Perceptions of opioid use and impact on quality of life in patients with musculoskeletal conditions within online health community forums

DEAR EDITOR, Opioids for chronic pain have received considerable notoriety, especially owing to an ‘opioid epidemic’ affecting North America and several Western countries [1, 2]. Qualitative work about patient perceptions through interviews has identified unintended consequences of this epidemic, including increased anxiety about being prescribed opioids in Canada [3]. Across Europe, research evaluating the patient perspective of prescribed opioid use is scarce. This was highlighted by a recent Public Health England report, reporting limited understanding of patients’ experiences of prescription opioids, which can include physical, emotional and social side effects [4]. Social media provides a valuable platform for patients living with disease to discuss their positive/negative medication experiences, holistic impact and to seek peer support/advice. Such patient perspectives may not always be communicated to health-care professionals or coded in electronic health records. Additionally, these data can complement traditional methods of drug safety, particularly for signal detection of rare events, niche areas such as drug abuse or evaluating the impact on quality of life (QOL) even of medically non-serious adverse drug reactions (ADRs) [5, 6]. An example of such a platform is HealthUnlocked, Europe’s largest social network for health, with >1.4 million members in 500 patient communities covering >250 conditions. With musculoskeletal conditions being one of the most common indications for prescribing opioids, the aim of this study was to evaluate perceptions and key themes from discussions within musculoskeletal communities on the HealthUnlocked platform. University ethics approval was obtained (UREC reference: 2019-7922-11765). HealthUnlocked posts have no character limit, which allows unique opportunities for more detailed patient accounts. Of the 8021 posts on HealthUnlocked between 1 September 2015 and 1 September 2019, 7481 (93%) posts were from opioid users who provided consent for research. The date range was chosen for data availability and being before 2020 to avoid potential impact of the coronavirus disease 2019 pandemic on the results. Of the 7481 posts, 574 posts were from musculoskeletal-specific forums (RA, lupus, SSc, PMR/GCA, OA, FM and joint hypermobility), then de-identified by HealthUnlocked for research. Where geolocation was available, the majority of posts were from the UK. The average (mean) word count was 202 words per post and ranged between 8 and 1655 words. A manual thematic analysis was performed on a sample of 100 randomly selected posts, independently by two authors (H.R. and M.J.). To evaluate the impact on QOL, where described, themes associated with opioids were mapped to domains and subdomains (facets) of a QOL tool using all posts. The choice of which health-related QOL instrument to use was made by comparing multiple instruments, including musculoskeletal-specific QOL tools, by mapping a small number of posts with mention of a medication to each domain/facet (using a different HealthUnlocked dataset). WHO-QOL, developed by the World Health Organization, was found to be the best tool to capture the spectrum of QOL issues in this context. It yields a multi-dimensional profile across six domains, namely physical, psychological, level of independence, social relationships, environment and spirituality/religious beliefs [7]. Thematic analysis revealed key opioid-related themes related to ADRs (n1⁄445), ineffectiveness (n1⁄4 29) and opioid alternatives (n1⁄4 8), including using cannabidiol (CBD) oil, turmeric, probiotics and kratom. Less common themes were opioid withdrawal (n1⁄4 5), buying illicit opioids to alleviate chronic pain (n1⁄43) and effectiveness (n1⁄4 3). The ADRs most discussed were cognition-related effects, including brain fog (n1⁄4 8) and confusion (n1⁄4 4); gastrointestinal effects, such as nausea/vomiting (n1⁄411) and constipation (n1⁄44); and effect on sleep, including both insomnia (n1⁄47) and somnolence (n1⁄4 3). Also reported were headaches (n1⁄4 5), dizziness (n1⁄4 3) and rarer ADRs, including itching, sweating, seizures, memory loss, poor concentration and irritability. The presence of ADRs was accompanied by rich narratives highlighted in Table 1 (e.g. ‘Tonight, I have hit the Tramadol, so off to wacky land I go again’). Discussions about ADRs focused on several subthemes include their nature, probability of ADRs, impact on QOL, balancing harms with benefits, frustrations about the lack of information provided regarding ADRs and their addictive potential (Table 1). Effects on QOL most frequently mapped to the WHOQOL physical domain, which included facets of pain/discomfort, energy/fatigue and sleep/rest (n1⁄4 53). This was followed by impact on the psychological domain (n1⁄419; thinking, memory/concentration, self-esteem and negative Key message

Perceptions of opioid use and impact on quality of life in patients with musculoskeletal conditions within online health community forums DEAR EDITOR, Opioids for chronic pain have received considerable notoriety, especially owing to an 'opioid epidemic' affecting North America and several Western countries [1,2]. Qualitative work about patient perceptions through interviews has identified unintended consequences of this epidemic, including increased anxiety about being prescribed opioids in Canada [3]. Across Europe, research evaluating the patient perspective of prescribed opioid use is scarce. This was highlighted by a recent Public Health England report, reporting limited understanding of patients' experiences of prescription opioids, which can include physical, emotional and social side effects [4].
Social media provides a valuable platform for patients living with disease to discuss their positive/negative medication experiences, holistic impact and to seek peer support/advice. Such patient perspectives may not always be communicated to health-care professionals or coded in electronic health records. Additionally, these data can complement traditional methods of drug safety, particularly for signal detection of rare events, niche areas such as drug abuse or evaluating the impact on quality of life (QOL) even of medically non-serious adverse drug reactions (ADRs) [5,6]. An example of such a platform is HealthUnlocked, Europe's largest social network for health, with >1.4 million members in 500 patient communities covering >250 conditions. With musculoskeletal conditions being one of the most common indications for prescribing opioids, the aim of this study was to evaluate perceptions and key themes from discussions within musculoskeletal communities on the HealthUnlocked platform. University ethics approval was obtained (UREC reference: 2019-7922-11765).
HealthUnlocked posts have no character limit, which allows unique opportunities for more detailed patient accounts. Of the 8021 posts on HealthUnlocked between 1 September 2015 and 1 September 2019, 7481 (93%) posts were from opioid users who provided consent for research. The date range was chosen for data availability and being before 2020 to avoid potential impact of the coronavirus disease 2019 pandemic on the results. Of the 7481 posts, 574 posts were from musculoskeletal-specific forums (RA, lupus, SSc, PMR/GCA, OA, FM and joint hypermobility), then de-identified by HealthUnlocked for research. Where geolocation was available, the majority of posts were from the UK. The average (mean) word count was 202 words per post and ranged between 8 and 1655 words. A manual thematic analysis was performed on a sample of 100 randomly selected posts, independently by two authors (H.R. and M.J.). To evaluate the impact on QOL, where described, themes associated with opioids were mapped to domains and subdomains (facets) of a QOL tool using all posts. The choice of which health-related QOL instrument to use was made by comparing multiple instruments, including musculoskeletal-specific QOL tools, by mapping a small number of posts with mention of a medication to each domain/facet (using a different HealthUnlocked dataset). WHO-QOL, developed by the World Health Organization, was found to be the best tool to capture the spectrum of QOL issues in this context. It yields a multi-dimensional profile across six domains, namely physical, psychological, level of independence, social relationships, environment and spirituality/religious beliefs [7].
Thematic analysis revealed key opioid-related themes related to ADRs (n ¼ 45), ineffectiveness (n ¼ 29) and opioid alternatives (n ¼ 8), including using cannabidiol (CBD) oil, turmeric, probiotics and kratom. Less common themes were opioid withdrawal (n ¼ 5), buying illicit opioids to alleviate chronic pain (n ¼ 3) and effectiveness (n ¼ 3). The ADRs most discussed were cognition-related effects, including brain fog (n ¼ 8) and confusion (n ¼ 4); gastrointestinal effects, such as nausea/vomiting (n ¼ 11) and constipation (n ¼ 4); and effect on sleep, including both insomnia (n ¼ 7) and somnolence (n ¼ 3). Also reported were headaches (n ¼ 5), dizziness (n ¼ 3) and rarer ADRs, including itching, sweating, seizures, memory loss, poor concentration and irritability. The presence of ADRs was accompanied by rich narratives highlighted in Table 1 (e.g. 'Tonight, I have hit the Tramadol, so off to wacky land I go again'). Discussions about ADRs focused on several subthemes include their nature, probability of ADRs, impact on QOL, balancing harms with benefits, frustrations about the lack of information provided regarding ADRs and their addictive potential (Table 1).
Effects on QOL most frequently mapped to the WHO-QOL physical domain, which included facets of pain/discomfort, energy/fatigue and sleep/rest (n ¼ 53). This was followed by impact on the psychological domain (n ¼ 19; thinking, memory/concentration, self-esteem and negative  Letter to the Editor feelings facets) and the level of independence domain (n ¼ 10), with impact on activities of daily living and work capacity. Less commonly discussed impacts were on the social relationships domain, with influence on personal relationships and sexual activity (n ¼ 4), and the environment domain that includes physical safety and financial resources (n ¼ 4). Patient narratives can reveal important and under-recognized themes that may not be available through traditional methods of pharmacovigilance and pharmacoepidemiology. Our analysis revealed that ADRs infrequently discussed in clinics, such as brain fog/confusion, withdrawal and addictive potential of prescribed opioids in musculoskeletal patients, can have considerable impact on patients' QOL. Patient-generated data have the ability to provide rich information using social media and mobile health, through exploration of the nature, probability and importance of ADRs to the person experiencing them [8]. In the future, automated text-mining using natural language processing to perform topic modelling and sentiment analysis on larger datasets could help to determine emergent themes that provide valuable insights about patient experiences.

Data availability statement
Anonymized data are available via HealthUnlocked for a fee and will be subject to ethics approval and contractual agreements. The dataset used for this analysis cannot be shared with third parties due to contractual agreements and information governance issues.