Tracking patient outcomes using cystic fibrosis (CF) national data registries, we have seen a dramatic improvement in patient survival. As there are multiple ways to measure survival, it is important for readers to understand these different metrics in order to clearly translate this information to patients and their families. The aims of this review were to describe measures of survival and to review the recent literature pertaining to survival in CF to capture the changing epidemiology.
Although survival has improved on a population level, several individual factors continue to impact survival such as sex, age of diagnosis, ethnic background and lung function. Survival estimates, conditional on surviving to a specified age, are more relevant to individuals living with CF today and are higher than the reported overall median age of survival. There is some evidence to suggest that newborn screening (NBS) has resulted in prolonged survival in CF.
Prognosis in CF is often described by reporting the median age of survival, the median age of death, the median survival conditional on living to a certain age and the survival by birth cohort. Each of these metrics provide useful information depending on an individual’s personal disease trajectory. The median age of survival continues to increase in CF in many countries while mortality rates are decreasing. Several factors have been associated with worse survival such as female sex, ethnicity, worse nutritional status, lower lung function and microbiology. When comparing survival between countries, one needs to ensure that similar data collection and processing techniques are used to ensure valid and robust comparisons.
-
A key metric in summarizing CF survival is the median age of survival, which is the age at which 50% of patients are expected to live beyond.
-
As the median age of survival is only applicable for babies born today, conditional survival (the age at which 50% of patients are expected to live given that they have already survived to a certain age) is a useful measure for patients currently living with CF.
-
International comparisons of survival using registry data require a unified approach with harmonization of the variable definitions and data processing in order to produce robust results.
-
Factors determined to be associated with an increased risk of death include lower FEV1% predicted, infection with Burkholderia cepacia complex, gender, increased number of pulmonary exacerbations and lower nutritional status.
-
Survival in CF has increased greatly in the last decade and is expected to continue to improve with the advent of CFTR modulators and the introduction of newborn screening.
Median age at death: based on only the CF decedents and is the age at which 50% of the deaths that occurred were younger and 50% were older. The years of life lived for those still alive are not considered in this calculation. Median age at death fluctuates with the age structure of the population and will increase as the percentage of older patients within the cohort increase. The median age at death is less meaningful when the population’s mortality is low, and it is difficult to compare across countries because it depends on the current age distribution of that population.
Median age of survival: the age past which 50% of the population is expected to live. This calculation takes into consideration both those who have died and those who are still living. The median age of survival is most often used by national CF registries to monitor temporal trends in survival. It should not be mistaken for life expectancy, which refers to the average or mean age of survival.
Most often used by CF registries to estimate median age of survival for the population. Estimates a survivor curve using mortality rates at each age for people observed in the registry during a specified time period. Shorter time periods result in higher variation; very long time period may not accurately reflect current survival trends due to changes in standard of care. It has been proposed that a 5-year window produces more stable estimates.
The mortality rate at each age is estimated by the number of people alive at that age during the study period who died before their next birthday, divided by the total number of people alive at that age. The probability of surviving to each age is then calculated and the median age of survival is the age at which the probability of surviving beyond that age for the time period under study is 50%. In order to understand the trend over time, one would conduct this calculation using rolling 5-year windows for successive years.
The birth cohort method is a longitudinal method in which survival is estimated for individuals born in a specified time period. This method is useful to track the changes in survival of successive birth cohorts, to show how survival has changed. With this approach, individuals are followed from their age of diagnosis (typically birth). Second, with each successive birth cohort, the curve declines at a slower rate suggesting that the survival at any given age is better for more recent birth cohorts.
It is not possible to report the median survival age for all birth cohorts, as they need to be followed for longer to be able to calculate the median survival. Therefore, the median age of survival estimates is not necessarily applicable for babies born with CF today, as recent births will not have an estimate for several years in the future.
Conditional survival defines the probability of patients surviving ‘x’ number of years given the patient has already survived ‘y’ number of years.
Population-level survival estimates do not account for individual patient characteristics that impact survival. Prior literature has identified several prognostic variables linked to survival, which include CF transmembrane conductance regulator (CFTR) genotype, initial disease presentation, pulmonary function, nutritional status, sputum bacteriology, age, socioeconomic status, pulmonary exacerbations and sex.