Table 2.

Gathering information on search volume.

Step	Description
2.1	For each region, for every term in the master list, send a request to the getTimelinesForHealth function from the Google Health Trends application programming interface (API) to obtain relative search volume. Figure MA2-4 in Multimedia Appendix 2 shows an example of the API call in Python.
2.2	The process in Step 2.1 should be repeated 30 times to minimize error. We take the average of the 30 samples of relative search volumes, which represents the estimated search volume for a given term with the date and location restrictions provided. Figure MA2-5 in Multimedia Appendix 2 shows a sample response of relative search volumes given by the getTimelinesForHealth function in the United States.
2.3	To compare across regions, normalize the values for each region by dividing each term’s value with the aggregate search volume for the region. Before normalization, the value returned is skewed and does not take into account parameters such as geographical size differences. The normalized value will range from 0 to 1. The total sum of all values of the set of queries is 1 after normalization. The value from 0 to 1 allows for understanding of the relative search frequency within search queries. These data can then be used to define search frequencies for each term (see Figure 3). The normalization function used in this study is found in Figure MA2-6 in Multimedia Appendix 2.