Cystic fibrosis (CF; OMIM#219700) is an autosomal recessive disorder due to mutations in the CFTR (Cystic Fibrosis Transmembrane conductance Regulator) gene, which was cloned in 1989 1. Newborn screening (NBS) for CF, which has progressively been implemented worldwide over the past years, now allows better monitoring of the birth incidence of CF, which appears lower than in the past. Long estimated as 4.0/10,000 (i.e. 1/2500) live births in Caucasians 2, the average birth incidence rate is now 2.9/10,000 (i.e. 1/3500) in Europe 34.

Time trends in the birth incidence of CF have been investigated in several studies, but the magnitude of the reported changes varies according to the areas and to the length of the period under study 5678. The trends observed result from a complex mixture of phenomena and have been ascribed to various health policies (as prenatal diagnosis, family testing 9, population-based carrier screening 101112) and/or to demographic factors (as reduction of family size, increase of population mixing with consecutive decline of endogamy and inbreeding 8).

Time trends in the birth incidence therefore depend on the health policies implemented in the studied population, but also on the characteristics of the population (i.e. carrier rate, cultural attitudes toward prenatal diagnosis and pregnancy termination, birth rate, etc). Monitoring of CF incidence is therefore essential to be able to make reliable predictions for disease management in the future.

In this study, we examined time trends in the incidence of CF over a long period (the longest ever studied to date: 35 years, 1975-2009) in an area where CF is frequent (Brittany, western France), where NBS has been implemented for a long time (since 1989) but where carrier screening (both preconceptional screening and prenatal screening) is not underway.

This study deciphers how the incidence of CF has evolved over a long time (the longest ever studied to date: 35 years) in an area where CF is frequent (Brittany, France) and where population-based carrier screening-known to significantly decrease the incidence 12-is not underway. It also highlights how the incidence is influenced by the health policies implemented.

This study reveals a clear fall in the incidence rate, which dropped by 40% between the period 1975-79 and 2005-09. The observed breakpoint coincides with the more common use of prenatal diagnosis since the end of the 1980's. This test has been increasingly used notably by couples identified to be at risk following NBS, which was implemented in our area in 1989. This study also highlights the importance of ultrasound monitoring performed during pregnancy, which may identify new 1-in-4 risk couples following detection of echogenic bowel.

We have been able to measure accurately the impact of prenatal diagnosis in our area as we have an exhaustive collection of its data. By including pregnancy terminations in the calculations, we have shown that the incidence rate would have been 35.8% higher than that observed to date. The impact of this test appears particularly high in Brittany. Indeed, despite the progress accomplished over the past decades in the field of CF, most 1-in-4 risk couples in our area request prenatal diagnosis and 95% choose not to continue the pregnancy if the fetus is affected 21.

However, the impact of prenatal diagnosis on incidence rate remains relatively limited in areas where population-based carrier screening is not implemented. Indeed, this test can only be offered to 1-in-4 risk couples related to a CF patient or to a carrier child identified through NBS. Therefore, most carrier couples are at risk without knowing it, and can give birth to a CF child. In our area, one individual in 29 is a healthy carrier. Only the implementation of carrier screening may significantly reduce the incidence, as recently shown in an Italian area 12. Population-based carrier screening consists in offering genetic testing to all the couples planning a pregnancy in a given population. Aware of their high risk, the couples found to have a 1-in-4 risk may opt for prenatal diagnosis and may choose to terminate the pregnancy if the fetus is CF-affected. Such a program is not underway in our area, and this is probably why the incidence has remained stable since the early 1990's.

The present study provides an accurate picture of the evolution of the incidence of CF on the scale of an entire population over a long period. It relies on exhaustive collection of data in a large region whose geographical situation (peninsula) means that the data are well centralized and easier to collect. A few biases may have affected our study. Since 1989, an NBS program has been implemented in our area, which ensures quasi-exhaustive registration of new cases at birth. False-negative cases are always a possibility, but the cases reported so far (which are few in number in the last decade) have been included in the calculations (n = 10). Newborn screening may also diagnose atypical cases that do not meet all the criteria for CF 2223. To overcome this problem, which may alter time trends, equivocal diagnoses were excluded from the analysis 15. Another pitfall that could affect our findings is under-reporting and/or under-diagnosis of cases predating the setting of NBS. The active combination of data sources and the use of capture-recapture studies may have limited this phenomenon. Anyway, in such a situation, the incidence rate would have been still greater in that period, and the observed drop still higher.

Several studies have analyzed time trends in the incidence of CF. While most have reported a decline, some have not observed any temporal change 24. The findings are however difficult to compare because the studies have been conducted over various periods and in areas that have not implemented similar public health policies and that have different attitudes towards prenatal diagnosis.

The oldest studies were conducted from registry data (prone to under-diagnosis and/or under-reporting), mainly in countries that did not practice NBS. In Canada, time trends were analyzed over the period 1971-2000 and a linear decline was noted since 1988, which coincided with the increased use of family testing and prenatal diagnosis services 6. In the Netherlands, the incidence rate observed over the period 1974-1997 (2.1/10,000 i.e. 1/4780) was found to be significantly lower than that estimated over a previous period (1961-1965: 2.9/10,000 i.e. 1/3600) 725. In the UK, a constant decline in incidence rate was also observed over a 26-year period (1968-1994) from data of the UK Registry 5.

More recent studies have analyzed time trends in the incidence rate since the implementation of NBS 8. The region of East Anglia (UK), where NBS has been in place since 1981, experienced a constant decline over the period 1981-1990. In the State of Victoria (Australia), a 17% reduction in the birth incidence of CF was seen following the introduction of NBS in 1989 26. The incidence dropped from 3.96/10,000 (i.e. 1/2525) live births over the period 1979-1988 to 3.28/10,000 (i.e. 1/3050) over the period 1989-2006. This was attributed to the uptake of prenatal diagnosis, especially in families identified by NBS. Brittany appears to be similar in various respects, such as the date of implementation of an NBS program, the attitudes towards request of pregnancy terminations, etc. Nevertheless, the decline we observed in our area is rather higher and may, in part, result from the greater uptake of family testing and the efficiency of routine ultrasound monitoring of pregnancy, which is, in our country, supported by our health care system.

In the US, a significant decline has been observed in the States of Wisconsin and of Massachusetts since the implementation of NBS (in 1994 and 1999, respectively) 1127, whereas no temporal trend has been noted in the State of Colorado, which has been screening newborns for longer (since 1982) 24. In Massachusetts there has also been reported a change in the composition of the screened cohort (notably a decrease in the proportion of the main F508del/F508del genotype) which coincides with the implementation of carrier screening 11. As in Colorado, we did not find such a change in our population over time.

A greater decline has recently been found in an area of northern Italy (Veneto/Trentino Alto Adige) over the period 1993-2007, particularly in the region where carrier screening is underway (mean annual decrease of 0.24/10,000) 12. The authors have also noted a significant correlation between the increase in the number of 1-in-4 risk couples identified and the decline in incidence.

This study highlights that the time trends observed in the incidence rate depend on the health policies implemented in the studied population (i.e. newborn screening, prenatal diagnosis, carrier screening, etc), but also on the characteristics of the population (i.e. carrier rate, cultural attitudes toward prenatal diagnosis and pregnancy termination, birth rate, etc).

We show, in this study, that the birth incidence of CF has dropped in our area following the implementation of prenatal diagnosis. However, the birth incidence appears now quite stable. The sole strategy that could strongly modify the current incidence of CF would be the routine carrier screening of couples planning a pregnancy. Implementation of such a program has proven successful in reducing the incidence of CF in the State of Massachusetts 11 or in Northern Italy 12, but also the incidence of other diseases, such as thalassemia 28.

Improvement in survival of CF patients as well as in assisted reproduction techniques should help alter the disease dynamics in the future. It is therefore essential to keep on monitoring the incidence to be able to make reliable disease predictions. It is also crucial to maintain and strengthen efforts to improve clinical management and patients' quality of life.

APC: Average percent change; CF: Cystic fibrosis; CFTR: Cystic Fibrosis Transmembrane conductance Regulator; NBS: Newborn screening; OR: Odds-ratio; 95% CI: 95% confidence interval.

The authors declare that they have no competing interests.

VS and CF conceived and designed the study. VS carried out the statistical analysis and drafted the manuscript. ID managed data collection and participated with PS to data analysis and interpretation. MPA, GR and MR contributed to data collection. All authors revised the manuscript critically for important intellectual content and approved the final version.