| Panel cohort study | Prospective measurement of a range of air pollutants | Individual clinical data may not be available |
| Larger sample sizes possible | Possibly smaller effect sizes (population based) |
| Longer time course possible | Confounding factors present |
| | Difficult to assess biological phenotypes |
| | |
| Controlled exposure study of volunteers | Controlled environment and population, reduces confounders | Generally smaller sample sizes |
| Can study high risk populations | Multiple pollutants usually not studied |
| More practical to assess biological phenotypes | Relatively short time course |
| | |
| In vitro experiments of biological samples from volunteers | Controlled environment and population, reduces confounders | Generally smaller sample sizes |
| Detailed mechanistic study possible (eg, gene expression microarrays, proteomics) | Relatively short time course |
| | May not entirely reflect in vivo situation |
| | Requires biological samples to be obtained |
| | |
| Randomised controlled trial | Prospective interventional trial | Often need larger sample sizes and longer duration to show a clinically important difference |
| Can target high risk individuals | Needs replication to be generalisable to other settings |
| Directly applicable to clinical practice | |