Identify Data Sources
There are three key things to consider when choosing data sources:
Which source is likely to provide the best information?
For example, you could gather information about blood pressure levels by asking patients to recall them; however, their medical records might be a more accurate source.
How easy or difficult will it be to gather the information you need from a particular source?
Building on the previous example, how difficult will it be to access medical records? And, if you can access them, how much time and money will be needed to do a medical records review? Also, using medical records may require a human subjects' review by an Institutional Review Board (IRB), which will add to the time it takes to collect this data. Alternatively, a question about blood pressure could be easily added to a patient questionnaire. You will have to weigh the level of accuracy needed against the degree of difficulty in obtaining the information.
Is there more than one source for a particular item of information?
Using more than one source provides a check and balance on your data, which can help you avoid bias in your results—or at least call it out if it is there.
In the example, if you compared self-reported blood pressure levels with those reported in patients' charts, you would be able to detect the degree to which self-reported data were accurate. Once you have a sense of how well patients' self-reports reflect chart records, you may be able to use patient questionnaires for follow-up tracking.