Google Health API was queried for all searches related to HZ vaccination between January 1, 2011 and December 31, 2016. The relative frequency of searches for the terms “shingles vaccine,” “zoster vaccine,” and “zostavax” are shown in Figure 1. Trend lines for these terms exhibited a significant annual periodicity with a maximum in the fall (P<.001 by Fisher’s g test).

Our initial hypothesis on the possible cause of this seasonal trend is that it is related to insurance utilization. Since deductibles are generally reset on January 1^st in American health insurance systems, it could be that more HZ vaccine-eligible patients would seek out this elective immunization in the last few months of the year. Furthermore, the majority of patients receiving the HZ vaccine are Medicare-eligible. The open enrollment period for Medicare Advantage and the Medicare Drug Plan runs between mid-October and December [37]. Newly eligible patients may look online for ways to apply their new insurance policy immediately after they enroll. To test this hypothesis, we assessed the seasonality of searches for “medicare” (Figure 1, blue) and “medicare vaccine” (data not shown). These terms revealed a similar annual cycle as “shingles vaccine,” “zoster vaccine,” and “Zostavax,” with a significant seasonal trend peaking in the fall (P<.001 for both terms, Fisher’s g test).

In addition, it has been shown that covaccination for influenza and HZ is safe [38]. One could also hypothesize that the surge in interest for the HZ vaccine may be linked to the flu season. We assessed periodicity for searches for “flu vaccine” and found that it also exhibited a pronounced annual cycle (P<.001, Fisher’s g test) with a peak in the fall (data not shown).

We also considered that the seasonal pattern may be related to commercial advertising for the vaccine. Merck produced two widely viewed commercials featuring a popular former National Football League quarterback, initially released in September 2014 and 2015 [39,40]. These commercials may have led individuals to then search online for more information about the vaccine. However, this does not provide a complete explanation of the seasonal trend; Merck has also released other shingles-related commercials in February, April, May, June, July, and August.

In our analysis for a seasonal trend in HZ vaccination, we noted that there were certain time points at which the overall volume of searches for vaccination-related topics sharply increased. We inferred that these time points may correspond with key dates in the development, approval, and extension of the HZ vaccine. The FDA approved the live attenuated vaccine (Zostavax) for US patients aged 60 or older on May 25, 2006 and extended its recommendation to patients aged 50 or older on March 25, 2011 [41,42]. Furthermore, the CDC issued its recommendation for the live attenuated vaccine in adults aged 60 or older on October 25, 2006 [43]. Google Trends data were used to estimate US public interest in the vaccine before and after these key dates (Figure 2, red). Between April 2006 and May 2006, Google searches for the term “zostavax” increased 16-fold. Google searches for “zostavax” reached a maximum in November 2006, shortly after the CDC’s recommendation was broadcast. Google searches for “zostavax” decreased over time after the FDA and CDC recommendations and increased again around the date of the FDA extension in March 2011, when searches increased 1.5-fold over the subsequent 6 months and 2-fold in the following year.

The trend in searches for “zostavax” in the United States was compared to the trend in Canada, where the National Advisory Committee on Immunization (NACI) first recommended the live attenuated vaccine for patients 60 years of age or older in January 2010, and then extended this recommendation to patients 50 years of age or older in January 2014 [44,45]. In Canada (Figure 2, green), searches for “zostavax” increased 2-fold just before the initial offering of the vaccine in 2010, but their maximum did not occur until 2014, when the NACI extended its recommendation to people between 50-59 years of age. US searches for “zostavax” were also compared to searches in the United Kingdom, where National Health Service England first offered the live attenuated vaccine through its Shingles Immunisation Programme in September 2013 [46]. Google searches for “zostavax” in the United Kingdom (Figure 2, blue) reached a maximum in October 2013.

The FDA approved the new recombinant subunit vaccine (Shingrix) on October 20, 2017 [5]. In light of this recent event, we additionally analyzed Google searches before and after this approval. The results can be found in Figure 3. After FDA approval, Google searches for the term “shingrix” increased over 6-fold with an accompanying 2-fold increase in searches for the term “shingles vaccine” and 1.5-fold increase in searches for “zoster vaccine.”

While this study primarily focused on trends in HZ vaccination, we also used the same techniques to explore trends for other HZ-related terms that are less directly associated with vaccination, including “shingles contagious ” and “shingles medicine” (Figure 4); “shingles symptoms,” “shingles blisters,” “shingles rash,” and “eye shingles” (Figure 5); and “skin rash,” “shingles itch,” “shingles pain,” and “shingles treatment” (data not shown). Figure 4 also illustrates trends in searches for the possible covariate “roof shingles” that also exhibits an annual periodicity but has a slightly shifted peak season compared to the terms “shingles contagious” and “shingles medicine” (P<.001). Although we did not find sharp increases in searches for these other HZ-related terms, we did find a periodicity for many of them. The terms “shingles blisters,” “shingles itch,” “shingles medicine,” “skin rash,” and “shingles rash” showed a statistically significant annual trend (see Table 1). The terms “shingles contagious,” “eye shingles,” and “shingles pain” showed no evidence of periodicity. The term “shingles symptoms” showed a highly significant periodicity with a 20-week cycle (P<.001). The term “shingles treatment” showed a similar 20-week periodicity that did not reach significance (P=.11) according to Fisher’s g test (although Morlet wavelet decomposition revealed a highly significant periodicity which began in early 2012).

Our current analysis of historic changes and seasonal patterns of public search interest over time related to the HZ vaccine identified a highly significant annual trend for searches related to vaccination, which appeared in part to be related to utilization of Medicare benefits, covaccination with the influenza vaccine, and advertising campaigns. This finding provides us with a better understanding of public interest in the HZ vaccine, which can be applied to informing future efforts to increase vaccine coverage.

Our results also suggest that public interest in a vaccine may be significantly affected by recommendations from health-related governmental departments. In the United States, United Kingdom, and Canada, a spike in search interest occurred in each country for vaccine-related terms soon after a recommendation by a health-related governmental organization was broadcast. With this in mind, it may be helpful to increase vaccine production or clinical stocks in anticipation of the increase in vaccine interest caused by these announcements. This factor is particularly important given the recent release of the new recombinant subunit vaccine. When the live attenuated vaccine was first released, there was some delay in vaccination efforts due to an insufficient supply of the vaccine [13]. The recent release of the recombinant subunit vaccine should not fall into the same trap, and vaccine availability could be increased accordingly to anticipate the likely increase in interest after any FDA or CDC announcements.

Finally, we report on a possible seasonality for interest in HZ as measured by an increase in Google search intensity in the summer. While trends in Google searches could be representative of trends in HZ incidence, there have been mixed reports about whether or not HZ is seasonal. Many studies have been published on the topic, ranging from reports of a single practice to regional health registries [47-58]. Six of these studies found that HZ exhibits a statistically significant seasonality, while two found that HZ incidence varies by season, but this trend was either not significant or its significance was not tested. All of the studies that reported a seasonal trend found an increase of HZ cases in the summer, which could explain the increase in interest for HZ vaccination in the summer. However, Google searches are only representative of interest in a topic and do not necessarily mean that a person has the disease they have searched for. Since the seasonality of HZ has been inconsistently characterized, we believe that this trend should be evaluated further utilizing a large health-specific dataset.

The use of social media and Internet search data to assess health matters has become more useful in recent years, but its methods come with limitations including the following for our current study. Google API data are reported as a normalized relative volume and can over-represent fluctuations in searches for terms with low post volumes. Second, Google Searches also estimate interest in a given topic, but not the follow through of getting vaccinated. Likewise, the periodicity of searches for HZ vaccination may not correspond to trends in actual utilization of the vaccine. As mentioned previously, our findings could benefit from validation using large administrative clinical datasets.

Google searches about HZ vaccination exhibit an annual periodicity in the fall and significant peaks after recommendations from large public health organizations are broadcast. Given previous issues with HZ vaccine availability and coverage, these trends may help predict when to increase supply of the new recombinant subunit vaccine to meet changes in demand, and when to increase awareness to improve coverage.