The target population includes all the students who went to school near to the facility when the facility was operating. It may be possible to find them, but this proposal presents some difficulties. Residents and researchers would have to find the names of the people who attended school during the 30 years that the facility was operating. During this time period, each student only attended the school for three years.. One option could be looking up old school yearbooks. Although the yearbooks would most likely not provide records of students' birth dates, the researchers could still deduct students' birth year within a year or so.
The target population could be also very difficult to locate depending on the researchers' ability to obtain their correct name. Some female students change their names when they married, which would make them hard to locate. Their full names and identifying information were necessary in order to search for them in the cancer registry to see if they had gotten cancer. If researchers could not get the correct names for some people from school databases, they would not be able to find out if these residents had cancer by reviewing the cancer registry.
In this type of study, determining the level of exposure to residents is very important and often difficult to do. In this situation, some information was available to estimate exposures. Researchers knew how much of the chemical was emitted into the air and used special techniques (called modeling) to predict how the chemical might have spread in the air. Through modeling, researchers generated estimates of average exposure levels of the students at the school. However, even a best estimate is still an approximation of what a student's exposure might have been.
If researchers are studying cancer, then they would use information from the cancer registry, which has a record of all cancer cases that are diagnosed in California. This registry is very accurate. However, if someone moved out of California, it would become a challenge to find out that information. Researchers would have to search other data sources, possibly from other states. There is also the possibility that not all cancer cases were diagnosed. For example, some of the target population may have died from another cause, such as a car accident, before being diagnosed with cancer. Another example is that the cancer may not yet have been discovered in some of the target population at the time of the study.
This brings up the issue called "confounding." Study results could be distorted if researchers mistakenly concluded that exposure caused the diseases when in fact there was some other factor that caused or contributed to the causation of the disease. Researchers have to try to search for possible confounders and make sure, as best possible, that they are not the cause for the health outcome.
For example, the most common factor that causes lung cancer is smoking. Suppose that students attending this school ended up having higher rates of smoking as adults than most Californians. Then, increased rates of lung cancer may be primarily due to high rates of smoking rather than exposure to the air contaminant. Researchers can attempt to obtain smoking information for each student through a study questionnaire. Sending a questionnaire to everyone can be costly and time-consuming; residents and researchers would need enough time and funding to build this into their study. Even then, when the questionnaire is delivered, there is a possibility that many people may not receive or answer it.
Researchers could attempt to indirectly estimate how many people smoke in a community by looking at socioeconomic status of a population. Smoking is more common among people of lower socioeconomic status. Researchers could obtain socioeconomic information, such as income and education level of town residents, from the U.S. Census. Then, this information could be taken into account during the data analysis to account for differences in lung cancer rates that are possibly due to smoking habits. There are limitations to this approach, because people's socioeconomic status may change.
Even if researchers are able to find the correct people for this study,
they still have to determine if the study is likely to find a relationship between contamination and disease, if this relationship actually exists. This ability is known as "
study power."
Study power is not the only consideration when deciding whether to conduct a health study. However, it is a very important component of this decision. Conducting a study requires a great deal of time and money. Consequently, instead of devoting much of the available resources to a study with little chance of helping the community, the community and the researchers may want to instead put these resources towards other activities that would address community concerns about environmental contamination. Learn more about these types of activities.
In order to calculate study power, researchers had to answer these three questions:
- How many people are available for the study?
- How common is this health outcome?
- How many more cancer cases are estimated to be in the exposed group compared to what is normally expected for this population?
Learn more about how these factors affect study power.