Introduction

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is a novel coronavirus that emerged from Wuhan China in December 2019, causing the disease that is referred to as coronavirus disease 2019 (COVID-19). This disease spread rapidly throughout China and other regions of the world, resulting in the World Health Organization (WHO) declaring COVID-19 a pandemic [1]. Globally, a total of 28,329,790 confirmed cases of COVID-19, including 911,877 deaths, had been reported to the WHO, as of September 12, 2020. The country with the highest number of cases and deaths is the USA, with 6,341,309 confirmed cases and 190,787 deaths [2].

Patients taking anti-rheumatic drugs share concerns regarding a potential increased risk of contracting COVID-19 infection [3,4,5,6,7,8], and the public may have searched online information for the implications and risks of some anti-rheumatic drugs, which have been used in the management of COVID-19 infection and associated inflammatory sequelae or cytokine storm [9,10,11,12,13,14,15]. Therefore, population-level interest in anti-rheumatic drugs in the COVID-19 era should be investigated to increase our understanding of this new pandemic.

During the past decade, an increasing number of scientific studies have documented the importance of Google Trends in the detection of an epidemic and monitoring public interest [16,17,18,19,20]. It has been used to study previous epidemics such as influenza [21, 22], Dengue fever [23], and Zika virus [24]. In light of COVID-19, Google Trends data has been used to investigate population-level interest in several treatment approaches including urologic procedures [25], hip and knee arthroplasties [26], facial plastic surgery [27], and cosmetic procedures [28]. Furthermore, an interesting study investigating Google searches of hydroxychloroquine is available in the literature [29]. Our study would expand the knowledge on Google searches of anti-rheumatic drugs beyond hydroxychloroquine.

Therefore, we aimed to investigate the population-level interest in anti-rheumatic drugs in the COVID-19 era, by analyzing changes in Google search frequency data.

Materials and method

Google Trends presents the frequency of Google search terms in a normalized form as a relative search volume (RSV). Values of RSV range between 0 and 100, where 100 represents to the peak popularity for the search term [30]. The information on Google Trends and its data are presented in detail in the literature [17, 18].

We selected search terms that encompass anti-rheumatic drugs including non-steroidal anti-inflammatory drugs (NSAIDs), glucocorticoids, antigout agents, conventional Disease-modifying anti-rheumatic drugs (DMARDs), immunosuppressant agents, interleukin (IL)-6 inhibitors, IL-1 inhibitors, tumor necrosis factor (TNF) inhibitors, IL-17 inhibitors, IL-12/23 pathway targeting agents, T cell co-stimulation modulators, B cell-targeting agents, Janus kinase (JAK) inhibitors, and phosphodiesterase-4 (PDE4) inhibitor. Both generic and brand names of the drugs were included, and brand names were obtained from UpToDate [31]. We used a total of 78 search terms as the following: NSAIDs, Ibuprofen, Advil, Indomethacin, Indocin, Diclofenac, Voltaren, Naproxen, Aleve, Meloxicam, Mobic, Glucocorticoids, Prednisone, Deltasone, Colchicine, Colcrys + Mitigare, Allopurinol, Zyloprim, Febuxostat, Uloric, Hydroxychloroquine, Plaquenil, Methotrexate, Otrexup + Rasuvo + Rheumatrex + Trexall, Sulfasalazine, Azulfidine, Leflunomide, Arava, Mycophenolate mofetil, CellCept, Azathioprine, Imuran, Cyclophosphamide, Cytoxan, Tacrolimus, Prograf, Cyclosporine, Gengraf + Neoral + Sandimmune, Tocilizumab, Actemra, Sarilumab, Kevzara, Anakinra, Kineret, Canakinumab, Ilaris, Rilonacept, Arcalyst, Etanercept, Enbrel, Infliximab, Remicade, Adalimumab, Humira, Certolizumab, Cimzia, Golimumab, Simponi, Secukinumab, Cosentyx, Ixekizumab, Taltz, Ustekinumab, Stelara, Guselkumab, Tremfya, Abatacept, Orencia, Rituximab, Rituxan, Belimumab, Benlysta, Tofacitinib, Xeljanz, Baricitinib, Olumiant, Apremilast, and Otezla. To obtain the relative frequency of these selected search terms, we queried Google Trends with the selection of “United States,” “01/01/2017–09/09/2020,” and “Web Search” settings on September 9, 2020. We imposed no category restrictions when querying Google Trends and exported the Google Trends weekly RSV data for further analysis.

Search term RSV during the initial 8-week period (March 15–May 9, 2020), after US President Donald Trump declared a national emergency due to the COVID-19 outbreak, was compared with overlapping periods of the preceding 3 years (2017–2019) to investigate initial stage interest in the drugs. In addition, we compared May 10–July 4, 2020, and July 5–August 29, 2020, periods with overlapping periods in 2017–2019 to investigate short-term interest. Previous studies included the years of 2015 and 2016 as well [25, 26]; however, to avoid selection bias in our study, we excluded data from the years 2015 and 2016 because some anti-rheumatic drugs were less popular treatments during that time. To investigate whether RSV had changed between the periods, we performed generalized estimating equations using a model of gamma with log link. The analysis was conducted using SPSS version 21.0, IBM. The level of significance was established at the 0.05 level.

Results

During the initial period, March 15–May 9, 2020, the RSV of NSAIDs (% change: + 88.4%; p = 0.030), diclofenac (+ 7.7%; p = 0.018), colchicine (+ 24.4%; p < 0.001), febuxostat (+ 101.3%; p < 0.001), hydroxychloroquine (+ 3613.0%; p < 0.001), Plaquenil (+ 635.0%; p = 0.003), tacrolimus (+ 12.6, p = 0.030), tocilizumab (+ 1253.5%; p < 0.001), Actemra (+ 443.6%; p < 0.001), sarilumab (+ 442.5%; p < 0.001), Kevzara (+ 272.2%; p < 0.001), anakinra (+ 188.6%; p < 0.001), infliximab (+ 19.2%; p = 0.005), Stelara (+ 13.8%; p = 0.015), and baricitinib (+ 91.9%; p = 0.027) showed a statistically significant increase; conversely, the RSV of Indocin (− 39.4%; p < 0.001), Voltaren (− 26.2%; p < 0.001), meloxicam (− 6.5%; p = 0.012), Mobic (− 38.9%; p < 0.001), allopurinol (− 12.6%; p = 0.006), Zyloprim (− 27.6%; p = 0.034), Uloric (− 39.9%; p < 0.001), methotrexate (− 12.6%; p < 0.001), Arava (− 29.3%; p < 0.001), CellCept (− 15.4%; p = 0.003), Imuran (− 31.5%; p < 0.001), Cytoxan (− 24.5%; p < 0.001), Prograf (− 31.1%; p < 0.001), Gengraf + Neoral + Sandimmune (− 28.0%; p = 0.026), canakinumab (− 1.3%; p = 0.046), Ilaris (− 32.5%; p = 0.029), rilonacept (− 14.2%; p = 0.023), Remicade (− 12.0%; p = 0.031), golimumab (− 37.2%; p = 0.005), ixekizumab (− 32.7%; p = 0.008), Taltz (− 30.7%; p < 0.001), and Rituxan (− 34.4%; p < 0.001) displayed a statistically significant decrease as compared to overlapping periods of the preceding 3 years (Table 1).

Table 1 Relative search volume (RSV) of anti-rheumatic drugs
Full size table

During the May 10–July 4, 2020, period, Diclofenac (+ 14.8%; p < 0.001), Voltaren (+ 222.8%; p < 0.001), Febuxostat (+ 72.3%; p < 0.001), Hydroxychloroquine (+ 1375.0%; p = 0.017), Sulfasalazine (+ 20.0%; p = 0.010), Azathioprine (+ 11.6%; p = 0.055), Tocilizumab (+ 434.4%; p < 0.001), Actemra (+ 87.1%; p < 0.001), Anakinra (+ 91.0%; p < 0.001), Arcalyst (+ 36.7%; p = 0.049), Baricitinib (+ 153.2%; p < 0.001), and Olumiant (+ 42.8%; p = 0.039) showed a statistically significant increase; conversely, Advil (− 8.6%; p < 0.001), Indomethacin (− 10%; p = 0.023), Indocin (− 21.7%; p = 0.006), Naproxen (− 8.3%; p < 0.001), Aleve (− 11.8%; p < 0.001), Mobic (− 26.3%; p < 0.001), Prednisone (− 7.2%; p = 0.018), Methotrexate (− 6.6%; p = 0.071), Sulfasalazine (+ 20.0%; p = 0.010), Arava (− 20.6%; p = 0.010), CellCept (− 21.9%; p < 0.001), Imuran (− 22.5%; p < 0.001), Cytoxan (− 20.4%; p < 0.001), Gengraf + Neoral + Sandimmune (− 29.1%; p < 0.001), Etanercept (− 24.7%; p = 0.029), Enbrel (− 24.8%; p = 0.004), Remicade (− 24.6%; p < 0.001), Humira (− 15.4%; p = 0.002), Golimumab (− 40.0%; p = 0.002), Simponi (− 12.3%; p = 0.042), Taltz (− 21.6%; p = 0.004), Rituxan (− 25.0%; p < 0.001), and Tofacitinib (− 22.0%; p = 0.012) displayed a statistically significant decrease compared to preceding 3 years (Table 1).

During the July 5–August 29, 2020, period, Ibuprofen (+ 6.7%; p = 0.015), Advil (+ 3.4%; p = 0.036, Diclofenac (+ 11.6%; p < 0.001), Voltaren (+ 118.8%; p < 0.001), Glucocorticoids (+ 22.3%; p = 0.021), Colchicine (+ 17.1%; p < 0.001), Febuxostat (+ 63.3%; p < 0.001), Hydroxychloroquine (+ 2125.0%; p = 0.046), Tacrolimus (+ 18.2%; p = 0.005), Tocilizumab (+ 400.0%; p < 0.001), Actemra (+ 61.5%; p < 0.001), Anakinra (+ 71.2%; p < 0.001), Infliximab (+ 32.6%; p < 0.001), Tremfya (+ 24.0%; p = 0.017), and Baricitinib (+ 88.2%; p < 0.001) showed a statistically significant increase; conversely, Indocin (− 26.0%; p < 0.001), Naproxen (− 8.5%; p < 0.001), Aleve (− 7.7%; p = 0.004), Mobic (− 20.6%; p < 0.001), Uloric (− 45.3%; p < 0.001), CellCept (− 13.2%; p = 0.023), Imuran (− 19.2%; p = 0.003, Cytoxan (− 21.5%; p = 0.005), Cyclosporine (− 7.8%; p = 0.021), Gengraf + Neoral + Sandimmune (− 29.9%; p = 0.018), Arcalyst (− 43.4%; p < 0.001), Enbrel (− 19.6%; p < 0.001), Remicade (− 19.0%; p < 0.001), Humira (− 11.0%; p = 0.010), Certolizumab (− 27.8%; p = 0.027), Secukinumab (− 27.7%; p = 0.039), Cosentyx (− 14.2%; p = 0.011), Ixekizumab (− 29.6%; p = 0.005), Rituxan (− 22.1%; p < 0.001), and Tofacitinib (− 26.8%; p = 0.012) displayed a statistically significant decrease compared to preceding 3 years (Table 1).

Discussion

We found statistically significant increases in RSV for colchicine, hydroxychloroquine, tocilizumab (and its brand name-Actemra), and anakinra, and statistically significant decreases among brand names of immunosuppressive agents (i.e., mycophenolate mofetil, azathioprine, cyclophosphamide, tacrolimus, and cyclosporine) during both the initial- and short-term COVID-19 periods as compared to overlapping periods of the preceding 3 years.

In a previous study investigating Google searches indicative of increased purchases of chloroquine and hydroxychloroquine, Liu et al. showed that demand for chloroquine and hydroxychloroquine increased following endorsements by Elon Musk and President Donald Trump [29]. In our study, we detected a statistically significant increase in RSV of hydroxychloroquine in all three periods investigated, indicating continued interest in hydroxychloroquine by the general public.

Aside from hydroxychloroquine, we showed an increased interest in colchicine, tocilizumab, and anakinra in both initial and short terms of COVID-19. The increased public interest in colchicine, tocilizumab, and anakinra may be attributed to media coverage of scientific studies investigating these drugs as potential therapeutic agents for the COVID-19 [32,33,34,35,36,37]. This finding also indicates that the general public has informational needs for these drugs. Given that the evidence on their efficacy, safety, or use for COVID-19 has been evolving, we recommend readers to refer to the recent version of National Institutes of Health (NIH) and Infectious Diseases Society of America (IDSA) treatment guidelines regarding the treatment and management of COVID-19 [14, 15].

In the initial period (March 15–May 9, 2020), the RSV of the term “NSAIDs” showed a statistically significant increase in RSV. Although statistically insignificant, the RSV of ibuprofen has also increased 73% in this period compared to the preceding 3 years. Between March and April 2020, the WHO provided a series of recommendations for the use of NSAIDs to treat COVID-19 that garnered public attention. In March 2020, the WHO initially advised against using ibuprofen to treat COVID-19 [38]; however, they later updated its recommendation by tweeting, “based on currently available information, WHO does not recommend against the use of ibuprofen” [39]. In April 2020, WHO published an overview on the use of NSAIDs in patients with COVID-19 [40]. Therefore, the initial trends observed in NSAIDs searches might be due to the discussions on the use of NSAIDs in patients with COVID-19.

It is notable that the RSV of brand names of immunosuppressive agents (i.e., mycophenolate mofetil, azathioprine, cyclophosphamide, tacrolimus, and cyclosporine) was reduced during the COVID-19 period compared to 2017–2019. It can be speculated that these agents have been initiated/added to the treatment regimens of patients with a rheumatic disease less often during the COVID-19 period, as these drugs have previously been associated with an increased risk of certain viral infections [41]. However, this assumption needs further validation. The more likely explanation for the observed decrease is that these drugs were initiated/prescribed less frequently during the COVID-19 period due to a general decrease in transplant surgeries [42] as these drugs are also used to prevent tissue rejection after an organ transplant. The decrease in the initiation of these drugs may lead to an observed decrease in RSV of these drugs, as patients who were prescribed a new immunosuppressive drug may search for online information. This presumption requires confirmation as well. Additionally, to our knowledge, it should be noted that there is currently no evidence on a possible association between an increased risk of COVID-19 and these immunosuppressive agents. Furthermore, the American College of Rheumatology (ACR) guidelines regarding the management of rheumatic diseases during the COVID-19 pandemic (version 2) recommend that conventional (DMARDs), immunosuppressants, biologics, JAK inhibitors, and NSAIDs may be continued in patients with stable rheumatic disease in the absence of COVID-19 infection or SARS-CoV-2 exposure [9]. However, as the literature is rapidly evolving, we recommend readers to refer to the recent version of ACR, European League Against Rheumatism (EULAR), and African League Against Rheumatism (AFLAR) guidelines on the management of rheumatic diseases during the COVID-19 pandemic [9, 10, 43].

Previous studies investigated the population-level interest in several treatment approaches during the COVID-19 era [25,26,27,28]. Bhambhvani et al. examined the population-level interest in elective urology procedures categorized into 4 categories: male infertility, erectile dysfunction, Peyronie’s disease, and vasectomy. They showed a decrease across all categories in the 30 days prior to March 18, 2020, period compared to 30 days after and March–May 2020 period compared to January 2015–February 2020 [25]. Jella et al. showed a decrease in population-level interest in knee and hip arthroplasties during March 2020, which breeched the lower control limit of previous 5 years [26]. In a study investigating population-level interest in facial plastic surgery from January 1, 2020, to July 10, 2020, Dhanda et al. showed a significant decrease during March–April 2020 compared to January–February 2020 [27]. Guzman and Barbieri examined the population-level interest in cosmetic procedures. They showed a statistically significant decrease in March 15–29, 2020, period compared to April 28, 2019,–March 8, 2020 (pre-pandemic period) [28]. All these previous studies showed a decrease in interest in treatment approaches investigated. Distinct from these previous studies investigating the procedural treatments, our study investigated pharmacological treatments (i.e., anti-rheumatic drugs). Also, as our study was conducted later, it was able to include a wider period of 2020. Therefore, it provides information not only on initial stage but also on short-term interest.

Limitations

Our study has some limitations. We used Google Trends, which captures the search behavior of people using the Google search engine, exclusively. However, it most likely represents American search queries because the Google search engine accounts for over 85% of all internet search volume in the USA [44]. Furthermore, because Google Trends does not provide demographic characteristics of the people who search on Google, the interest cannot be assessed by stratifying specific subpopulations. Therefore, our results can only be applied to the general population. Also, to control for possible seasonal variation in the searches [18, 20], we compared the COVID-19 period to preceding years (as similarly performed in previous studies [25, 26]); however, this approach may also introduce a limitation. The observed differences might be caused by decreases/increases in searches in 2017–2019 (e.g., Food and Drug Administration warning on Uloric in 2019 [45]). Therefore, our results should be interpreted with caution and need to be verified by further studies. Despite these limitations, our study would contribute to the knowledge on Google searches of anti-rheumatic drugs during the COVID-19 pandemic.

Conclusion

There were significant increases in RSV of colchicine, hydroxychloroquine, tocilizumab, and anakinra during both initial and short-term COVID-19 periods when compared to overlapping periods of the preceding 3 years reflecting a heightened level of information-seeking on these drugs during the pandemic. Rheumatologists should address this increase in informational demand. Further research assessing medium- and long-term interests in anti-rheumatic drugs is required to increase our knowledge on this new pandemic.