Study and comparison cohorts can be chosen from special groups, exposure groups, hospital registers, medical records, and death certificates etc.

When choosing the special groups like doctors or occupation groups, special care to be taken regarding their representativeness.

When registers and records were used for selecting a cohort e.g. historical cohort, relevant and complete records if available are to be used.

Naturally, selection bias will creep in, as there is no randomization.