Figure 1.

Study index period and the start of follow-up. A) gives an example timeline for case enrolment to the cohort. Eligibility was determined by the presence of a previous COPD medcode and at least 1-year history of GP consultations. This is illustrated using a red circle and covariate values used within regression models are established as closely as possible to the point of case enrolment. In this case, the medcode was a new entry but for many cases in the cohort, the code for COPD was present at the onset of the CPRD cohort. The example shows a period of 1-year prior and a month after enrolment within which values were accepted. Outcomes of all-cause hospitalisation and mortality were recorded 12 months after the point of enrolment. (B) illustrates that the timespan of analysable patient data is highly variable depending on whether COPD was present at the onset of the CPRD cohort or diagnosed later during follow-up and whether the patient left the cohort (eg due to leaving the practice or death.