Interpretation of Surveillance Studies

Obtaining data for surveillance can be deceptively simple in concept: Simply collate information from diagnostic laboratories. However, interpreting that data is often not straightforward due to bias in the data collection process. For example, most samples sent to diagnostic laboratories are from hospitals and are thus biased toward patients who are quite ill. Another problem concerns longitudinal studies. The most informative surveillance reveals trends over time, because they enable policy decisions to be made. For such work, one needs the same patient population each year. Problems associated with interpretation and design of surveillance studies are sketched in Box 8-4.