Most annual CF registry reports give the median survival age. For example, the UK CF Registry 2016 Annual Data Report gives a median survival age of 47 years old. But what does this mean? My colleague Dr Sanja Stanojevic, from Sick Kids, Toronto, and I noticed that there is some confusion about the interpretation of ‘median survival age’, and that different national CF registries presented and explained the figure in different ways.
What is a national CF registry?
A national CF registry is a centralised database that collects clinical care and health outcome information from consenting people with CF in that country. Many countries across the world have a national CF registry, including the UK, USA, Australia, Canada, Brazil, France and Germany. Each of those registries publishes annual reports to give a picture of CF in that country each year.
How is the median age of survival calculated?
In our project we set out to clear up some of this confusion by providing a clear explanation of the meaning of the ‘median survival age’ and to make recommendations to registries about how they report information about survival. Our work has just been published in the Journal of Cystic Fibrosis and is freely available.
The median age of survival is calculated from a survivor curve. This statistical graph (right) shows the chances of living beyond a specific age and the median survival age is the age at which the survival chances are 50% (probability 0.5). The graph illustrates a survivor curve based on data from the UK CF Registry, indicating the median survival age of 47.
What does that mean?
The interpretation of this number is that we expect 50% of babies with CF born in the UK today to live to age 47. Importantly, this estimate is based on an assumption that the current mortality rates at each age will be the same in the future. This means that the estimate does not take into account recent and future improvements in care and treatment that may bring improved survival.
The median age of survival, as calculated, is useful for families of babies born today with CF, as well as for health care providers and policy makers planning future resource needs. However, information on survival from birth is not so useful for individuals who have already reached a certain age.
What about older people with cystic fibrosis?
In the paper we explain how ‘conditional survivor curves’ can be used to provide more relevant information for people who have already reached a given age. For example, based on analyses of the UK CF Registry data, we found that 50% of people living with CF aged 30 today are expected to live beyond age 55.
What are the limitations?
A limitation of the median survival age is that it is not person-specific, that is it does not take into account the many features of a person that impact on their expected survival. In other recent work also published in the Journal of Cystic Fibrosis, we have provided more detailed information about CF survival in the UK taking into account a person’s current age, sex, genotype and age at diagnosis, and trends towards improved survival over time.
In further research with a number of colleagues I am working on providing more personalised information about survival that takes into account a person’s current health measures, for example lung function. We look forward to presenting those results soon.
Here comes the science!
There are two main ways of calculating the survivor curve. The first method is to follow a group of individuals from birth and observe how long they live. The second is to observe all individuals during a particular calendar period (eg the five years from 2012–2016). For both methods a mathematical formula uses information about how many people die, and at what age they die.
Both methods also take into account that some people might move away, and that many individuals are still alive. The two approaches have advantages and disadvantages. The second approach is what is used to give the median survival ages presented in annual registry reports. The shaded areas in the graph above represent the uncertainty in the estimated survivor curve, which is also important to consider.
Sign up to our e-newsletter