congenital anomalies [4]. Only those EUROCAT defects considered as rare diseases with corresponding ICD-10 codes are displayed.

Figure 1.
Time trends in age-adjusted mortality rates per 100,000 inhabitants for rare congenital anomalies. Shading represents 95% confidence intervals. Figure 1. Time trends in age-adjusted mortality rates per 100,000 inhabitants for rare congenital anomalies. Shading represents 95% confidence intervals. Table 2 includes the annual age-adjusted mortality rates (ARs) by sex and type of rare CA. Apart from the data for the CA grouped by body systems, it also includes defect specific results for all the defects for which at least 180 deaths (i.e., at least an average of 12 deaths per year) were observed in the study period. According to data shown by type of rare CA, the greatest fall corresponded to transposition of the great vessels, followed by hydrocephalus, hypoplastic left heart, and severe CHD as a group. Time trends were not significant for rare CAs of the digestive system, as well as neural tube defects, and chromosomal abnormalities (not elsewhere classified) and trisomy 18 in males, and rare CAs of the respiratory system and urinary system in women (Table 2).      Table 3 shows the districts in which SMRs differed from the expected value for Spain along the 15 years studied (SMR = 1.00). As can be seen from the values registered for both sexes, lower-than-expected mortality was detected in 14 districts situated in the provinces of Alicante, Badajoz, Balearic Islands, Barcelona, Girona, Madrid, Pontevedra, Cantabria and Toledo. SMRs were higher than expected in 24 districts: of these, 20 (83.3%) corresponded to provinces lying in the south of Spain (Almería, Cádiz, Córdoba, Granada, Jaén, Málaga, Las Palmas, Tenerife, Seville, Ceuta and Murcia), two corresponded to provinces lying in the north (Asturias and León), and two corresponded to provinces lying in the west (Cáceres and Badajoz).  Detailed mapping made it easier to monitor spatial differences in risk of death due to rare CAs. Figure 2 depicts the geographical variability in smoothed SMRs, taking into account mortality registered in each district and its adjacent districts. According to the PP values associated with these smoothed SMRs, risk of death due to rare CAs was significantly higher than expected in districts situated in the south of Spain, with some exceptions (Figure 3). This geographical pattern remained unchanged when males and females were analysed separately.

Geographical Distribution
Lastly, the above mapping exercise was completed by identification of spatial clustering, with a higher-than-expected risk of mortality being detected in 6 clusters of municipalities situated in the south, and a lower risk being identified in the north (3 clusters), east (1 cluster) and centre (1 cluster) of the country. As before, the geographical distribution of clusters from both sexes combined, remained unchanged when males and females were analysed separately, and less significant groups were identified (Figure 4).

Discussion
Mortality is a major health-status indicator which is fairly well monitored for some common diseases [23]. Unfortunately, there is still a sizeable knowledge gap for large groups of rare diseases, and too many aspects relating to rare-disease mortality remain unknown. Rare CAs are not an exception, and not many studies have addressed the associated mortality. Such research is essential because, while mortality remains unknown, any interpretation of lifetime prevalence could well prove misleading. Failure to consider mortality data could lead to the conclusion that a disease was negligible if it caused death very early in life. This, in turn, could prevent the allocation of the precise health resources that could increase survival or survival under better conditions. This population-based study on rare CAs shows the continuous downward trend in CA-related mortality and the geographical distribution of risk of death from these causes in Spain. In addition, it confirms that most deaths occur below 5 years of age, with the first year of life accounting for half of all CA-related deaths. Moreover, the average age at death caused by rare CAs corresponds to early adulthood (around 20 years old). Although such average age has significantly increased along the time (in accordance to the downward trend in CA-related mortality), it seems clear that greater efforts

Discussion
Mortality is a major health-status indicator which is fairly well monitored for some common diseases [23]. Unfortunately, there is still a sizeable knowledge gap for large groups of rare diseases, and too many aspects relating to rare-disease mortality remain unknown. Rare CAs are not an exception, and not many studies have addressed the associated mortality. Such research is essential because, while mortality remains unknown, any interpretation of lifetime prevalence could well prove misleading. Failure to consider mortality data could lead to the conclusion that a disease was negligible if it caused death very early in life. This, in turn, could prevent the allocation of the precise health resources that could increase survival or survival under better conditions. This population-based study on rare CAs shows the continuous downward trend in CA-related mortality and the geographical distribution of risk of death from these causes in Spain. In addition, it confirms that most deaths occur below 5 years of age, with the first year of life accounting for half of all CA-related deaths. Moreover, the average age at death caused by rare CAs corresponds to early adulthood (around 20 years old). Although such average age has significantly increased along the time (in accordance to the downward trend in CA-related mortality), it seems clear that greater efforts are needed to ascertain the exact determinants of such early death, in order to establish the most appropriate prevention measures. Nevertheless, this scenario is not stable. In a matter of just 15 years, the average age of CA-related death has doubled, which is a promising sign. This goes to show that, insofar as CAs are concerned, a change is taking place in the Spanish population. Moreover, given the degree of geographical heterogeneity that was detected, the determinants of CA-related mortality may not be uniformly distributed across the country, and special attention should therefore be paid to areas with the highest rates, and even to those with the lowest rates because, to some extent, such areas might serve as models. Comparison of these two types of areas might conceivably yield some clues for prevention.
One of our study's most relevant findings is the above mentioned observed fall in age-adjusted mortality rates from 1999 to 2013 across all the different subgroups of rare CAs (with the exception of those of the digestive system, for which no statistically significant decrease was found). This decrease (for which we ruled out methodological issues as a cause) is in line with the increase in the average age at death attributed to rare CAs, since both are interrelated. Both findings could also be due to a number of other reasons. The treatment and care of patients with CAs-even prenatally-likely had an influence on the mortality figures, though lack of data means that this cannot be quantified. Similarly, advances in prenatal diagnosis make for better preparedness and the referral of the deliveries of severely affected pregnancies to tertiary hospitals, where more adequate care can be provided at birth and during the neonatal period. It has been shown, for instance, that prenatal diagnosis of congenital heart defects allows for early preemptive stabilisation, and is associated with improved early clinical status [24]. Improved prenatal diagnosis also must have an impact, in the sense that better detection of CAs in the foetus increases the likelihood of interruption of affected pregnancies. This means that a higher number of elective terminations of pregnancy due to foetal anomalies (ETOPFA) will reduce the number of affected newborn infants, and this may in turn affect the mortality figures. In fact, infant CA mortality in a given country is higher when prevalence of ETOPFA is lower, and it thus follows that increases over time in the ETOPFA rate would tend to lower the infant mortality rate [25]. Furthermore, the severity of the defects has to be considered, in that the most severely affected cases (those with a higher risk of postnatal death) will more probably be detected prenatally, with a considerable number of subsequent ETOPFA, thereby also influencing mortality figures by reducing them.
It should be said that, based on EUROCAT data, the prevalence of some CAs, particularly severe congenital heart defects, was reported to have increased in Europe, from 2004 to 2012 [25]. While these data also included ETOPFA, the number or proportion of ETOPFA among the cases was not specified. Hence, the increase per se could not be taken to mean that the number of newborn infants with these types of defects also increased. The authors speculated that this might reflect increases in maternal obesity and diabetes, both of which are well-known risk factors for CAs. For the purposes of our study, if the number of ETOPFA was high, this could in fact have reduced the postnatal mortality figure. This is a good example of the complexity of the situation and its interpretation.
It is noteworthy that 56.1% of deaths attributed to rare CAs occur before the age of 5 years, and even more striking that 49.9% occur in the first year of life. This is an important item of information because it narrows down the age at which the risk of death is highest, and consequently the age at which follow-up should somehow be different, so as to ensure that the determinants of early death are properly approached. Furthermore, it provides some evidence of the need for strategic allocation of resources specifically required for that segment of the population. This facilitates analysis and identification of the hospital(s) and local or district services that should be reinforced, through strengthening the workforce and/or providing adequate materials and infrastructure. Another strategy for reducing CA-related mortality could lie in designating some national reference hospitals, services and units specialised in the care of specific diseases (here in Spain the equivalent designation is known as Reference Centres, Services, and Units of the National Health System; Centros, Servicios y Unidades de Referencia, CSUR). There are data in the literature showing, for instance, that the lower number of patients attended at heart surgery centres is associated with higher neonatal mortality among cases with transposition of the great arteries [26].
In terms of geographical distribution, risk of death was observed to be consistently higher in the south of Spain, and specifically in certain districts. While this type of epidemiological finding can sometimes generate concern among the population [11], it should preferably be seen as indicating where in-depth research would be needed into the causes of that distribution of mortality. Further analyses, beyond the scope of this study, could reveal some possible ways of minimising the risk.
Our findings could prove useful as a base for comparison with other countries, or as a reference in time to assess possible changes and the impact of different measures in the future. In contrast to other approaches that use multisource data-integration for rare-CA prevalence estimates [7], we propose that the results of this nationwide registry-based study on mortality be used as a potential contributor for the monitoring of rare CAs. This could be a simple, effective, complementary way of improving epidemiological surveillance in countries addressing current difficulties in pooling cases from different registries (i.e., individual linkage between CAs and rare-disease registries) although of course, mortality statistics should not be considered an optimal source of data for case identification. It could also benefit those programmes that form part of the EUROCAT network but have incomplete geographical coverage of their country, such as the population-based CA registries in Spain [4].
Regarding the clinical impact of CAs, it is evident the influence of the important cost associated with birth defects [27], as well as the increase in disability adjusted life years (DALYs) rates and years lived with disability (YLDs) for CAs [1,28]. In addition, terminations of pregnancy for CAs were almost three times more frequent than the combination of infant deaths and stillbirths with CA, which clearly must affect the global burden of disease [17], and this should be taken into account when interpreting any figure.
This retrospective descriptive study has several limitations, such as the inability to link exposure to outcome in individuals, and to control for confounding factors. Therefore, it cannot be used to determine an association between a risk factor and disease. Consequently, additional research in this sense is needed for rare CAs.
Although the use of underlying cause of death underestimates case identification when compared to multiple-cause analysis, it is nonetheless an effective approach to mortality directly attributable to rare CAs. Some authors have estimated that mortality due to congenital anomalies for the under-5 age group is likely to be a fourfold underestimate [29]. If this were also applied to Spain, it would mean that the situation could be rather striking, something that yet again would make it advisable to focus attention on this younger stratum of the population. On the other hand, differences in diagnostic quality or coding practices over time or among regions might bias our results, even though the death registry officially follows a standardised, uniform methodology.
The use of ICD for rare diseases research is challenging and in fact the lack of appropriate coding makes difficult, and sometimes impossible, the study of a particular CA without a specific ICD code [8]. In this paper, the analysis of rare CAs (globally and by system) provides a general view of this public health problem, even though some misclassification issues cannot completely be ruled out.
Considering quality issues of deaths certificates, the differences and incompatibilities between original underlying cause of death and final main condition were assessed previously [30]. In that study, Johansson and Westerling reported the lowest percentage of differences for CAs, in comparison to other ICD chapters, which is reassuring.
Despite these limitations mortality statistics provide broad temporal and geographical coverage and continue to be a very useful tool for studying the epidemiology of low-prevalence diseases [9,20], either for epidemiological research as well as for health monitoring [30]. In addition, mortality databases enhance the ability to collect cases diagnosed after the early neonatal period (lifetime detection), thereby becoming a complementary data source for rare-CA studies.

Conclusions
In conclusion, this is the first nationwide population-based study to focus on mortality due to rare CAs. Our results contribute to the monitoring of rare CAs along 15 years in Spain, by providing evidence of the continuous decline in mortality rates and illustrate some geographical differences in the risk of death. These findings are not only useful for assessing the burden of low-prevalence CAs in Spain, but also serve as evidence which might be taken into account by health authorities, in order to identify possible risk factors, adopt appropriate preventive measures, implement and evaluate health policies and healthcare plans ensuring equality and equity, with the ultimate purpose of achieving better health for all.