Google search data as a novel adjunct to patient and public involvement in rheumatology research

Patient and public involvement is essential in the design and implementation of research studies to ensure research remains relevant and in line with public priorities. Public views on a given area of research may be sought via platforms such as focus groups or surveys. Here, we present the use of an openly available Google search data query tool, which may be used alongside traditional forms of patient and public involvement in research to highlight public perceptions and priorities. We used an online search query tool (“AnswerThePublic.com”) to explore public Google searches relating to “arthritis,” and an exemplar rheumatic disease, “rheumatoid arthritis.” The most common searches relating to these diseases included quality of life, treatment, prognosis, as well as impacts on life, including work. However, they also reveal concerns that may be more difficult to elicit in face-to-face focus groups, such as questions on alcohol consumption in arthritis, and impacts on mental health. Using public search engine data in research, alongside the important traditional methods of patient and public involvement, is a cost-effective and time-efficient method of gauging public views and concerns on a given topic. It may facilitate broad scoping searches of public priorities and help to guide future research questions.


Introduction
Patient and public involvement (PPI) is integral to the design and prioritisation of research studies. The National Institute for Health Research (NIHR) defines public involvement in research as "research being carried out 'with' or 'by' members of the public rather than 'to', 'about' or 'for' them" [1]. PPI are typically conducted by funders and researchers to prioritise research questions, and offering advice as part of a project steering group. The term 'public' denotes a wide range of individuals, including patients, potential patients, carers, people using health and social care services, and those who represent them [1]. It is important to distinguish between the views of the public, and those with a professional role in health and social care, especially when considering research design and prioritisation. This facilitates optimal impact and improved enrolment to studies, especially if those with the condition under study are involved from the study's inception [2]. In the rheumatic diseases, undertaking PPI ensures research remains patient-centred and clinically relevant, across a range of chronic and often debilitating conditions, treated with a wide range of drugs, including immunosuppressants and analgesics.
There is ample evidence demonstrating that PPI is necessary to optimise research design and output [2][3][4]. However, at the earliest stages of designing a study, it can be useful to gain a broad overview of public perceptions of a given topic, ensuring increased relevance and potential for wider benefit. Furthermore, it can be difficult to have access to focus groups, especially for junior researchers and clinicians. This has been made all the more difficult by social-distancing measures implemented worldwide due to the COVID-19 pandemic, which are likely to have lasting consequences for months to years. In outpatient specialties such as rheumatology, a quick and easy method of gauging public priorities for research can be through questionnaires and surveys conducted in clinic, but access to this may also be limited in the short-medium term due to increased use of telephone consultations.
There is no replacement for PPI, for example in the form of stakeholder involvement, individual interviews, and focus groups, which will always be necessary to ensure adequate prioritisation and relevance of research. However, largescale search query tools can overcome some of the issues described above. These openly available online tools are able to collate and display keyword suggestions and predictions based on searches within Google, which is used for 78% of internet searches. With 3.5 billion searches conducted per day through Google, summarising searches this way can provide an excellent overview of public concerns and queries on a given topic [5]. They are already commonplace in marketing, to ensure content is relevant and up-to-date. An example of such a tool includes "AnswerThePublic.com" [6], which is able to fetch data on phrases and questions entered into Google, featuring certain keywords (as entered into the tool by the user), Results are presented in both a tabulated and infographic format, making it efficient to ascertain search patterns and popular phrases.
In healthcare, search query tools may not only serve to confirm prior knowledge from PPI and clinician perceptions, but also reveals searches on more sensitive topics, such as opioid dependence and psychological impact of rheumatic disease, which service-users may not necessarily wish to discuss face-to-face.
We aimed to demonstrate the utility of a search query tool to complement PPI in the rheumatic diseases, using an exemplar rheumatic disease, rheumatoid arthritis.

Methods
We used an openly available search query tool to collate and map Google searches for two separate terms: "arthritis" and "rheumatoid arthritis." We opted to use "AnswerThePublic. com", due to ease of use and the resulting visual format of the results. Data on searches performed in Google are presented in real-time, so results are up-to-date at the time of using the search query tool. Since the results at any given time are identical regardless of the individual conducting the search, one author conducted the search for the chosen terms. Search queries are based on Google searches in the region in which it is accessed (e.g. UK); therefore, factors such as the user's own Google search history are irrelevant.
We first explored public Google searches for "arthritis" to gauge a broad overview of public queries on one of the most commonly-used terms in rheumatology, also used by patients and public. We then assessed search suggestions for "rheumatoid arthritis," as an exemplar rheumatic disease. Rheumatoid arthritis was selected as it is the most prevalent chronic inflammatory arthritis, and, therefore, likely to be a commonly-searched term in this field [7]. All searches were performed on 19 May 2020.
The output of results for each of the above two queries were delivered in the following categories: questions on the term; prepositions (i.e. searches incorporating additional terms such as "with" and "near"); comparisons (i.e. searches incorporating additional terms such as "and" and "versus"). Results were presented in lists as well as data visualisation images. Search query data for the UK, English language only, were interrogated. Figure 1 displays the results for search terms relating to questions on "arthritis," with up to eight of the most common terms shown for each, as an example of output produced by the search query tool. Within each sub-category, the darker stems denote more popular search terms, while paler stems denote less popular search terms.

Results
Subsequent searches are displayed in Tables 1, 2 and 3. Table 1 displays results for search terms relating to prepositions and comparisons to "arthritis". Tables 2 and 3 display the results for search terms relating to questions, prepositions, and comparisons to the term "rheumatoid arthritis," in the same manner as for "arthritis." Public searches performed on these topics predominantly relate to impacts on quality of life, relief of debilitating features including pain, and prognosis. On review of questions asked on arthritis, approximately 30% referred to debilitating features or quality of life, while just over 20% were on treatment or prognosis. In the prepositions' category, approximately 20% of searches were on debilitating features or quality of life, with a further 20% on treatment or prognosis.
For on "rheumatoid arthritis" searches, approximately 20% of search questions were on debilitating features or quality of life, with approximately 30% on treatment or prognosis. Looking at searches on rheumatoid arthritis, by prepositions, approximately 15% were on debilitating features or quality of life, with 30% on treatment or prognosis. The remainder of searches for both arthritis and rheumatoid arthritis covered topics including but not limited to anatomical sites of symptoms, possible associations with other rheumatological diseases (e.g. gout, systemic lupus erythematosus), investigations and diagnosis, and symptoms in animals.

Discussion
We used an openly available online tool to investigate Google searches for the terms "arthritis" and "rheumatoid arthritis," conducted by the public. Our results reveal striking patterns in search themes, with a focus on treatment, prognosis, quality of life, and debilitating features, including impact on work. While some results were novel, consistencies were seen with previous research in these areas, such as quality of life and patient-reported outcomes. These similarities with traditional methods of PPI suggest a role for a search query tool in this area.
One of the most common themes in searches for both arthritis and rheumatoid arthritis was on treatment. Specifically, some of the most frequently searched questions were "can arthritis be cured?", "can arthritis be treated?", "will arthritis go away?", and "rheumatoid arthritis-can it be cured?", as well as variations on these questions. This is consistent with previous studies on patient-reported health service needs. A recent scoping review identified that patients with inflammatory arthritis value discussions on treatment outside conventional medicine, such as alternative and complimentary therapies and dietary advice, as well as practical tools to help with daily activities [8]. Indeed, our review of Google searches revealed the following as some of Fig. 1 Data visualisation of the 8 most common questions asked on "arthritis," using the interrogative words: are, which, how, what, where, when, can, will, why, who. Darker green denotes an increased number of searches, while paler green denotes fewer searches the commonest searches: "arthritis and diet" and "arthritis can opener", consistent with the findings in this review and elsewhere in the literature [9,10].
Another common theme in the searches centred on quality of life, and debilitating aspects of arthritis relating to this. Some specific frequently-searched examples include "when arthritis becomes debilitating", "will rheumatoid arthritis cripple me?", and "what rheumatoid arthritis feels like". Quality of life outcomes have been explored increasingly in recent years in the rheumatic diseases. This has sought to increase clinical awareness of the burden of disease on mental health and daily living, with recommendations for factors such as health-related quality of life and fatigue to be routinely assessed in those with conditions including rheumatoid arthritis [11][12][13]. Our review of searches reveal that this remains an important priority in the public domain.
In addition, some specific frequent searches relating to quality of life, revealed public interest in aspects relating to work and arthritis. These included such examples as: "arthritis-can I claim PIP?", "rheumatoid arthritis-can you work?", and "is rheumatoid arthritis a disability?". We know from previous studies and patient-reported Table 1 Top: the eight most common search terms on "arthritis," using the prepositions: can, without, is with, for, to, near, bottom: the 8 most common search terms on "arthritis," using the comparators: outcomes that productivity at work, as well as presenteeism and absenteeism, are affected by all forms of arthritis [14][15][16]. However, these searches demonstrate more sensitive topics, such as eligibility for benefits, that patients may not feel able or comfortable to discuss during appointments. This highlights the need for clinicians to facilitate conversations regarding work and disability during clinic appointments, and the importance of the multi-disciplinary team, including psychologists and occupational therapists, and patient support services. Patient and public involvement, in rheumatology and elsewhere, has repeatedly highlighted the value placed on their relationship with health professionals, including ease of communication and approachability [8,17]. Consistent with this, some common Google searches included "rheumatoid arthritis doctors near me" and "rheumatoid arthritis support group near me". This again demonstrates the importance of access to patient, as well as clinician, support.
The co-existence of fibromyalgia with other rheumatic diseases has become more apparent in recent years, associated with poorer clinical outcomes in inflammatory arthritis [18,19]. Some of the most common public searches seek to explore this relationship e.g. "arthritis versus fibromyalgia" and "are arthritis and fibromyalgia connected?". These searches may also represent questions asked by patients during the diagnostic process for inflammatory arthritis or fibromyalgia, as they seek to explore the overlapping symptoms and relationship between the two diseases. This is also consistent with studies demonstrating a sense of 'invalidation' experienced by those with fibromyalgia, due to the invisibility and medically-unexplained nature of the syndrome, and the subsequent impact on quality of life outcomes in these patients [20,21]. Some searches did not follow a theme, or revealed unexpected patterns of search behaviour. One of the commonest phrases was "rheumatoid arthritis and COVID", consistent with the time at which we conducted this study (i.e. during the UK lockdown and shielding period for those at high-risk of complications of COVID-19, including some patients with rheumatoid arthritis). Other commons searches were on the genetics of rheumatoid arthritis ("is rheumatoid arthritis hereditary?" and "rheumatoid arthritis-is it genetic?"). This perhaps demonstrates a sensitive area which clinicians may not routinely address in clinic, but of importance to the patient. On a similar theme, searches such as "arthritis and alcohol" and "arthritis and sex" reveal similar sensitive topics which may prove potentially embarrassing for the patient to ask, leading them to seek answers on the Internet and online forums.

Future directions
Our use of a search query tool to explore searches conducted on "arthritis" and "rheumatoid arthritis" reveal known concerns from patients and the public, but also demonstrate areas of concern which may not be immediately apparent to researchers and clinicians. There is, therefore, scope to apply this tool as an adjunct to traditional methods of PPI, for example, when planning research in sensitive fields, such as opioid addiction and medication side-effects, for example, corticosteroids. It may also enhance and encourage research in areas including the impact of rheumatic disease on mental health. In the era of COVID-19 and telemedicine, such a tool provides a quick method of ascertaining public concerns and questions on its use, not only in rheumatology, but also other fields. Finally, as our internet and social media habits change, search query tools which are able to capture phrases used on social media may provide additional information on patient perceptions on healthcare-related topics. This has already been done through applications which are able to capture increased social media activity in relation to disease activity, including pain, but there is scope to extend further to incorporate use of specific phrases [22].

Limitations
A key limitation of such a tool is the inability to attribute all searches to a patient body. However, the tool can provide broad indications of public perceptions of a topic, within a given geographical region, e.g. UK. Furthermore, while searches on a topic such as arthritis may not change significantly from one day to another, searches conducted in Google on current and fast-changing situations may do so, which will affect the output from search query tools if conducted at two separate times. A prime example of this is search terms relating to COVID-19. In addition, as described above, search query tools are currently limited in number and can only capture data from the most commonly-used search engine, Google. Search query tools are complimentary, but not identical, to online platforms such as Google Trends which is able to display volumes and patterns of searches on a given topic, but no granularity on the nature of these searches. It is this additional information on extended phrases and questions entered into search engines which may enable their use as an adjunct to traditional PPI. As their utility grows in areas outside marketing, including in healthcare, possibly so too will their scope, including in areas such as social media.