2009 French CF Registry Report

Question

Hello

I have a question about the graph (Figure 4) on page 4 of the french CF registry: what explains the lower probability of survival of the cohort 2007-2009 as compared to other birth cohorts? (dropping out of the curve between 2-3 years duration).

Answer

Hello,

Figure 4 page 4 of the 2009 French CF Registry expresses the length of life in terms of survival probabilities. It was calculated for four different cohorts of patients according to their year of birth: 1992-1996 (in 2009 this cohort had been followed for a maximum of 18 years), 1997-2001 (followed for 13 years maximum), 2002-2006 (followed for 8 years maximum) and 2007-2009 (followed for 3 years maximum). A total of 3,356 patients were included in the analysis, among whom 68 had died.

This method, known as the Kaplan-Meier, is widely used in epidemiology to identify the occurrence of a given event (ie, death) in a given population (ie. patients with cystic fibrosis born in France in the years corresponding to the four cohorts and reported to the French CF Registry) for a given time period (ie. 1 January 1992 to 31 December 2009).

Whenever a death occurs, the survival probability of the corresponding cohort is diminished, and this gradual decrease is materialized by a survival curve-shaped staircase.

As you noticed, the curve of the 2007-2009 cohort drops to reach a probability of survival (99.3%) lower than the 2002-2006 cohort (99.6%) even at age 3. This finding may indeed surprise since the two cohorts are subsequent to the widespread use of newborn screening and the implementation of accredited French CF Centers: it could be assumed that the most recent cohort benefits most from the new organization of care.

However, these so-called survival curves require clarification and caution in interpretation. Indeed, the difference in probability of survival at 3 years between these two cohorts is, on the one hand, extremely low (99.6 - 99.3 = 0.3% or 3 ‰), on the other, due to a very small number of deaths, which correspondingly increases the effect of chance. This difference, observed with a hindsight of only 3 years for the cohort 2007-2009, is therefore absolutely not significant. Indeed, the difference is not found any more over the age of 4 in the same figure of the upcoming 2010 report (due to the occurrence of several deaths in 2010 in the cohort 2002-2006 but none in the cohort 2007-2010).

So, hindsight is necessary to make statistically significant differences between cohorts. These precautions being taken, the fact remains that the probability of survival calculated for patients in cohorts of children born after the widespread neonatal screening in France (2002), on the one hand are reliable because of the existence of a comprehensive collection of information by the French CF Registry, on the other hand, are high with a life expectancy which will increase with the constant advances in treatment.

Hope that answers your question. Sincerely.

Gilles RAULT, MD

02.05.2012