Also, consider how you plan to synthesize the outcome data you’re extracting. If you’re clear about the data you’ll need for your final analysis, you can avoid collecting information you won’t use. For instance, if you know you aren’t interested in outcomes related to high blood pressure, don’t bother extracting data on that outcome.
The reverse is also true. If you don’t think about the specific types of calculations you plan to make, you might fail to collect the raw numbers you need to standardize your results.
For example, If you only collect odds ratios and don’t extract the raw data that could be used to make this calculation, you won’t be able to reconstruct that data for any studies that didn’t report odds ratios.
Every study will have a slightly different approach to reporting data, so your data extraction form will need to include a place for both raw numbers and calculated results to capture the range of reporting methods.
In general, try to capture data in as raw a form as possible so that you can reconstruct anything you need.