Online interest in calcitonin gene‐related peptide‐therapies: An infodemiology study using Google trends Authors: Bradley Ong, Jad El Ahdab, Nicolas Thompson, Maithili Bhagat, John Ney, Allison L. Weathers Published: 2026 ABSTRACT AbstractObjectiveTo evaluate changes in public interest in calcitonin gene‐related peptide therapies associated with major product lifecycle milestones using online search data from the United States and Europe.BackgroundCalcitonin gene‐related peptide therapies have transformed migraine treatment, but the impact of key product lifecycle milestones on public awareness remains unclear. As patients increasingly turn to online platforms for health information, events such as trial publications, regulatory approvals, commercial launches, and advertising campaigns may drive search activity and signal broader awareness.MethodsWe conducted a retrospective ecological infodemiology study using a quasi‐experimental interrupted time‐series design with monthly Google Trends data (January 2017–July 2025). Search volume index (0–100 scale) was assessed for seven CGRP‐therapies. Milestone events (trial publications, regulatory approvals, commercial launches, and media exposure) were defined a priori as potential drivers of digital impact. We evaluated immediate and sustained changes in search interest following major milestones over time. The primary analysis included all therapies in the United States; secondary analyses evaluated erenumab across five European countries (Germany, France, Italy, Spain, and United Kingdom).ResultsIn the United States, commercial launch was associated with significant immediate increases in search interest for erenumab (β = 102.54; 95% confidence interval [CI], 17.05–188.02, p = 0.021) and galcanezumab (brand search term) (β = 42.36; 95% CI, 15.80–68.91, p = 0.002). Sustained post‐launch growth was observed for fremanezumab (brand search term) (β = +0.50/month; 95% CI, 0.27–0.73, p < 0.001), eptinezumab (β = +0.45/month; 95% CI, 0.26–0.64, p < 0.001), rimegepant (brand search term) (β = +0.97/month; 95% CI, 0.47–1.47, p < 0.001), atogepant (brand search term) (β = +1.28/month; 95% CI, 0.70–1.85, p < 0.001), and ubrogepant (brand search term) (β = +0.59/month; 95% CI, 0.42–0.77, p < 0.001). Media exposure was associated with significant immediate increases for erenumab (β = 94.33; 95% CI, 11.17–177.50, p = 0.029) and ubrogepant (β = 29.77; 95% CI, 11.12–48.41, p = 0.003). Regulatory approvals and trial publications were not consistently associated with significant changes in search interest. In Europe, commercial launch was associated with significant immediate increases for erenumab (generic search term) in Italy (β = 126.82; 95% CI, 62.12–191.52, p < 0.001), erenumab (brand search term) in Italy (β = 142.98; 95% CI, 23.88–262.08, p = 0.021) and the United Kingdom (β = 69.44; 95% CI, 6.71–132.16, p = 0.033). Post‐launch trends were generally negative across countries, indicating gradual declines after initial spikes.ConclusionsCommercial launch was the most consistent driver of public search interest in CGRP‐targeting therapies. Media exposure produced additional increases in selected cases, whereas regulatory approvals and clinical trial publications had limited measurable impact. These findings underscore the dominant role of commercial and consumer‐facing communication in shaping public awareness of new migraine treatments. Abbreviations augmented Dickey‐Fuller Akaike information criterion autoregressive calcitonin gene‐related peptide confidence interval European Medicines Agency United States Food and Drug Administration generalized least squares interrupted time series locally estimated scatterplot smoothing search volume index INTRODUCTION In the recent years, preventive options for migraine have expanded quickly. 1 The introduction of calcitonin gene‐related peptide (CGRP)‐therapies has reshaped the landscape of migraine prevention, generating substantial clinical interest and public attention. 2 In parallel with these therapeutic advances, there has been a growing shift in how patients access and engage with health information. 3 Patients with migraine frequently use digital technology and online search platforms, like Google, to seek information about new therapeutic options. 4 Analyzing these search behaviors offers a real‐time lens into public interest and health‐seeking behavior. Prior studies have shown that online search trends can reflect public interest in newly approved medical therapies; however, using search trends to study awareness of CGRP‐therapies remains underexplored. 5 , 6 This is especially relevant when considering how major digital milestones, such as regulatory approvals by United States Food and Drug Administration (FDA) or European Medicines Agency (EMA), high‐profile clinical trial publications, commercial launches, and/or media campaigns, may influence public awareness and perceptions of these therapies. In this study, we use the term milestones to describe key regulatory, commercial, and media events in the life cycle of CGRP‐therapies. These events are not inherently digital; rather, they become digitally visible through dissemination on regulatory websites, journal portals, press releases, advertising campaigns, and subsequent coverage on online platforms. We focused on four milestone categories: (1) regulatory approvals, (2) commercial launches, (3) trial publications, and (4) media exposure, because they represent consistent, well‐documented points where public awareness could plausibly be influenced. 7 , 8 , 9 Tracking online interest in CGRP‐therapies around these events can offer important insights into how awareness spreads, where information gaps exist, and whether media coverage is associated with increased online engagement with migraine‐related content. Although search data cannot reveal whether individuals ultimately discuss therapies with clinicians or initiate treatment, prior studies suggest that heightened online awareness may influence subsequent health‐seeking behaviors and patient‐provider conversations. 10 , 11 The objective of this study was to evaluate changes in public interest in CGRP‐therapies in response to major milestones using Google Trends data from the United States and Europe. By comparing online search trends across therapies, regions, and milestone types, this analysis aimed to elucidate how specific digital events influence public awareness and identify opportunities to enhance outreach and patient education in migraine care. We hypothesized that major milestones, including regulatory approvals, commercial launches, trial publications, and media exposure, would be associated with measurable shifts in online search interest that differ in magnitude across milestone type and region. METHODS This was a retrospective ecological infodemiology study using a quasi‐experimental interrupted time‐series framework to quantify changes in public interest in CGRP‐therapies in response to major milestones. We focused on two geographic regions (i.e., the United States and five individual European countries, including Germany, France, Italy, Spain, and the United Kingdom) where CGRP‐therapies were approved and commercially available during the study period and where Google is a widely used search engine. 12 Data source We used Google Trends (https://trends.google.com) as the primary data source to evaluate public interest in CGRP‐therapies. Google Trends is a freely accessible platform providing anonymized, aggregated data reflecting the relative popularity of specific search terms over time and geography. Google Trends data are expressed as a search volume index (SVI), ranging from 0 to 100, where 100 represents peak relative popularity for the search term within the selected time frame and geographic region. Each search term is scaled independently, meaning SVI reflects relative changes over time rather than absolute search volume and does not allow direct comparison of raw search magnitude across different terms or time windows. This approach does not provide absolute search numbers but offers relative trends valuable for infodemiological research. Outcomes and inclusion criteria The primary outcome was the monthly search volume index SVI for each CGRP‐therapy, representing relative public search interest. We included therapies approved by the FDA/EMA for migraine prevention during the study period and countries where these therapies were commercially available and Google is a widely used search engine (United States, Germany, France, Italy, Spain, and the United Kingdom). Data extraction and study period Search data were extracted monthly between January 1, 2017 through July 5, 2025. Search queries were entered manually into the Google Trends web interface using exact‐match formatting (quotation marks), and monthly SVI data were downloaded directly from the platform in comma‐separated values format for each search term and geographic region. The downloaded files were then imported into Microsoft Excel (Microsoft Corp., Redmond, WA, USA; version 16.0) and R version 4.4.3 (R Foundation for Statistical Computing, Vienna, Austria) for preprocessing and statistical analysis. The start date was selected to coincide with the period immediately preceding the first pivotal phase 3 CGRP monoclonal antibody trial publications, which marked the initial public dissemination of efficacy data. Values before these publications were excluded to maintain comparability across therapies and to focus analyses on digital milestones occurring after initial trial dissemination. The end date reflected the most current data available at the time of extraction. This study relied solely on publicly available, anonymized Google Trends data and did not involve human participants. The Cleveland Clinic institutional review board determined that the project did not constitute human subjects research; therefore, no institutional review board approval was required. Reporting followed STROBE guidelines, with adaptations for infodemiological research, consistent with prior digital health studies. 13 SVI were downloaded in comma‐separated values format and organized by month. Study design and overview We included seven CGRP‐therapies approved by the FDA or EMA for the acute and preventive treatment of migraine during the study period. Zavegepant was not included because its approval occurred late in the study period, resulting in insufficient post‐approval longitudinal data to permit robust analysis. These therapies comprised both monoclonal antibodies and small‐molecule antagonists. The monoclonal antibodies included erenumab, fremanezumab, galcanezumab, and eptinezumab. The gepants included rimegepant, atogepant, and ubrogepant. Both generic and brand names were used as individual search terms in Google Trends to comprehensively capture search behavior. All search terms were entered using quotation marks to restrict results to exact‐match queries. Searches were conducted using the “Web search” domain and “All categories” filter. For example, “erenumab” (brand search term) and “erenumab” (generic search term) were entered separately as exact‐match queries to restrict results to the specified term. Key milestones were identified for each therapy based on regulatory approvals (FDA or EMA), commercial launch dates, landmark phase 3 trial publications, and major media exposure. Dates were compiled from regulatory databases, pharmaceutical press releases, official drug websites, and peer‐reviewed literature. These milestones were grouped into four categories: (1) trial publication (reference category), (2) regulatory approval, (3) commercial launch, and (4) media exposure, to allow comparative analysis of their relative impact on search interest. For therapies with multiple regulatory approvals (e.g., rimegepant for acute and later preventive indications), approval events were modeled separately when applicable, with preventive indication approval included as a distinct regulatory milestone. Media exposure events were identified through structured review of publicly available sources, including pharmaceutical press releases, national advertising campaigns, and widely reported news or celebrity endorsements. Events were selected by consensus among the authors before modeling and were included regardless of whether coverage was positive or negative; however, we did not systematically code or stratify media exposure by sentiment (e.g., safety concerns or adverse event publicity). Analyses were structured around two prespecified outcomes. The primary outcome focused on changes in search interest for all seven CGRP‐therapies in the United States, where the most complete longitudinal data were available. The secondary outcome evaluated changes in search interest for erenumab, the first CGRP‐therapy to receive regulatory approval and commercial launch in both the United States and Europe, and the only therapy approved across all five European countries included in this study. Its longer market presence and broader availability made it the most suitable candidate for cross‐regional comparisons. Trial publication effects were analyzed using generic search terms, consistent with how clinical trial results are reported. Commercial launch and media exposure effects were analyzed primarily using brand search terms, reflecting how therapies are marketed and searched by the public. Both terms were retained when generic and brand series diverged substantially. Statistical analysis We first determined which search term (generic or brand) to analyze for each therapy, based on analyses of long‐run equilibrium of time series. We first tested each search term for stationarity using the augmented Dickey‐Fuller (ADF) test. When both ADF tests revealed that the brand and generic name time series were nonstationary, the Engle‐Granger method was used to test if the two series were cointegrated. When the two series were cointegrated, only the generic name search term was modeled. When the two series were not cointegrated, both search terms were modeled. When the ADF test revealed both search terms were stationary, we used vector autoregression and a Granger causality test. When Granger‐causality was present, we modeled only the generic name search term. When Granger causality was not present, we modeled both search terms. When the ADF test revealed one search term was stationary and the other was not, we modeled both search terms. Retained series were then entered into interrupted time‐series models. Google Trends may return zero values for months with low search volume. No months were missing within the selected study period. Zero values were retained as observed and were not imputed or transformed before modeling. Baseline search trend was defined as the period beginning with the first published phase 3 trial for each therapy and extending to its initial regulatory approval (FDA or EMA). Related search topics were reviewed to ensure the terms reflected therapeutic interest in migraine prevention and unrelated contexts. To evaluate how digital milestones influenced public search interest, we conducted interrupted time‐series (ITS) analysis to examine changes in the level and trend of a measured outcome following a specific event. We constructed segmented regression models to quantify both immediate shifts (level changes) and sustained alterations (trend changes) in SVI values surrounding each milestone. Segments were defined by consecutive milestone events, beginning with trial publication, followed by regulatory approval, commercial launch, and media exposure, with each segment length varying by drug, depending on event timing. Because trial publication was the first milestone and time was centered at that event, the intercept (β0) represented the immediate level change at trial publication, whereas the Time × Trial Publication term (β1) represented the post‐trial change in trend. Specifically, the model was defined as follows: SVIt = β0 + β1(Time × Trial Publication) + β2(FDA/EMA Approval) + β3(Commercial Launch) + β4(Time × Commercial Launch) + β5(Media Exposure) + εt, where SVIt is the monthly Google Search Volume Index at time t; Time represents monthly increments starting from the study's initial observation; and εt represents the residual error term. For rimegepant only, an additional binary indicator term was included for date of preventive approval. The primary outcome was change in SVI for all seven CGRP‐therapies in the United States, analyzed using ITS models. The secondary outcome was change in SVI for erenumab across the United States and Europe, representing the first CGRP‐therapy to reach approval and commercial launch. Model residuals were inspected for autocorrelation using Durbin‐Watson tests and partial autocorrelation function plots. When significant autocorrelation was present, models were refitted using generalized least squares (GLS) with appropriate autoregressive (AR) correlation structures. Models fitted with and without AR correlation structure were further compared using Akaike information criterion (AIC). AIC is a measure of model fit that balances the likelihood function with the number of estimated parameters. The model with AR correlation structure was deemed superior when its AIC was at least two units smaller than that of the model without AR correlation. Seasonal patterns were also visually evaluated and tested with the Ollech‐Webel combined test for seasonality and, when present, accounted for by including harmonic seasonal terms in the final models. Post‐event changes in SVI were summarized and compared across therapies, geographic regions (United States vs. Europe), and milestone types. We examined both the immediate level change and the sustained trend following each digital milestone. Comparisons were based on visual inspection of time‐series plots, descriptive summaries of percentage change from baseline, and the duration of elevated interest after each event. We generated time‐series line graphs to illustrate both raw and modeled SVI over time for each therapy, with annotated milestone timelines indicating the occurrence of regulatory approvals, trial publications, commercial launches, and media events. All data processing and cleaning were performed in Microsoft Excel (Microsoft Corp.; version 16.0). Statistical modeling was conducted using R version 4.4.3 (R Foundation for Statistical Computing, Vienna, Austria). Interrupted time‐series models were estimated using the nlme package (version 3.1‐165), autocorrelation diagnostics were performed using the forecast package (version 9.0.1), the aTSA package (version 3.1.2.1), and the seastests package (version 0.15.4). Data management and visualization were completed using tidyverse (version 2.0.0). A two‐sided α level of 0.05 was considered statistically significant for all analyses. No a priori statistical power calculation was conducted. The sample size was determined by the availability of historical Google Trends data for the selected therapies from January 2017 through July 2025, which defined the temporal scope of the interrupted time‐series analyses. RESULTS Long‐run equilibrium between generic and brand‐name search terms We evaluated stationarity and long‐run equilibrium between brand‐name and generic search terms for each CGRP‐therapy to determine whether analyzing both terms was necessary. Table S1 shows results of tests for stationarity (ADF test), cointegration (Engle‐Granger cointegration test), and Granger causality tests. In the United States, only two pairs of brand‐name/generic search terms had both time series nonstationary by ADF test (eptinezumab [brand and generic search terms] and rimegepant [brand and generic search terms]). Engle‐Granger cointegration tests revealed that the brand/generic search term time series were not cointegrated. Thus, all four of these search terms were retained for analysis. Three pairs of brand‐name/generic search terms had time series where the ADF test revealed the brand‐name search term time series was stationary and the generic search term time series was nonstationary (fremanezumab [brand and generic search terms], galcanezumab [brand and generic search terms], and ubrogepant [brand and generic search terms]), so we modeled all of these search terms. The ADF test for erenumab (brand and generic search terms) revealed both time series were stationary. Vector autoregression and the Granger causality test revealed that the time series for erenumab Granger‐caused the time series for erenumab (brand search term). Thus, we excluded erenumab (brand search term) from modeling in the United States. Finally, we note that the search term atogepant had SVI 0 for all time points. Thus, we only model the search term atogepant (brand search term) for this pair of search terms. Across European countries, analyses were restricted to erenumab (brand and generic search terms). ADF tests revealed each of erenumab (brand and generic search terms) time series were nonstationary for Germany and the United Kingdom. Engle‐Granger cointegration tests revealed the time series for each term were not cointegrated for both countries, so both search terms were modeled for these countries. For France and Spain, the ADF test revealed both brand‐name and generic search terms had stationary time series. Vector autoregression and Granger‐causality tests revealed that the time series for erenumab Granger‐caused the series for erenumab (brand search term), so erenumab (brand search term) was not modeled for these countries. For Italy, the ADF test revealed that the time series for erenumab (brand search term) was nonstationary and the time series for erenumab was stationary, so both search terms were modeled. Interrupted time‐series analysis of digital milestones All final analyses used GLS models with AR (1) correlation structure, which provided superior fit after diagnostic testing. Ollech‐Webel combined tests for seasonality revealed the need to include harmonic seasonal terms for the erenumab, galcanezumab (brand search term), and galcanezumab models. Full diagnostic results are provided in Table S2 and Figures S1 and S2. We performed segmented regression analyses to quantify the immediate (level) and sustained (trend) changes in public search interest associated with each milestones type. Detailed model results are shown in Table 1 (all drugs, United States) and Table 2 (erenumab [brand and generic search terms], Europe). TABLE 1Results of interrupted time series segmented regression models for the United States of America.Search termModel termMilestone dateβ estimate (95% CI) p valueErenumab (generic search term) a Trial Publication Immediate Level Change2017–119.52 (−17.73 to 36.77)0.495Trial Publication Change in Trend4.59 (−0.79 to 9.97)0.098FDA Approval Immediate Level Change2018–0525.31 (−2.71 to 53.34)0.080Commercial Launch Immediate Level Change2018–0648.46 (18.92–78.00)0.002Commercial Launch Change in Trend−0.33 (−0.71 to 0.06)0.099Media Exposure Immediate Level Change2019–0143.85 (14.52–73.18)0.004Fremanezumab (brand search term)Trial Publication Immediate Level Change2017–11−0.92 (−18.49 to 16.65)0.918Trial Publication Change in Trend−0.25 (−2.67 to 2.18)0.844FDA Approval Immediate Level Change2018–0923.77 (2.40–45.15)0.032Commercial Launch Immediate Level Change2018–1041.98 (20.52–63.45)<0.001Commercial Launch Change in Trend0.50 (0.27–0.73)<0.001Fremanezumab (generic search term)Trial Publication Immediate Level Change2017–1157.04 (41.63–72.45)<0.001Trial Publication Change in Trend−2.05 (−4.88 to 0.77)0.158FDA Approval Immediate Level Change2018–0936.66 (10.02–63.29)0.008Commercial Launch Immediate Level Change2018–10−19.95 (−37.09 to −2.80)0.025Commercial Launch Change in Trend0.05 (−0.08–0.19)0.415Galcanezumab (brand search term) a Trial Publication Immediate Level Change2018–0822.75 (−2.26 to 47.77)0.079FDA Approval Immediate Level Change2018–0916.70 (−1.46 to 34.85)0.075Commercial Launch Immediate Level Change2018–1044.25 (20.42–68.09)<0.001Commercial Launch Change in Trend0.09 (−0.26 to 0.44)0.597Galcanezumab (generic search term) a Trial Publication Immediate Level Change2018–088.18 (−86.13 to 102.48)0.866FDA Approval Immediate Level Change2018–093.84 (−12.81 to 20.49)0.653Commercial Launch Immediate Level Change2018–104.33 (−19.78 to 28.43)0.726Commercial Launch Change in Trend0.87 (−0.53 to 2.26)0.228Eptinezumab (brand search term)Trial Publication Immediate Level Change2020–010.66 (−518.38 to 519.69)0.998FDA Approval Immediate Level Change2020–026.00 (−8.52 to 20.52)0.421Commercial Launch Immediate Level Change2020–043.71 (−17.54 to 24.97)0.733Commercial Launch Change in Trend1.43 (−0.39 to 3.25)0.128Eptinezumab (generic search term)Trial Publication Immediate Level Change2020–0133.18 (10.15–56.21)0.006FDA Approval Immediate Level Change2020–0221.68 (−5.17 to 48.52)0.118Commercial Launch Immediate Level Change2020–04−4.19 (−28.35 to 19.96)0.735Commercial Launch Change in Trend0.45 (0.26–0.64)<0.001Rimegepant (brand search term)Trial Publication Immediate Level Change2019–087.70 (−20.14 to 35.55)0.590Trial Publication Change in Trend2.47 (−3.58 to 8.52)0.426FDA Approval Immediate Level Change2020–0217.60 (−13.42 to 48.62)0.270Commercial Launch Immediate Level Change2020–0323.91 (−8.49 to 56.32)0.153Commercial Launch Change in Trend0.97 (0.47–1.47)<0.001Preventive Approval Immediate Level Change* 2021–0542.71 (10.66–74.77)0.011Media Exposure Immediate Level Change2021–0658.68 (26.70–90.67)<0.001Rimegepant (generic search term)Trial Publication Immediate Level Change2019–0814.86 (−34.97 to 64.69)0.561Trial Publication Change in Trend2.24 (−7.31 to 11.79)0.648FDA Approval Immediate Level Change2020–0217.54 (−32.74 to 67.81)0.497Commercial Launch Immediate Level Change2020–0310.35 (−44.57 to 65.27)0.713Commercial Launch Change in Trend0.98 (−0.04 to 2.00)0.065Preventive Approval Immediate Level Change* 2021–0533.94 (−18.87 to 86.75)0.212Media Exposure Immediate Level Change2021–0638.55 (−14.19 to 91.30)0.157Atogepant (brand search term)Trial Publication Immediate Level Change2021–0812.19 (−11.36 to 35.74)0.316FDA Approval Immediate Level Change2021–099.47 (−10.09 to 29.03)0.348Commercial Launch Immediate Level Change2021–1030.84 (6.69–54.98)0.016Commercial Launch Change in Trend1.28 (0.70–1.85)<0.001Ubrogepant (brand search term)Trial Publication Immediate Level Change23.07 (8.74–37.40)0.002Commercial Launch Immediate Level Change2020–0121.64 (7.33–35.95)0.004Commercial Launch Change in Trend0.59 (0.42–0.77)<0.001Media Exposure Immediate Level Change2020–0829.77 (11.12–48.41)0.003Ubrogepant (generic search term) b Trial Publication Immediate Level Change13.99 (−4148.16 to 4176.14)0.995Commercial Launch Immediate Level Change2020–01−5.36 (−21.01 to 10.28)0.504Commercial Launch Change in Trend1.36 (−0.55 to 3.27)0.167Media Exposure Immediate Level Change2020–085.05 (−18.21 to 28.30)0.672Abbreviations: CI, confidence interval; FDA, Food and Drug Administration. a Harmonic seasonal terms included. b Model intercept term had a very large standard error and corresponding wide 95% CI, indicating model instability. We also reported the results of the simpler linear regression model in the main text.*Preventive indication approval refers to the separate FDA approval of rimegepant for migraine prevention, modeled as a distinct regulatory milestone. Results of interrupted time series segmented regression models for the United States of America. Abbreviations: CI, confidence interval; FDA, Food and Drug Administration. Harmonic seasonal terms included. Model intercept term had a very large standard error and corresponding wide 95% CI, indicating model instability. We also reported the results of the simpler linear regression model in the main text. Preventive indication approval refers to the separate FDA approval of rimegepant for migraine prevention, modeled as a distinct regulatory milestone. TABLE 2Results of interrupted time series segmented regression models for erenumab for five European countries.CountrySearch termModel termMilestone dateβ estimate (95% CI) p valueFranceErenumab (generic search term)Trial Publication Immediate Level Change2017–114.54 (−14.50 to 23.58)0.641Trial Publication Change in Trend−0.59 (−5.04 to 3.86)0.795EMA Approval Immediate Level Change2018–0715.12 (−16.79 to 47.03)0.356Commercial Launch Immediate Level Change2018–0843.06 (22.38–63.75)<0.001Commercial Launch Change in Trend−0.44 (−0.58 to −0.30)<0.001GermanyErenumab (brand search term)Trial Publication Immediate Level Change2017–11−4.78 (−45.21 to 35.65)0.817Trial Publication Change in Trend0.83 (−4.59 to 6.25)0.765EMA Approval Immediate Level Change2018–078.21 (−31.31 to 47.72)0.685Commercial Launch Immediate Level Change2018–0818.34 (−24.80 to 61.47)0.407Commercial Launch Change in Trend0.69 (0.03–1.35)0.043Erenumab (generic search term)Trial Publication Immediate Level Change2017–11−6.52 (−52.60 to 39.57)0.782Trial Publication Change in Trend1.95 (−4.13 to 8.03)0.531EMA Approval Immediate Level Change2018–0711.90 (−32.49 to 56.30)0.601Commercial Launch Immediate Level Change2018–0830.29 (−18.30 to 78.89)0.225Commercial Launch Change in Trend0.64 (−0.11 to 1.40)0.100ItalyErenumab (brand search term)Trial Publication Immediate Level Change2017–11−5.54 (−39.79 to 28.71)0.752Trial Publication Change in Trend4.86 (−3.14 to 12.87)0.237EMA Approval Immediate Level Change2018–0763.78 (6.34–121.21)0.032Commercial Launch Immediate Level Change2018–0881.81 (44.61–119.01)<0.001Commercial Launch Change in Trend−0.70 (−0.95 to −0.44)<0.001Erenumab (generic search term)Trial Publication Immediate Level Change2017–11−2.22 (−20.65 to 16.21)0.814Trial Publication Change in Trend3.62 (−0.74 to 7.98)0.107EMA Approval Immediate Level Change2018–0742.88 (10.45–75.31)0.011Commercial Launch Immediate Level Change2018–0880.15 (60.26–100.03)<0.001Commercial Launch Change in Trend−0.62 (−0.75 to −0.49)<0.001SpainErenumab (generic search term)Trial Publication Immediate Level Change2017–11−0.39 (−16.31 to 15.54)0.962Trial Publication Change in Trend0.47 (−3.25 to 4.18)0.807EMA Approval Immediate Level Change2018–0712.57 (−13.97 to 39.11)0.356Commercial Launch Immediate Level Change2018–0836.64 (19.33–53.95)<0.001Commercial Launch Change in Trend−0.15 (−0.27 to −0.03)0.017UKErenumab (brand search term)Trial Publication Immediate Level Change2017–11−4.23 (−22.82 to 14.35)0.657Trial Publication Change in Trend3.17 (−0.99 to 7.33)0.139EMA Approval Immediate Level Change2018–0725.26 (−2.85 to 53.36)0.082Commercial Launch Immediate Level Change2018–0833.35 (12.72–53.98)0.002Commercial Launch Change in Trend−0.11 (−0.27 to 0.04)0.161Erenumab (generic search term)Trial Publication Immediate Level Change2017–1113.77 (−6.66 to 34.20)0.190Trial Publication Change in Trend0.46 (−4.05 to 4.96)0.843EMA Approval Immediate Level Change2018–0726.53 (−3.67 to 56.73)0.089Commercial Launch Immediate Level Change2018–0825.00 (2.18–47.82)0.035Commercial Launch Change in Trend−0.19 (−0.36 to −0.01)0.038Abbreviations: CI, confidence interval; EMA, European Medicines Agency; UK, United Kingdom. Results of interrupted time series segmented regression models for erenumab for five European countries. Abbreviations: CI, confidence interval; EMA, European Medicines Agency; UK, United Kingdom. Primary outcome Commercial launches were the most consistent drivers of search interest across all seven therapies. Significant immediate increases were observed for erenumab (generic search term) (β = 102.54; 95% confidence interval [CI], 17.05–188.02, p = 0.021) and galcanezumab (brand search term) (β = 42.36; 95% CI, 15.80–68.91, p = 0.002). Sustained post‐launch growth was also evident for fremanezumab (brand search term) (β = +0.50 per month; 95% CI, 0.27–0.73, p < 0.001), eptinezumab (generic search term) (β = +0.45 per month; 95% CI, 0.26–0.64, p < 0.001), rimegepant (brand search term) (β = +0.97 per month; 95% CI, 0.47–1.47, p < 0.001), atogepant (brand search term) (β = +1.28 per month; 95% CI, 0.70–1.85, p < 0.001), and ubrogepant (brand search term) (β = +0.59 per month; 95% CI, 0.42–0.77, p < 0.001). Media exposure produced strong spikes in search interest, particularly for erenumab (generic search term) (β = 94.33; 95% CI, 11.17–177.50, p = 0.029) and ubrogepant (brand search term) (β = 29.77; 95% CI, 11.12–48.41, p = 0.003). Regulatory approvals showed more variable effects with no significant differences. Trial publications, analyzed using generic terms, did not significantly alter search patterns. We note that the model for ubrogepant had a very wide 95% CI for the intercept term, which indicates model instability. We therefore chose to also report the simpler model fit via linear regression. The direction of post‐launch trend remained positive in the simpler model; however, estimates for immediate level change and media exposure differed from the primary GLS results. Given superior model fit based on AIC and autocorrelation diagnostics, the GLS model with AR (1) structure was retained as the primary analysis. Secondary outcome There were significant immediate increases in search volume following commercial launch for erenumab (brand search term) in Italy (β = 142.98; 95% CI, 23.88–262.08, p = 0.021), erenumab (generic search term) in Italy (β = 126.82; 95% CI, 62.12–191.52, p < 0.001), and erenumab (brand search term) in the United Kingdom (β = 69.44; 95% CI, 6.71–132.16, p = 0.033). Sustained post‐launch slopes were slightly negative for erenumab (brand search term) in Germany and Italy and also negative for erenumab (generic search term) in France, Spain, and the United Kingdom. The post‐launch slope for erenumab (brand search term) in Germany had a significant positive slope (β = +0.69 per month; 95% CI, 0.03–1.35, p = 0.043). Regulatory approvals had one significant increase, which was for erenumab in Italy (β = 82.73; 95% CI, 12.10–153.36, p = 0.024). Trial publications showed no significant trends. Data visualization of search trends Scatterplots and locally estimated scatterplot smoothing (LOESS)‐smoothed curves illustrating changes in public interest over time are provided in Figure 1 (United States) and Figure 2 (Europe, erenumab [brand and generic search terms] only). Each figure is annotated with milestone events, including regulatory approval, commercial launch, trial publication, and media exposure (United States only). Public search interest in CGRP‐therapies in the United States (2017–2025). Scatterplots with LOESS‐smoothed curves show monthly Google search trends for selected monoclonal antibodies and gepants. CGRP, calcitonin gene‐related peptide; EMA, European Medicines Agency; FDA, United States Food and Drug Administration; LOESS, locally estimated scatterplot smoothing; SVI, search volume index. [Color figure can be viewed at wileyonlinelibrary.com] Public search interest in erenumab across the United States and five European countries (2017–2025). Scatterplots with LOESS‐smoothed curves show monthly Google search trends for erenumab by country and search term. EMA, European Medicines Agency; FDA, United States Food and Drug Administration; LOESS, locally estimated scatterplot smoothing; SVI, search volume index. [Color figure can be viewed at wileyonlinelibrary.com] In the United States, multiple therapies demonstrated visible increases in search interest following commercial launches, with the magnitude and statistical significance of effects varying across drugs. For erenumab (generic search term), repeated media exposure sustained elevated search volumes well beyond the initial peak. Fremanezumab (brand search term), galcanezumab (brand search term), and atogepant (brand search term) also showed pronounced launch effects, whereas rimegepant (brand search term) and ubrogepant (brand search term) demonstrated additional boosts linked to media campaigns. In Europe, analyses were restricted to erenumab. LOESS curves revealed consistent spikes at commercial launch across all five countries, with gradual declines in subsequent months. These visual patterns closely mirrored the interrupted time‐series results shown in Table 2. Across both regions, trial publications rarely produced noticeable shifts in search interest, whereas regulatory approvals yielded modest or inconsistent effects. In contrast, commercial launches and media campaigns generated the clearest and most sustained impacts, consistent with the regression models. DISCUSSION Between January 2017 and July 2025, public interest in CGRP‐therapies demonstrated clear responsiveness to key milestones. In the United States, several therapies demonstrated significant immediate or sustained increases following commercial launches, with additional boosts from regulatory approvals and media exposure in selected cases. Baseline search interest appeared to vary by therapy class, with monoclonal antibodies generally showing earlier public interest than gepants. Erenumab showed parallel patterns in the United States and across five European countries. Commercial launch produced the most pronounced and durable increases in search interest across all regions, whereas regulatory approvals had more variable and modest effects. Trial publications typically showed little measurable impact. Media campaigns were associated with significant increases in search interest for selected therapies in the United States, highlighting the influence of consumer‐directed exposure. This study systematically quantified how major milestones, such as regulatory approvals, commercial launches, trial publications, and media exposures, affected online public interest in CGRP‐therapies. Analyses were conducted for all seven therapies in the United States and, as a secondary outcome, for erenumab across five major European countries. Our findings highlight specific milestones that may influence public engagement and patient‐level awareness of novel migraine treatments, offering practical insights for providers, health care policymakers, and pharmaceutical stakeholders. We identified significant launch‐related effects across multiple therapies in the United States, with some demonstrating immediate level increases and others exhibiting sustained post‐launch growth. In Europe, significant immediate launch effects were observed in selected countries, particularly Italy and the United Kingdom, whereas other countries showed more modest or nonsignificant responses. Our findings suggest that commercial launch and associated promotional campaigns strongly influence public awareness and interest. Previous Google Trends analyses of newly introduced therapies in other clinical domains, including glucagon‐like peptide‐1 receptor agonists and thyroid eye disease treatments, have likewise demonstrated marked increases in online search interest following commercial launch and direct‐to‐consumer advertising, with comparatively smaller effects associated with regulatory approval or publication of trial data. 14 , 15 , 16 In contrast, regulatory approval events had a varied and generally modest impact on search trends. In the United States, regulatory approval events were not associated with statistically significant changes in search interest across therapies. In Europe, regulatory approval effects were generally modest and not statistically significant; however, Italy demonstrated a significant immediate increase in search interest following EMA approval of erenumab. These findings align with prior studies indicating that regulatory processes may remain relatively opaque or less directly relevant to patients and the general public. 5 , 17 Interestingly, landmark trial publications, despite their centrality in medical literature and potential impact on clinical guidelines, rarely prompted significant shifts in public online interest. This underscores an apparent disconnect between scientific communication within the medical community and broader patient awareness or engagement. 18 Bridging this gap may require targeted dissemination strategies tailored for patient populations but also stronger engagement of the media by journals and industry partners when reporting landmark studies. Future work could examine whether trials that were actively promoted through press releases or picked up by lay media generated greater public awareness compared to those that were not. 19 , 20 Notably, media exposure markedly increased search interest, but these effects were observed only in the United States. Pronounced spikes for erenumab, rimegepant (brand search term), and ubrogepant (brand search term) were linked to direct‐to‐consumer advertising, celebrity endorsements, or national campaigns, whereas no comparable media‐driven effects were observed in Europe. Previous studies have provided similar findings, underscoring the powerful role of strategic media campaigns in influencing patient awareness and health care decision‐making. 5 , 21 The relatively subdued impact of media exposure in European countries reflects more stringent regulations governing pharmaceutical marketing directly to consumers, 22 , 23 reinforcing the need for region‐specific patient engagement strategies. 24 Direct‐to‐consumer pharmaceutical advertising is prohibited across the European Union under Directive 2001/83/EC, and no equivalent consumer‐facing media campaigns occurred for CGRP‐therapies during the study period. 25 As a result, our models contained no media exposure events for Europe, underscoring a structural difference rather than a measurement limitation. The comparative analysis of monoclonal antibodies versus gepants in the United States revealed clear differences between drug classes in public interest profiles. Monoclonal antibodies demonstrated consistently greater baseline search interest and more pronounced, sustained increases following milestones compared to gepants. This difference may partly result from earlier approval timelines and broader professional familiarity with monoclonal antibodies, highlighting the importance of timing, cumulative clinical experience, and/or sustained marketing in establishing robust patient‐provider awareness. 21 , 26 In addition, consumer interest may have become progressively saturated as multiple therapies entered the same clinical niche, leading to smaller relative increases in search activity for later products. Study limitations This study has important limitations. First, the reliance on Google Trends provides a relative measure of public interest rather than absolute numbers of searches or actual medication use. Thus, higher relative interest does not necessarily equate to higher real‐world uptake or utilization of therapies. Because SVI is a normalized, relative metric calculated separately for each search term, cross‐therapy comparisons should be interpreted with caution and considered descriptive rather than absolute. Each SVI value is self‐referential, scaled from 0 to 100 relative to the peak search volume for that specific term within a given region and timeframe. Consequently, analyses for different drugs and for erenumab across European countries are internally valid but not directly comparable across terms or regions. This normalization framework means our results capture within‐term changes over time rather than absolute differences in search volume between therapies or countries. Because SVI values are independently normalized for each search term, constructing a composite class‐level index by summing or averaging individual therapies would not yield a valid measure of overall search magnitude. Any such aggregation would mix independently scaled series and potentially distort interpretation. Therefore, analyses were conducted at the individual therapy level to preserve internal validity. Google Trends does not disclose the extent to which automated or algorithmic search queries contribute to overall search volume. Although the platform is designed to filter irregular or nonhuman traffic, we cannot exclude the possibility that some portion of observed search activity reflects automated processes. If present, such activity would likely represent broader digital engagement patterns rather than targeted manipulation of specific milestone events; however, this remains a potential source of residual confounding. We also cannot determine whether any automated search activity remains constant over time or fluctuates in response to external events. If algorithmic searches are temporally stable, they would be incorporated into the baseline trend and would not materially affect the estimated post‐milestone changes. Only systematic variation in automated activity coinciding with specific milestone dates would bias our results, which we consider unlikely but cannot definitively exclude. Although Google Trends has proven valuable for epidemiologic research, interpreting these findings as direct proxies for patient behavior or clinical practice requires caution. 6 , 27 Additionally, we identified and categorized milestone events primarily based on pharmaceutical press releases, regulatory documentation, and published literature, potentially overlooking smaller‐scale, but still influential, local events or online activities. Furthermore, our analysis did not differentiate between search interest driven by patients versus health care professionals. Because professionals may also contribute substantially to search volume, particularly around regulatory approvals or scientific publications, these trends should be considered a combined reflection of public and professional interest. 28 , 29 The study period (2017–2025) overlapped substantially with the corona virus 2019 pandemic, which altered digital health‐seeking behaviors and baseline internet usage patterns globally. Although ITS modeling accounts for underlying temporal trends, pandemic‐related shifts in health care access and online engagement may have influenced baseline search activity in ways not fully captured by milestone‐specific effects. Additionally, Google Trends data may disproportionately reflect the behavior of younger, more digitally literate, and higher‐income populations with consistent internet access. Older adults and individuals with limited digital access may therefore be underrepresented in search‐based analyses, potentially limiting generalizability. Analyses outside the United States were restricted to erenumab, the first approved CGRP‐therapy, which may not fully capture class‐level or drug‐specific variation in European regions. Although we accounted for autocorrelation in search behavior using appropriate statistical modeling, unmeasured external factors, such as changes in clinical guidelines, reimbursement decisions, competitor drug introductions, or unrelated news events, may also influence search trends and could not be fully controlled or analyzed in this study. Lastly, correlating Google search interest with actual medication sales or prescription data would provide a stronger link to real‐world uptake, but these figures are typically proprietary and not publicly available. As a result, our analysis was limited to relative online search patterns rather than direct measures of therapy utilization. Future directions Further research should aim to connect online public interest measures directly with real‐world outcomes such as drug prescriptions, sales data, or treatment adherence rates. Integrating digital data with clinical and administrative data sets could help clarify how online interest translates into actual health care utilization. Future studies should also investigate the impact of specific online platforms, including social media, video‐sharing platforms, and health forums, to better understand the digital ecosystem of patient information‐seeking behavior. Understanding patient motivations behind online searches could further illuminate the practical implications of these infodemiological findings for patient education and clinical care delivery. Finally, qualitative exploration of why patients and providers respond differently to various milestone types could yield insights for optimizing communication strategies. In addition to trial results and regulatory approvals, future studies might also examine events that were not captured in our analysis, such as negative press related to safety concerns, publicity surrounding adverse events, or personal milestones of celebrity endorsers. Understanding how these different events are amplified or overlooked could help close current gaps in public awareness and strengthen engagement with scientifically important developments. CONCLUSION Our findings highlight that commercial launches and targeted media exposure have significant influence on public search interest in novel migraine preventive therapies. In the United States, all seven therapies demonstrated strong responses to these milestones. In Europe, analyses were limited to erenumab, the first CGRP‐targeting therapy approved in the region, which also showed significant effects at launch. Regulatory approvals produced variable effects in the United States but had no measurable impact in Europe. Understanding these infodemiological trends can help clinicians, policymakers, and pharmaceutical companies optimize strategies for communicating therapeutic advances, ultimately promoting better patient awareness, informed decision‐making, and equitable access to emerging migraine treatments. AUTHOR CONTRIBUTIONS Bradley Ong: Conceptualization; investigation; writing – original draft; methodology; validation; visualization; writing – review and editing; resources; data curation. Jad El Ahdab: Investigation; methodology; validation; formal analysis; data curation; resources; writing – original draft. Nicolas Thompson: Software; formal analysis; project administration; supervision; methodology; writing – review and editing. Maithili Bhagat: Writing – original draft; resources; data curation. John Ney: Supervision; writing – review and editing; validation; resources; project administration; methodology; visualization. Allison L. Weathers: Resources; supervision; project administration; writing – review and editing; validation; methodology; visualization. FUNDING INFORMATION The author(s) received no financial support for the research, authorship, and/or publication of this article. Authors have nothing to disclose. CONFLICT OF INTEREST STATEMENT Bradley Ong serves as a member of the Neurology Resident & Fellow Section Editorial Board and has participated as an uncompensated host for the Neurology Minute podcast of the American Academy of Neurology. Bradley Ong has also received research support from the American Academy of Neurology and the American Headache Society/International Headache Academy. John Ney is an associate editor for Neurology Clinical Practice. He receives compensated consultation in the amount of $10,000–$49,0000 from Ceribell, Inc. He is a federal government employee; work on this research was performed outside of his duties as a federal employee. Allison L. Weathers is a member of the Neurology Clinical Practice editorial board. She receives compensation in the amount of $2100 from the American Academy of Neurology for work as a CME question writer for Continuum and for her work on American Academy of Neurology produced educational content. She has noncompensated physician advisory roles with Epic and Oracle. Jad El Ahdab, Nicolas Thompson, and Maithili Bhagat have no conflicts of interest to disclose. Supporting information Figure S1. Partial autocorrelation function (PACF) plots for U.S. therapies. The dashed horizontal lines represent the 95% confidence bounds under the null hypothesis of no partial autocorrelation at each lag. Figure S2. Partial autocorrelation function (PACF) plots for erenumab in European countries. The dashed horizontal lines represent the 95% confidence bounds under the null hypothesis of no partial autocorrelation at each lag. Table S1. Results of augmented Dickey‐Fuller (ADF) tests, Engle‐Granger cointegration tests, and Granger causality tests. Statistically significant p values from the ADF test indicate evidence of a stationary time series. Statistically significant p values from the Engle‐Granger cointegration test indicate evidence that the two time series are cointegrated. Significant p values from the Granger causality test indicate evidence that one time series Granger‐causes the other time series. Table S2. p values from Durbin‐Watson Test, Akaike Information Criterion values from linear regression models and generalized least squares models with AR(1) correlation structure, and p values from Ollech‐Webel combined test for seasonality.