It can take time to recruit participants into a study, whether observational or interventional and sometimes there are differences between those recruited earlier in the process and those recruited later; this constitutes chronological bias. When there is a relationship between the time of recruitment and the observed outcome, or another factor causative of the outcome, chronological bias can cause confounding of observed associations and effects. Chronological bias can also occur as a consequence of changing disease definitions or categorizations over time, and if the abilities of the clinicians undertaking the diagnosing change.
As an example, grading of prostate cancer can change depending on the scoring system used and how people apply it. In a study of how clinicians graded the severity of disease based on samples taken from patients, the estimated severity differed according to when the categorisation was done: in more recent years, more samples were categorised as being indicative of higher grade disease.
Prostate cancer grade assignment: the effect of chronological, interpretive and translation bias. J Urol 2003
Another problem can arise if inclusion criteria or outcome assessment methods change during the study so that participants recruited earlier differ in a systematic way from those recruited later.
Time trends may be more likely to affect studies with slow recruitment rates, for example in studies of rare conditions. Chronological bias can also occur when a study uses historical controls to provide a comparison for a set of cases: known or unknown factors relating to the time of diagnosis or recruitment of these cases and controls may bias observed relationships.
Chronological bias has been investigated in randomised controlled trials, specifically looking at the extent of time-related bias in trials in which block randomisation is used. Chronological bias can strongly influence estimates of treatment effects if randomisation sequences are used that are not balanced over time. Undertaking block randomisation with small sample blocks can reduce this bias, but this needs to be balanced against the problems that small block sizes can lead to.
Chronological bias in randomized clinical trials arising from different types of unobserved time trends. Methods Inf Med. 2014.
In randomised trials, using small block sizes for randomization can reduce chronological bias, but must be balanced against the resulting risk of selection bias. Analysis of trials should, therefore, include assessment of the possibility of chronological bias. If it is found, attempts should be made to adjust for it analytically. In observational data, it is essential to analyse any differences in exposure, treatment, or diagnostic criteria that may have varied over time and could affect the results.