There has been questionable effectiveness about the long-term use of paracetamol (Freo et al., 2021), with limited evidence supporting the efficacy of long-term use of paracetamol in RMDs (Abdel Shaheed et al., 2021). A systematic review showed paracetamol (4 g/day for 3–12 weeks) provided modest pain relief by 3.23 points on a 0–100-point pain scale (95% CI = −5.43, −1.02) for people with knee or hip OA (Leopoldino et al., 2019). The rest of the RMDs, by contrast, lack high-quality evidence on efficacy, in which chronic LBP, RA, non-cancer pain in children and adolescents, and neuropathic pain are supported by very low-quality evidence. More importantly, the evidence from a previous Cochrane review concluded that paracetamol (up to 4 g/day for up to 12 weeks) was not effective in reducing acute LBP (Saragiotto et al., 2016). In light of a short follow-up ranging from a few hours to 2 weeks after administration, most systematic reviews assessed the immediate treatment effect, making the effectiveness of paracetamol for chronic pain management in RMDs difficult to thoroughly evaluate.

Whilst generally deemed fairly safe there has been emerging evidence of specific adverse effects in chronic use (McCrae et al., 2018). Regular long-term use at higher doses has been associated with an increased risk of gastrointestinal (GI) bleeding and a small increase in systolic blood pressure (BP) (2–4 mmHg) (Table 2). Regular use of daily doses of ≥2–3 g paracetamol was associated with a potentially increased risk of upper GI bleeding, with most being observational studies in participants aged 40 or older or in those with a history of ischemic stroke (García Rodríguez and Hernández-Díaz, 2001; González-Pérez and Rodríguez, 2006; Rahme et al., 2008; Doherty et al., 2011; Gonzalez-Valcarcel et al., 2016). An randomised controlled trial (RCT) also supported a decrease in haemoglobin (≥1 g/dl) at 13 weeks observed in 20.3% of participants on the treatment of paracetamol 3 g/day (Doherty et al., 2011). This effect was additive when combined with NSAIDs.

BP increase has been another emerging concern with paracetamol (Turtle et al., 2013). The earliest study published in 1984 found an average of 4 mmHg increase in systolic BP when 3 g of paracetamol daily was administered for 4 weeks among patients with hypertension or OA (Chalmers et al., 1984). Subsequent RCTs had small sample sizes and showed inconsistent results, of which some supported an increase in BP (Radack et al., 1987; Sudano et al., 2010) but some against (Lewis et al., 1986). Similarly, most of the observational studies suggested that paracetamol in long-term use increased the risk of developing hypertension (Curhan et al., 2002; Dedier et al., 2002; Forman et al., 2007), with some conflicting evidence (Kurth et al., 2005; Dawson et al., 2013). Whilst several observational studies showed an association, the pain was often not measured and adjusted accordingly. An important confounding—uncontrolled pain could lead to high BP—of observational studies, however, would underestimate the association between paracetamol use and the change in BP, given the baseline BP might be higher. This could possibly explain the non-significant finding of the change in systolic BP in a cohort study (Dawson et al., 2013). The most recent NICE guideline defines hypertension as 140/90 mmHg and above, with a 10% or greater risk of developing cardiovascular disease (CVD) within the next 10 years (The Lancet, 2019; National Institute for Health and Care Excellence, 2019). A small increase in BP may be clinically important, especially for those with an increased baseline CVD risk. However, studies to date do not demonstrate these modest increase in BP has led to an increase in clinical endpoints such as stroke or myocardial infarctions (Table 2).

The PRECISION study (Nissen et al., 2016) was the first study using NSAIDs in high-CV-risk patients with OA or RA to assess the CV risk of COX2-selective (i.e., celecoxib 200 mg/day) and non-selective NSAIDs (i.e., ibuprofen 1,800 mg/day or naproxen 750 mg/day) (Table 2). Similar CV-event rates were observed between celecoxib vs. naproxen and ibuprofen (hazard ratio (HR) for celecoxib vs. naproxen: 0.90, 95% CI = 0.71, 1.15; HR for celecoxib vs. ibuprofen: 0.81, 95% CI = 0.65, 1.02), but GI tolerability was better for celecoxib (serious GI events: HR for celecoxib vs. naproxen: 0.71, 95% CI = 0.54, 0.93; HR for celecoxib vs. ibuprofen: 0.65, 95% CI = 0.50, 0.85) (Nissen et al., 2016). The risk of renal events was significantly lower with celecoxib than with ibuprofen but not with naproxen. The MEDAL program (Combe et al., 2009) that evaluated long-term use of COX2-selective NSAIDs also reported no difference in risk of thrombotic CV events in arthritis patients on long-term therapy with etoricoxib (90 mg/day) compared to diclofenac (150 mg/day). Despite the reassuring results, an increase in BP was noticed in both groups, and the rate of discontinuation due to hypertension was higher in the etoricoxib group (Combe et al., 2009). The CONCERN study (Chan et al., 2017) assessed the risk of GI events between COX2-selective (i.e., celecoxib 200 mg/day) and non-selective NSAIDs (i.e., naproxen 1,000 mg/day), both in combination with a prophylactic proton pump inhibitor. This trial included arthritis patients who had cardiothrombotic diseases requiring low-dose aspirin and a history of upper-GI-tract bleeding and followed up recurrent upper-GI-tract bleeding for 18 months. Celecoxib was found to be associated with fewer adverse GI-tract events than naproxen (HR = 0.44, 95% CI = 0.23, 0.82).

Over the last two decades, an increase in CVD risk has been the major concern with NSAIDs especially following two COX2-selective NSAIDs, rofecoxib and valdecoxib, were withdrawn from the market in 2004 and 2005 and deemed unsafe (U.S. Food and Drug Administration, 2018). Additionally in 2013, following a Europe-wide review of CVD safety, diclofenac was issued safety warnings and contraindicated in patients with established ischaemic heart disease, peripheral arterial disease, cerebrovascular disease and congestive heart failure (Medicines and Healthcare products Regulatory Agency, 2014; Coxib and traditional NSAID Trialists' (CNT) Collaboration Emberson et al., 2013; Schmidt et al., 2018). Regarding the newer COX2-selective NSAIDs, reassuringly a meta-analysis (Bally et al., 2017) showed no significant difference in the rate of acute myocardial infarction between celecoxib and non-selective NSAIDs, and celecoxib was the only COX2-selective NSAID with a lower risk of adverse CV and GI events compared with non-selective NSAIDs. The risk was greatest during the first month of NSAID use and with higher doses. Summarising the latest evidence it indicates that long-term use of celecoxib 200 mg/day may be considered for patients at increased CV risk, given the comparable risk of CV events and favourable profile of GI adverse events compared to non-selective NSAIDs (Bally et al., 2017). The 2007 scientific statement from the American Heart Association, however, advised that COX2 inhibitors should be used at the lowest possible dose and for the shortest possible time to minimise the risk of CV events until more long-term data on CV safety is available (Antman et al., 2007). In patients where vascular risks are a concern, the safest option appears to be naproxen when compared to other NSAIDs (Coxib and traditional NSAID Trialists' (CNT) Collaboration Emberson et al., 2013), emphasising the need to personalise the approach to prescribing all analgesics based on an individual’s baseline risk.

Recent evidence from a meta-analysis suggests that opioid use was associated with statistically significant but small clinical improvements in pain and physical functioning/disability among people with CNCP, accompanied by a higher risk of adverse effects (Busse et al., 2018). Opioid use was associated with reduced pain [weighted mean difference (WMD) = −0.69 cm (95% CI = −0.82, −0.56) on a 10-cm visual analogue scale] and improved physical functioning [WMD = 2.04 points (95% CI = 1.41, 2.68) on the 100-point 36-item short form physical component score] compared to the placebo (Busse et al., 2018). Compared with non-opioid alternatives including NSAIDs, tricyclic antidepressants (TCAs), or anticonvulsants, opioids showed similar associations with improvements in pain and physical functioning, with low-to moderate-quality evidence only (Busse et al., 2018). Another meta-analysis focusing on chronic LBP reported the short-term (4–15 weeks) use of strong opioids might have clinically relevant reductions in pain compared to placebo (Nury et al., 2022). However even short-term use was associated with increases in GI and nervous system adverse events.

Despite frequent long-term opioid use for CNCP management in RMDs, there has been no scientific evidence to support its efficacy but with increasing evidence of adverse events in this population (Anastasiou and Yazdany, 2022). To date, there is limited evidence on the efficacy and safety of opioid use in RA and SLE, with scarce data in other RMDs (Anastasiou and Yazdany, 2022). Research on RA cohorts supported weak opioids in short-term (<6 weeks) use for pain control, with a relative risk (RR) = 1.40 (95% CI = 1.07, 1.85) that favours opioids over placebo (Whittle et al., 2013). Opioids were also superior to placebo in RA patient-reported global impression of change, with an absolute risk difference of 18% (95% CI = 1, 41), a relative percent change of 44% (95% CI = 3, 103), and numbers needed to treat (NNT) as 6 (95% CI = 3, 84) (Whittle et al., 2011). In OA, an RCT examining chronic opioid therapy for moderate to severe chronic back pain or hip or knee OA found no significant difference in pain-related function over 12 months (Krebs et al., 2018). The pain intensity was however significantly better in the non-opioid analgesic group which also had fewer adverse medication-related symptoms (Krebs et al., 2018). For inflammatory RMDs that require biologic disease-modifying antirheumatic drugs (DMARDs) treatments, early opioids may improve pain in the short term, resulting in delayed DMARD initiation or reduced DMARD use (Boytsov et al., 2019; Kimsey et al., 2019).

Opioid use was related to increased vomiting (RR = 2.50, 95% CI = 1.89, 3.30) among people with CNCP (Busse et al., 2018). Opioid treatments also showed 3 times higher odds of adverse events (OR = 2.94, 95% CI = 2.33, 3.72) and 4 times higher odds of treatment discontinuation due to adverse events (OR = 4.04, 95%CI = 3.10, 5.25) among older adults with musculoskeletal pain (Megale et al., 2018).

There has been limited RMD-specific research on opioid safety for pain management, in which RA and OA are the most commonly studied. For RA patients, opioids in short-term use were more likely to report adverse events such as nausea, vomiting, dizziness and constipation (OR = 3.90, 95% CI = 2.31, 6.56) (Whittle et al., 2011), but the risk of withdrawals due to adverse events was inconsistent between studies (Whittle et al., 2011; Whittle et al., 2013). Opioid use was also associated with an increased risk of fracture in the RA cohort, with a greater risk observed in strong opiates (HR = 1.53, 95% CI = 1.24, 1.88) than in weak opiates (HR = 1.37, 95% CI = 1.18, 1.59) (Ozen et al., 2019). Moreover, both RA and systemic lupus erythematosus (SLE) patients had a higher risk of hospital admissions due to opioid overdose compared to other hospitalisations (Anastasiou et al., 2019). Regarding OA, several RCTs have shown that patients receiving opioid treatments experience more adverse events and have higher proportions of dropout due to adverse events than those on placebo or non-opioids (Etropolski et al., 2014; Serrie et al., 2017; Krebs et al., 2018; Kawai et al., 2022). In the 12th month, patients on opioids had a significant increase in medication-related symptoms by 0.9 (95% CI = 0.3, 1.5) compared to those on non-opioid treatments (Krebs et al., 2018). Previous work also suggested that tapentadol PR seemingly had a better GI tolerability profile (GI adverse event = 47.3%) than oxycodone CR (GI adverse event = 65.4%) (Etropolski et al., 2014; Serrie et al., 2017) (Table 2).

To date, there are only a few studies investigating long-term opioid therapy for more than 6 months, with no study following up more than 1 year (Chou et al., 2015; Krebs et al., 2018; Nury et al., 2022). Long-term opioid therapy (defined in this study as ≥6 months) for CNCP appears not to be superior to non-opioids in improvements of pain, disability, or pain-related function but shows more adverse events, including treatment discontinuation, opioid abuse or dependence and all-cause mortality (Krebs et al., 2018; Nury et al., 2022). In addition, there is no study evaluating the long-term effectiveness of different opioid dosing strategies such as short- plus long-acting opioids vs. long-acting opioids alone (Chou et al., 2015). In light of a lack of evidence demonstrating consistently improved pain control with long-term opioid use in RMDs but with increased risks of adverse events, the current evidence strongly suggests that opioids do not have a routine role in the CNCP management of inflammatory rheumatic diseases (Anastasiou and Yazdany, 2022).

Antidepressants for the treatment of LBP (Ferraro et al., 2021) and OA (Ferreira et al., 2021) have been found associated with small reductions in pain intensity or disability scores but the effect on pain might not be clinically important. In the meta-analysis assessing LBP (Ferraro et al., 2021), antidepressants showed a reduction of 4.33 points (95% CI = −6.15, −2.50, on a 0–100 scale) in pain intensity but had increased odds of stopping treatment for any reason (OR = 1.27, 95% CI = 1.03, 1.56) or due to adverse effects (OR = 2.39, 95% CI = 1.71, 3.34). In the meta-analysis evaluating back pain and OA (Ferreira et al., 2021), serotonin and norepinephrine reuptake inhibitors (SNRIs) reduced back pain by 5.30 points (95% CI = −7.31, −3.30, on a 0–100 scale, moderate evidence level) and OA pain by 9.72 points (95% CI = −12.75, −6.69, low evidence level) at 3–13 weeks. SNRIs were also found to decrease disability from back pain at 3–13 weeks (−3.55, 95% CI = −5.22, −1.88) and disability due to OA at 2 weeks or less (−5.10, 95% CI = −7.31, −2.89), with moderate evidence level respectively (Ferreira et al., 2021). Despite the efficacy, SNRIs were related to a high risk of adverse effects (RR = 1.23, 95% CI = 1.16, 1.30). Adverse events in this meta-analysis referred to various symptoms such as nausea (most prevalent), somnolence, back pain, diarrhoea, dizziness, transient ischaemic attack, and myocardial infarction, because they were defined by each study and varied noticeably across trials (Skljarevski et al., 1976; Skljarevski et al., 2009; Ferreira et al., 2021). TCAs and other antidepressants, by contrast, did not reduce pain or disability from back pain and had no available information about the treatment of OA.

The current evidence on the efficacy of antidepressants for musculoskeletal pain appears conflicting, leading to the discrepancy in guideline recommendations. NICE does not recommend the use of antidepressants for LBP (National Institute for Health and Clinical Excellence, 2016), while the American College of Physicians guidance suggests considering duloxetine as a second-line drug treatment for chronic LBP (Chou et al., 2017). Similarly, NICE does not make a specific recommendation on antidepressants for OA (National Institute for Health and Care Excellence, 2022b) but OARSI guidance makes a conditional recommendation on duloxetine for people with OA and widespread pain or depression (Bannuru et al., 2019). Despite the small effects at the group level reported in the present evidence, some treated individuals may gain a worthwhile benefit from antidepressants. For example, absolute effect sizes for physical treatments for LBP are of similar magnitudes to those reported in the previous review (Ferreira et al., 2021) and translate into NNT of between 5 and 9 (Froud et al., 2009; Underwood and Tysall, 2021).

Antidepressants such as amitriptyline, citalopram, duloxetine, fluoxetine, paroxetine or sertraline are suggested by NICE guidelines to manage chronic primary pain (e.g., fibromyalgia) for adults (National Institute for Health and Care Excellence, 2021a). Most antidepressants are off-label use since, to date, only three pharmacological agents—pregabalin (i.e., gabapentinoids), duloxetine and milnacipran—approved by the United States Food and Drug Administration to treat fibromyalgia. A meta-analysis evaluated the efficacy and harms of antidepressants in the management of fibromyalgia syndromes (Hauser et al., 2012). The standardised mean differences of SNRIs, selective serotonin reuptake inhibitors (SSRIs), and TCAs on pain, sleep, fatigue, depression and health-related quality of life were all significant, despite only a small effect size reported. The NNT was estimated as 10.0 (95% CI = 8.0, 13.4) for SNRIs, 6.3 (95% CI = 4.1, 14.1) for SSRIs, and 4.9 (95% CI = 3.5, 8.0) for TCAs (Hauser et al., 2012). The RR of dropouts due to adverse events was higher for SNRIs (1.83, 95% CI = 1.53, 2.18) but was not statistically different for SSRIs and TCAs (Hauser et al., 2012). Although antidepressants are increasingly prescribed for fibromyalgia as per guideline recommendations, physicians and patients should be realistic about the balance between the benefits and harms of antidepressants.

The inconsistent recommendations of antidepressants across different guidelines are challenging, given the limited evidence on antidepressants’ efficacy for different RMDs. The current evidence indicates that for non-fibromyalgia RMDs, antidepressants have no important benefit that is less acceptable, less safe, and less tolerable (Ferraro et al., 2021).