Epidemiological Study Designs When Doing

Epidemiological Study Designs

When doing an epidemiological study there are specific designs that are most commonly used. Cohort studies are done by looking at a healthy group of people and then following them to see if they develop a particular disease over time (Goodman, Buehler, & Koplan, 1990). This helps provide information about both the progression and the onset of a disease. It is a prospective study which can only prevent bias if the study subjects and the researchers neither one have any idea who will get the disease or the odds of each person getting the disease. Studies like this can be very expensive, though, because some people have to be followed for many years to determine when and if they will contract a particular type of disease. (Goodman, Buehler, & Koplan, 1990)

Randomized controlled trials are another type of study that is conducted, and this kind of study is considered to be the gold standard (Goodman, Buehler, & Koplan, 1990). In this kind of study one group gets one treatment and another group does not. None of the people in the study are told whether they are the experimental group (the group getting the treatment) or the control group (the group not getting the treatment). It is best if the researchers are also not told which group is which. That way the study is double-blind and there are no biases on the part of the researchers (Goodman, Buehler, & Koplan, 1990). It provides the closest real-life scenario, but only if it is done correctly. This is costly, and it does not work well for everything. Human beings can also not be used in this type of trial if it is possible that they will be harmed by the experiment. When they can be conducted safely, however, these tests bring the best opportunity for a realistic outcome.

There are other studies, however, including the case series. The case series is the most basic kind of study that can be done, and it simply involves the researcher's description of an observed case. In other words, the patient's characteristics and presentation -- the story that unfolds as the case is addressed -- is all that is used (Goodman, Buehler, & Koplan, 1990). There is no control group, and there are no other patients that are observed. This kind of study is strictly about one specific patient and what he or she is doing at the time of study (Goodman, Buehler, & Koplan, 1990). The case is thoroughly described, though, so that it can be used to gain ideas and for reference in the future.

This case series type of study has a large margin for error when it comes to bias, though, because it is easy for a researcher to misinterpret something that the patient is doing. Without a control group, this becomes more likely (Yehuda & McFarlane, 1995). The case can provide several plausible factors or scenarios, and the researcher will not know which one of them is correct. It becomes guesswork, which is generally not a good choice where science is concerned. The use of it is limited and should only be used for observation and reference in the monitoring of new patients who seem to have cases that are unusual.

Ecological studies are also used. Instead of only looking at one patient or subject, this kind of study provides for comparison. This is not done between individuals, but rather between groups of individuals. It would seem like bias would be eliminated that way, but this is not the case. There can be associations found on both group and individual levels, and that can be misleading (Goodman, Buehler, & Koplan, 1990). A researcher who notices this kind of association can be led to believe that it is something taking place on a group level, where it should actually be attributed to the people in the group on an individual level. It is called ecological bias when this happens, but these studies still have their place. They are less expensive than many other types of studies, and they are also quicker, as they can often be conducted with information that has already been published by other researchers or with public records information (Yehuda & McFarlane, 1995).

The cross sectional study is designed to gather information from a point in time, and from a specific population. It is like taking a snapshot to look only at that group at that chosen time, rather than at a group over a period of time. The study population is very well-defined, and the characteristics of that population are studied at a precise point (Goodman, Buehler, & Koplan, 1990). One example of this kind of cross sectional study is the exit polling that is done right after someone votes in an election. The people for this type of study are only contacted one time and specific information is gathered.

The study is very inexpensive but there can be problems with it. For example, in the case of disease this kind of study can detect something taking place at that moment but it cannot detect why it is taking place, how long it has been ongoing, or where it came from, which are all things that can be very important (Goodman, Buehler, & Koplan, 1990). If a person has changed since contracting that disease, this also cannot be seen with this kind of study. It works well for some kinds of information, but it will not work well for all applications.