Following licensure of a vaccine, typically based on a demonstration of efficacy in a clinical efficacy trial, almost all subsequent research and regulatory approvals – extensions of indications, approvals in other jurisdictions, development of combination vaccines, duration of protection, assessment of second-generation vaccines – are based on a 'correlate of protection', a threshold level of an immunological assay which is believed to differentiate individuals protected against disease from those susceptible. However, the scientific basis and properties of many established correlates of protection are unclear. Do they represent a ‘population average’ level of immunological assay value required for protection, or a level at which, say, 90% of individuals are protected? Are they a level to be achieved post-vaccination, when the immure response is likely to be at its highest, or a value that indicates protection at the time of exposure? Is the threshold required for protection the same for infants, children, adolescents, adults and elderly?

Standard statistical methods have not always provided clear answers to such questions. In addition, opportunities for collecting data based on which an assessment of the relationship between immunological assay value and disease can be made are limited. Post-licensure placebo-controlled clinical efficacy trials can be considered unethical, and active control trials have few cases of disease.

Some recent quantitative methods for estimating correlates of protection will be presented, factors to take into account in experimental designs considered, current research reviewed, and methodological and data challenges outlined.