Evidence-based living recommendations for the treatment of inflammatory arthritis

The evidence considered by the panel for this recommendation specifically relates to the choice of methotrexate monotherapy versus methotrexate in combination with (one or more) other DMARDs as the initial therapy in people with rheumatoid arthritis (RA) who have not previously used DMARDs. The evidence is primarily drawn from a Cochrane systematic review and network meta-analysis of 93 studies in DMARD-naive participants ¹²⁷, and two RCTs published subsequently ¹²⁰¹²⁶.

Compared to MTX oral monotherapy:

• Several treatment combinations with methotrexate resulted in a better ACR50 response rate: methotrexate + sulfasalazine + hydroxychloroquine ('triple therapy'), methotrexate + several biologics (abatacept, adalimumab, etanercept, infliximab, rituximab, tocilizumab), and methotrexate + tofacitinib. The estimated probability of ACR50 response was similar between these treatments (range 56-67%), compared with 41% for methotrexate alone.
• ACR50 response rate was similar in people receiving 'triple therapy' (methotrexate + sulfasalazine + hydroxychloroquine) and in those receiving methotrexate + b/tsDMARDs.
• Radiographic progression was similar in people receiving triple therapy and methotrexate monotherapy. Compared with methotrexate alone, methotrexate combined with adalimumab, etanercept, certolizumab, or infliximab slowed radiographic progression slightly (but less than the minimal clinically important difference of five units on the Sharp-van der Heijde scale over 1 year).
• Withdrawals due to adverse events were similar in people receiving triple therapy compared with those receiving methotrexate monotherapy. Methotrexate + azathioprine had more withdrawals due to adverse events compared with methotrexate monotherapy, and triple therapy had fewer withdrawals due to adverse events than methotrexate + infliximab.

The panel elected to vote on the benefits and harms of MTX monotherapy vs all MTX-based combinations (including combination therapy with b/tsDMARDs that are not currently PBS-subsidised in Australia for initial therapy) but noted that, in practice, the available choices for Australian patients are limited to combinations of csDMARDs.

For treatment response (measured by ACR50), evidence was moderate certainty for csDMARD triple therapy versus methotrexate monotherapy (downgraded for indirectness), and moderate or high certainty for most other comparisons.

For radiographic progression and withdrawals due to adverse events, evidence was moderate certainty for csDMARD triple therapy versus methotrexate monotherapy (downgraded for imprecision), and low to moderate certainty for most other comparisons.

There are some limitations in applicability of the evidence, including the following:

• the effect of glucocorticoid therapy in addition to DMARDs was not assessed
• the analysis did not directly compare the strategy of starting with methotrexate monotherapy in methotrexate-naïve patients and progressing to triple therapy versus the strategy of starting with triple therapy from the outset
• MTX combination therapies with newer b/tsDMARDs are not included in the evidence here (such trials will be captured in a new systematic review and network meta-analysis currently underway)
• adverse event data are limited to events likely to be reported in RCTs; potential rare long term adverse events are best captured within registries and do not form part of this evidence base

A recent systematic review of patient preferences for DMARD treatment in RA included 36 studies published between 1990 and 2018, although the included studies generally involved those who had already commenced DMARDs and only two studies involved participants with early RA (i.e., those who are most likely to be making decisions about the initial choice of DMARD) ³⁵. The review found that, in general, people with RA consider treatment benefits (particularly improvement in symptoms and function) to be more important than both serious and non-serious adverse events (AEs). However some studies found evidence of important risk aversion (particularly regarding serious AEs), which suggests that there is considerable variation in preferences between individuals. Dosing and administration considerations, including route and frequency of medications, were generally found to be less important than treatment benefits but more important than AEs, although this also varied between and within studies. Overall, the results suggest that individual preferences are likely to be an important factor in shared decision-making regarding treatments for RA.

A study of 152 patients with early RA (<2 years since diagnosis) that used a discrete-choice experiment design also identified important heterogeneity in preferences ¹²³. In general, treatment benefits were considered the most important treatment attribute, however latent-class analysis also identified that approximately half of the patients were more risk averse (particularly to serious adverse effects) in comparison to the other half who were more benefit-driven. In this study sample, participants would be willing to tolerate an increase in pill burden (triple therapy versus weekly oral medication) for a relatively small increment in treatment benefits (<5% absolute increase in chance of major symptom improvement), although the incremental improvement required was considerably higher for those in the risk averse group.

The panel considered that while many people would be likely to choose the treatment regimen that offers the highest chance of improving their current symptoms (including pain), there was also evidence that there is likely to be substantial variability between individuals in their preferences and values.

We have not systematically reviewed the cost-effectiveness of methotrexate monotherapy versus combination therapy in DMARD-naïve patients with rheumatoid arthritis. There is some evidence that the use of triple therapy as the initial therapy is more cost-effective than initial therapy with methotrexate monotherapy in people with very early inflammatory arthritis in a European setting ¹²², although the estimates in this population (in which relatively large doses of glucocorticoids were used as an adjunct to csDMARDs) may not apply to the current Australian context.

Although there are no publicly available data regarding the actual comparative costs of different bDMARDs to the payer (i.e. the Federal Government) in Australia, b/tsDMARDs are generally far more expensive than csDMARDs and therefore the cost implications of b/tsDMARD use are likely to be an important consideration at the societal level and for funders, particularly if there are marginal differences in efficacy.

At the individual patient level the direct cost implications are less profound for those who meet Government funding criteria, however currently DMARD-naïve patients with RA are not eligible for subsidised treatment with any b/tsDMARD in Australia. The impact of prescription dispensing fees may be higher for those using combination csDMARDs versus monotherapy, although the absolute cost difference and its impact is likely to vary between individual patients.

We did not identify any data regarding the impact of methotrexate monotherapy versus combination therapy with csDMARDs or b/tsDMARDs on the environment. It is plausible that combination therapy (particularly with bDMARDs) may have a larger negative impact on the environment due to manufacturing, packaging and transport, however the magnitude of such an impact (if present) or other indirect effects is currently unknown.

The panel discussed the implications of the large cost difference between csDMARDs and b/tsDMARDs. The cost implications of MTX monotherapy versus combination therapy with csDMARDs are likely to be small, but the current cost of b/tsDMARDs would have a major impact on their potential use as an initial therapy for RA.

There are a number of factors that may influence the health opportunities and outcomes in individual patients. Such factors (based on the PROGRESS-Plus characteristics) might include poor health literacy; residence in rural, remote or relatively under-serviced locations; primary language other than English; low educational attainment; the presence of disability; or adverse socioeconomic or social circumstances. Some of these may be important considerations in the choice of initial DMARD therapy, particularly DMARD regimens that involve complex dosing, regular dose adjustments, or frequent monitoring for adverse effects, and as a result may have an impact on equity if the opportunity to access initial combination therapy varies between groups.

The panel noted the importance of access to a rheumatologist. It considered that while some other practitioners (e.g., GPs, general physicians) may consider prescription of csDMARD monotherapy for people with early RA who are unable to promptly access a rheumatologist, initial combination therapy is likely to require more specialist expertise. As such, people who are less able to access immediate specialist care may be less likely to receive combination therapy.

The panel also noted that some members of the Australian community do not have access to Medicare (e.g., refugees, international students). These people may experience additional barriers to the use of combination DMARD therapy (including the cost of medical appointments and medications), which may lead to additional health inequity.

There is evidence to suggest variation in the use of csDMARDs in combination versus monotherapy as initial therapy for RA ^125129130. It is likely that this heterogeneity in practice is influenced by several factors, including variation in the preferences of patients and prescribers for the various treatment options. Rheumatologists' beliefs regarding the effectiveness of triple therapy have been shown to vary considerably and to impact on their interpretation of the evidence base and choice of treatment options ¹²⁵.

A study that compared RA treatment recommendations developed by separate panels comprised of either consumers or clinicians found that when presented with the same data regarding the benefits and harms of triple therapy versus methotrexate monotherapy in DMARD-naïve RA, the consumer panel conditionally favoured triple therapy whereas the clinician panel conditionally favoured MTX monotherapy ¹²¹.

A discrete choice experiment that asked patients with early RA to choose between hypothetical treatment options estimated that, after weighing network meta-analysis data on benefits and harms in addition to other considerations including dosing, the majority of patients (78%) would prefer triple therapy over MTX monotherapy as the initial treatment for RA ¹²⁸.

The consumer panellist noted that many people who have been recently diagnosed with RA are anxious about commencing MTX and that, in comparison, the additional burden of combination therapies in addition to MTX is often considered to be relatively minor.

Initial treatment of rheumatoid arthritis with methotrexate alone or in combination with other csDMARDs is consistent with current practice in Australia and there are unlikely to be important feasibility implications. Use of b/tsDMARDs as part of an initial treatment combination in DMARD-naïve patients does not meet current subsidised funding criteria in Australia and therefore would not be feasible to widely implement. As a result of this dichotomy, the panel voted in favour of 'important issues or potential issues not investigated'.

Compared with oral methotrexate, subcutaneous or intramuscular methotrexate may very slightly increase the number of people achieving ACR50 and may improve pain. However, compared with oral methotrexate, subcutaneous or intramuscular methotrexate may or may not improve disease activity or function. No studies compared the effect of subcutaneous or intramuscular methotrexate with oral methotrexate on quality of life.

We are uncertain whether subcutaneous or intramuscular methotrexate results in fewer withdrawals due to adverse events or fewer serious adverse events as the certainty of the evidence is very low. Subcutaneous or intramuscular methotrexate may have little or no effect on the number of adverse events compared with oral methotrexate.

There is LOW certainty evidence that subcutaneous or intramuscular methotrexate may improve pain and may slightly increase the number of people achieving ACR50 compared to oral methotrexate, and may or may not improve disease activity or function.

There is LOW certainty evidence that subcutaneous or intramuscular methotrexate may have little or no effect on the number of adverse events, and VERY LOW certainty evidence regarding withdrawals due to adverse events or serious adverse events compared to oral methotrexate.

No studies reported data on quality of life.

A systematic review of patient preferences for DMARD treatment in RA was identified that included 36 studies published between 1990 and 2018. The included studies generally involved participants who had already commenced DMARDs ⁸⁴. The review found that, in general, people with RA consider treatment benefits (particularly improvement in symptoms and function) to be more important than both serious and non-serious adverse events (AEs). However some studies found evidence of important risk aversion (particularly regarding serious AEs), which suggests that there is considerable variation in preferences between individuals. Dosing and administration considerations, including route and frequency of medications, were generally found to be less important than treatment benefits but more important than AEs, although this also varied between and within studies.

Evidence also exists that patients' perception of competing risks of benefits and harms is affected by prior experience with DMARDs, including prior DMARD-related adverse effects ¹⁸³.

The anticipated variability in preferences and values between patients underscores the importance of shared decision-making between clinician and patient regarding treatments for RA.

In general, methotrexate is an inexpensive medication. Injectable preparations of methotrexate, particularly pre-filled syringes, are more expensive than oral methotrexate. While the absolute cost difference to the individual patient is relatively low, the impact is likely to vary. Optimisation of the efficacy and tolerability of methotrexate might plausibly reduce the proportion of people with RA who progress to b/tsDMARDs, which are considerably more expensive medications.

We did not identify any data regarding the differential impact of oral versus injectable methotrexate on the environment. It is plausible that injectable methotrexate (especially pre-filled syringe formulations that are administered via single-use plastic devices) may have a larger negative impact on the environment due to manufacturing, packaging and transport, however the magnitude of such an impact (if present) or other indirect effects is currently unknown.

The panel debated the extent to which resource considerations might be important for this recommendation. Overall, despite the relatively low absolute cost of methotrexate, the potential for even relatively minor cost differences to have a large impact on some consumers and the potential additional environmental impact of injectable methotrexate were judged by the panel to be important considerations.

There are a number of factors that may influence health opportunities and outcomes in individual patients. Such factors (based on the PROGRESS-Plus characteristics) might include poor health literacy; residence in rural, remote or relatively under-serviced locations; primary language other than English; low educational attainment; the presence of disability; or adverse socioeconomic or social circumstances. Some of these may have an impact on equitable access to the use of injectable methotrexate.

The panel considered the degree to which equity factors may influence the recommendation regarding the route of administration of methotrexate, and whether this was sufficient to represent ‘important issues’ for this recommendation. Overall the panel agreed that equity considerations are relevant to this recommendation, particularly in regard to people with limited access to healthcare and those for whom self-administration of an injectable medication is likely to be difficult or unachievable due to hand arthritis or other disability.

It is likely that subcutaneous administration of methotrexate would be acceptable to most stakeholders, including many people with RA, although there may be variation between individuals based on their particular circumstances, prior treatment experience, and preferences. There is some evidence that people with RA often prefer medications administered orally, although this preference varies according to age, duration of disease, and other factors ¹⁸². In general, people with RA are more likely to accept changes to their treatment regimen when they perceive that the change may result in an important improvement in their current symptoms and future prognosis ¹⁸³.

The panel agreed that acceptability of methotrexate injections may vary between individuals. While this is unlikely to be a major consideration, the panel preferred a judgment of 'important issues, or potential issues not investigated' to 'no important issues'.

Both oral and injectable methotrexate are generally widely available and affordable in Australia.

Injectable methotrexate is currently available either in vials (50mg/2mL) or pre-filled syringes. Use of the vial requires the user to access needles and syringes (typically an insulin syringe) and draw up the correct dose of methotrexate before administration. This requires manual dexterity and sufficient numeracy and visual acuity, which may be a barrier for some patients, particularly those without ready access to a carer or healthcare provider at the time of each weekly dose.

The pre-filled syringe may afford fewer barriers to use, although it also requires confidence in self-administration of an injection or access to a carer or healthcare provider. Currently the pre-filled syringe (brand name Trexject) is only available on the Australian Pharmaceutical Benefits Scheme via a Streamlined Authority prescription for people with "severe active rheumatoid arthritis" who "must be unsuitable for administration of an oral form of methotrexate for this condition". This is also likely to represent a barrier to the use of injectable methotrexate for the treatment of RA, particularly among people with early RA or those who are methotrexate-naïve.

The panel felt that while the potential feasibility barriers described were minor, they may still be a factor in some cases, particularly for people with RA who live with a physical or cognitive disability, and therefore may impact on this recommendation.

The panel considered a set of outcomes from a living network meta-analysis (NMA) comparing the benefits and harms of different DMARDs in adults with RA who have failed to respond to a TNF inhibitor (TNFi).

A summary of the initial key findings of the NMA is as follows:

1. Compared with placebo, at 6 months follow-up:
   • Continuation or switch in route of administration of the failed TNFi did not result in an improvement in disease activity as measured by ACR50 response or DAS28
   • Several biologic or targeted synthetic DMARDs (b/tsDMARDs) resulted in a better ACR50 response rate: a TNFi not previously tried, tocilizumab (4mg/kg IV Q4W or 8mg/kg IV Q4W), abatacept (10mg/kg IV Q4W with or without IV loading dose), rituximab (1000mg IV), tofacitinib (5mg po BD), baricitinib (4mg po OD), upadacitinib (15mg po OD), sarilumab (200mg sc)
   • Abatacept (125mg sc Q1W), baricitinib (2mg po OD) and upadacitinib (6mg BD) were not superior to placebo in terms of ACR50 response rate
   • A different TNFi not tried previously, abatacept (125mg sc weekly) and rituximab (1000mg IV) improved mean disease activity. DAS28 data were not available in the network meta-analysis for tocilizumab or the JAK inhibitors (tofacitinib, baricitinib, upadacitinib)
   • Upadacitinib (15mg po OD and 6mg BD) improved mean pain but baricitinib (2mg and 4mg po OD) did not; the network did not contain pain data for any other intervention
   • None of the biologics assessed in the network meta-analysis improved mean function
   • None of the biologics resulted in more withdrawals due to adverse events, or more serious adverse events; the network did not contain data on withdrawal due to adverse events for upadacitinib 15mg
2. None of the effective interventions were found to differ from any other in any of the efficacy or safety outcomes measured in the network meta-analysis

The panel considered the evidence from the NMA at length. There was general agreement that the NMA data demonstrated that in people with RA who have had an inadequate response to a TNFi, switching to a different therapy (either another TNFi or a b/tsDMARD with a different mechanism of action) was superior to continuation of the current therapy, or discontinuation of b/tsDMARD therapy altogether. The panel also agreed that there is likely to be little difference in efficacy between the various b/tsDMARD options, including TNFi, abatacept, IL-6 inhibitors, rituximab and the JAK inhibitors (JAKi). As a result, the panel considered that the balance of potential benefits and harms clearly favoured a switch in therapy but that the choice of b/tsDMARD would be determined by individual factors, including values, preferences and clinical circumstances, and also by an evaluation of potential harms.

The panel noted that the NMA did not demonstrate any difference in safety or tolerability between the various treatment options. However, RCTs designed to evaluate efficacy are typically not powered to estimate differences in adverse events and may not have long enough follow-up periods to detect important long-term harms. The panel also discussed the results of the ORAL Surveillance trial ¹⁸⁴. ORAL Surveillance was an RCT designed to assess the risk of important adverse events (major adverse cardiovascular events (MACE) and cancers) in people with RA who were 50 years of age or older and had at least one cardiovascular risk factor. Participants were randomised to treatment with either the JAKi tofacitinib or a TNFi (etanercept or adalimumab). In brief, the trial did not demonstrate non-inferiority of tofacitinib 5mg bd (the approved dose in Australia) versus TNFi for MACE (cardiovascular death, non-fatal myocardial infarction or stroke) or cancers (excluding non-melanoma skin cancers). The estimated number needed to treat for harm (NNH) with tofacitinib compared with a TNFi over a 5-year period for one additional MACE was 113. The 5-year NNH for cancers was 55. The incidence of serious infections (particularly herpes zoster) was also higher in those taking tofacitinib, and venous thromboembolism was more common in those allocated to the higher dose of tofacitinib compared with a TNFi. As a result of this trial, the European Medicines Agency (EMA) has recommended that JAKi be used in people at increased risk of cardiovascular disease or cancer only if no suitable treatment alternatives are available. Similarly, the US FDA has mandated that manufacturers provide warnings about an increased risk of MACE, malignancy, thrombosis, and mortality with the JAKi and has suggested that these medications be reserved for patients who have had an inadequate response or intolerance to one or more TNFi. The panel noted that the current recommendation applies to this population of people with RA who have already failed to respond to at least one TNFi.

The panel decided to vote separately on the balance of benefits and harms for bDMARDs and tsDMARDs. There was consensus that for both bDMARDs and tsDMARDs, there was a net benefit compared with no treatment (or continuation of the ineffective treatment) but that there was uncertainty regarding the comparative net benefit between bDMARDs and tsDMARDs.

Placebo was chosen as the primary comparator. The network estimates for the comparisons between the interventions and placebo were mostly of moderate to high certainty. Some comparisons were downgraded from high to moderate certainty for imprecision. For some interventions, primarily continuation or switch in route of the failed TNFi, the certainty of evidence was lower due to both imprecision and risk of bias in the included trials.

A systematic review of patient preferences for DMARD treatment in RA was identified that included 36 studies published between 1990 and 2018. The included studies generally involved participants who had already commenced DMARDs, including bDMARDs ⁸⁴. The review found that, in general, people with RA consider treatment benefits (particularly improvement in symptoms and function) to be more important than both serious and non-serious adverse events. However some studies found evidence of important risk aversion (particularly regarding serious adverse events), which suggests that there is considerable variation in preferences between individuals. Dosing and administration considerations, including route and frequency of medications, were generally found to be less important than treatment benefits but more important than AEs, although this also varied between and within studies.

It is plausible that individuals with RA who have not responded to a TNF inhibitor may place more weight on improvement in disease activity, although there is still likely to be important variation between individuals. A recent systematic review of patient-based benefit-risk assessment in people with a range of chronic diseases indicates that variation in individual risk tolerance is common, and may be higher in those with greater disease severity and longer treatment experience ¹⁸¹. Patients' perception of competing risks of benefits and harms are also affected by prior experience with DMARDs, including prior DMARD-related adverse effects ¹⁸³.

The anticipated variability in preferences and values between patients underscores the importance of shared decision-making between clinician and patient regarding treatments for RA.

We have not systematically reviewed the relative cost-effectiveness of the various interventions included in the current network meta-analysis. There are few existing data on the comparative cost-effectiveness of different sequences of b/tsDMARDs in RA ¹⁸⁰. However, in patients for whom a decision to institute b/tsDMARD therapy has already been taken, resource implications are unlikely to be a major determinant of the choice of subsequent DMARDs unless there is a clear difference in cost-effectiveness between the available alternative interventions.

Although there are no publicly available data regarding the actual comparative costs of different bDMARDs to the payer (i.e. the Federal Government) in Australia, b/tsDMARDs are expensive therapies and represent an important fraction of total pharmaceutical expenditure. While there are likely to be some cost differences between b/tsDMARDs, these differences may vary over time due to various factors including statutory requirements and market forces (including the advent of biosimilar drugs), and represent a relatively small proportion of the total cost. Therefore the cost implications of the choice between individual b/tsDMARDs are currently unlikely to be an important consideration at the societal level and for funders, particularly if there are marginal differences in efficacy, although this may vary over time.

At the individual patient level the direct cost implications are less profound for those who meet Government funding criteria. Currently, almost all Australian patients with RA who have had an inadequate response to a TNFi are eligible for subsidised treatment with an alternative b/tsDMARD. There may be some differences between medications with respect to prescription dispensing fees, particularly for bDMARDs for which co-prescription of methotrexate is required, although the absolute cost difference and its impact is likely to vary between individual patients.

We did not identify any data regarding the differential impact of the included interventions on the environment. It is plausible that some b/tsDMARDs (e.g. those that are administered via single-use plastic devices) may have a larger negative impact on the environment due to manufacturing, packaging and transport, however the magnitude of such an impact (if present) or other indirect effects is currently unknown.

There are a number of factors that may influence health opportunities and outcomes in individual patients. Such factors (based on the PROGRESS-Plus characteristics) might include poor health literacy; residence in rural, remote or relatively under-serviced locations; primary language other than English; low educational attainment; the presence of disability; or adverse socioeconomic or social circumstances. Some of these may be important factors in the choice of subsequent DMARD therapy, particularly DMARD regimens that involve complex dosing, regular dose adjustments, or close specialist supervision, and may have a particular impact on equity if the opportunity to access specialist care varies between groups.

It is likely that a switch to an alternative DMARD would be acceptable to most people with active RA who have not responded to a TNF inhibitor, although this may not apply equally to all treatment options and there may be variation between individuals based on their particular circumstances, prior treatment experience, and preferences. There is some evidence that people with RA often prefer medications administered orally, although this preference varies according to age, duration of disease, and other factors ¹⁸². In general, people with RA are more likely to accept changes to their treatment regimen when they perceive that the change may result in an important improvement in their current symptoms and future prognosis ¹⁸³.

In people with RA, switching from an ineffective or poorly-tolerated TNFi to an alternative b/tsDMARD (including a switch to a different TNFi) is consistent with current rheumatology practice in Australia. All of the interventions included in the network meta-analysis that informs this recommendation (aside from sarilumab) are currently subsidised under the Pharmaceutical Benefits Scheme for the treatment of active RA in people who have had an inadequate response to a TNF inhibitor. For this reason, there are unlikely to be important feasibility implications.

Dose reduction and discontinuation of conventional synthetic disease-modifying antirheumatic drugs (csDMARDs) were considered separately and each compared to continuation of csDMARDs. For each comparison, the analysis pooled three groups of participants:

In people using csDMARDs alone, dose reduction or discontinuation of csDMARDs versus continuation of csDMARDs In people using background b/tsDMARDs, dose reduction or discontinuation of csDMARDs (with continued background treatment of b/tsDMARD) versus continuation of csDMARDs (and continued dose of b/tsDMARD) In people using background b/tsDMARDs, dose reduction or discontinuation of both csDMARDs and b/tsDMARDs versus continuation of csDMARDs (with or without continued b/tsDMARD use)

Compared to continuation of treatment at the standard dose, dose reduction of csDMARDs (methotrexate monotherapy without background b/tsDMARDs in the majority) in participants with RA in remission or with low disease activity for at least 6 months:

• may increase the number of people experiencing a flare
• is probably comparable in terms of the number of people maintaining remission, disease activity and function
• may be comparable in terms of the number of people with progression of joint damage
• may be comparable in terms of the number of people experiencing a serious adverse event and the number of people withdrawing due to adverse events

Compared to continuation of treatment at the standard dose, discontinuation of csDMARDs (methotrexate combined with continued use of bDMARDs or a tsDMARD in the majority) in participants with RA in remission or with low disease activity for at least 6 months:

• slightly reduces the number of people maintaining remission and slightly worsens disease activity and function
• probably increases the number of people experiencing a flare and slightly increases the number of people with progression of joint damage
• is probably comparable in terms of the number of people experiencing a serious adverse event and the number of withdrawals due to adverse events

The characteristics of the participants recruited to the trials seem to largely reflect those in practice who may be considering discontinuation of their DMARD treatment; participants were in remission or low disease activity for 6 to 12 months in the majority of trials, apart from a single trial that included participants with low disease activity for only 4 months.

The panel debated the relative benefits and harms of both dose reduction and discontinuation of csDMARDs, and noted that the balance between the potential benefits and harms of dose modification was likely to be different for dose reduction compared to discontinuation. While the panel was initially split on whether the risk of increased disease activity with csDMARD dose modification represented a net “important harm”, there was consensus that the risk of disease flare in those who abruptly discontinue DMARDs was likely to be harmful whereas the risk of flare was much lower in those who reduce their DMARD dose, in which case the potential benefits may outweigh the potential harms for a proportion of patients. The consumer panelists both described the significant impact of disease flare or loss of remission, and noted that the experience and impact of a change in RA disease activity or medication adverse effects may vary between individuals and within the same individual over time.

Given the potential for a significant negative impact on RA disease control with discontinuation of csDMARDs, the panel voted for an overall judgment of “important harms” but noted that the risk of important harm is likely to be minimal in those who reduce the csDMARD dose rather than discontinue csDMARDs completely.

Dose reduction

There was MODERATE certainty evidence that dose reduction of csDMARDs probably has little to no effect on the number of people maintaining remission, on disease activity or physical function.

There was LOW certainty evidence that dose reduction of csDMARDs may increase the number of people experiencing a flare, and may have little to no effect on the number of people with progression of joint damage, on the number of people experiencing a serious adverse event or withdrawing due to adverse events.

Discontinuation

There was HIGH certainty evidence that discontinuation of csDMARDs slightly reduces the number of people maintaining remission, and slightly worsens disease activity and physical function.

There was MODERATE certainty evidence that discontinuation of csDMARDs probably increases the number of people experiencing a flare, probably slightly increases the number of people with progression of joint damage, and probably has little or no effect on the number of serious adverse events or the number of withdrawals due to adverse events.

A systematic review of patient preferences regarding DMARD treatment found that, in general, people with RA consider treatment benefits (particularly improvement in symptoms and function) to be more important than both serious and non-serious adverse events (AEs) ⁸⁴. However some studies found evidence of important risk aversion (particularly regarding serious AEs), which suggests that there is considerable variation in preferences between individuals. A scoping review of patient preferences for treatment modification in people with RA also identified variability in preferences between individuals ¹⁸³. While a common theme was a desire to return to a "normal life", the effect of this desire on treatment choices was dependent on the impacts of the disease and its treatments in the individual's unique life circumstances.

A qualitative study of patients with well-controlled RA indicated a general preference for reduction of medication burden, particularly in regard to concerns about the cumulative toxicity of DMARDs, but this was tempered by concern about the impact of a possible disease flare ¹.

The variation between individuals in their assessment of competing risks between disease and treatment, suggests that an individualised shared decision-making framework is of particular importance when considering treatment modification, including dose reduction or discontinuation of DMARDs.

The panel unanimously agreed that there is likely to be substantial variability in preferences between individuals based on their unique circumstances, goals, disease course and impact, comorbidity, and experience with previous and current medications.

In contrast to b/tsDMARDs, csDMARDs are relatively inexpensive medications. Therefore, the direct cost impact of dose reduction or discontinuation at both the individual and system level is likely to be relatively minor. The importance of any out-of-pocket cost difference may vary between individual patients. There is evidence that out-of-pocket healthcare costs may act as a barrier to accessing treatment for some people - as many as 25% of Australians living with arthritis (of any type) report skipping healthcare treatment due to cost ²⁴³. The proportion of Australians who delayed or did not fill a medication prescription due to cost increased to 7.6% in 2022-23, from 5.6% in 2021-22 ²⁴⁷.

Most data on the cost-effectiveness of DMARD tapering strategies in people with RA are focused on b/tsDMARDs. A cost-effectiveness analysis of the TARA trial, which compared csDMARD tapering with TNF inhibitor tapering in participants with RA who had achieved low disease activity with combination therapy, found that both strategies were cost-effective ²⁴⁵.

While there are no data that specifically estimate the effect of csDMARD dose reduction or discontinuation on the environment, it is plausible that reduced prescription drug use may lower any negative impact on the environment attributable to manufacturing, packaging and transport. The magnitude of such an impact (if present) or other indirect effects is currently unknown.

The panel noted that at the time of this panel meeting, the relative cost of living in Australia is very high, and therefore even relatively small changes in out-of-pocket costs may have a large impact for some people. It was also noted that the cost implications of csDMARD dose reduction may be unpredictable and any putative cost savings attributable to lower medication use may potentially be offset by increased direct healthcare costs related to more frequent disease monitoring or management of flares, and indirect costs (e.g. an impact on work status) in the event of an increase in disease activity.

Many factors influence individual patient's health opportunities and outcomes. Such factors (based on the PROGRESS-Plus characteristics) might include poor health literacy; residence in rural, remote or relatively under-serviced locations; primary language other than English; low educational attainment; the presence of disability; or adverse socioeconomic or social circumstances. Some of these may affect csDMARD dose reduction, including an impact on access to urgent specialist care or advice in the event of a flare in disease activity following DMARD dose adjustment.

The panel noted that any adjustment of treatment regimen, and the potential impact on disease activity, may present a number of challenges for individuals living with RA and that this may affect some individuals disproportionately.

The panel also highlighted that the current severe shortage of rheumatologists in Australia may increase the risk that access to urgent specialist care may be limited, and that this risk may be greater for disadvantaged groups or those living in rural or remote areas.

The acceptability of csDMARD dose reduction or discontinuation is likely to vary between patients. In addition, the perspective of people living with RA on the acceptability of DMARD tapering differs from the perspective of clinicians ²⁴⁶²⁴⁴, highlighting the importance of shared decision-making tailored to the individual patient's preferences and context.

The panel considered that while csDMARD dose reduction was likely to be feasible for many individuals, there was the potential for barriers to implementation in some settings, particularly where access to care and support may be limited during the period of dose modification.

Dose reduction

Compared to continuation of treatment at the standard dose, dose reduction of methotrexate (MTX) alone, or sequential reduction of MTX and concomitant conventional synthetic DMARDs (csDMARDs) or biological DMARDs (bDMARDs), in participants with psoriatic arthritis in remission or with low disease activity for at least 6 months:

• may increase the number of people experiencing a flare
• may increase mean disease activity slightly
• may be comparable in terms of the number of people maintaining remission, the number of people maintaining low/minimal disease activity and the number of people experiencing an adverse event
• We are uncertain whether dose reduction of csDMARDs results in fewer withdrawals due to adverse events or fewer serious adverse events, as no events were reported and the certainty of the evidence is very low.

No studies were found that measured or reported the number of people with radiographic progression of joint damage. Function was measured but not reported by one study.

The panel noted that the single randomised controlled trial (RCT) that provided data on dose reduction ¹⁷³ included 17 patients, of whom only ten were eligible for tapering of csDMARDs, and that therefore any estimates of the effect of csDMARD dose reduction based on this trial alone were unlikely to be reliable.

Discontinuation

Compared to continuation of oral MTX plus tofacitinib, abrupt discontinuation of oral MTX and continuation of tofacitinib in participants with psoriatic arthritis:

• may be comparable in terms of the number of people experiencing a flare, the number of people maintaining low/minimal disease activity, the number of withdrawals due to adverse events, the number of people experiencing a serious adverse event or any adverse events, mean disease activity and function

No studies were found that measured or reported the number of people maintaining remission or progression of joint damage.

Only one RCT was found that evaluated the effect of csDMARD dose reduction or discontinuation in people who had achieved good control of cutaneous psoriasis ²⁸⁸. This trial compared three strategies for dose reduction or discontinuation of treatment in people who achieved a PASI75 response with MTX (sudden discontinuation, halving of the MTX dose, doubling of the MTX dosing interval), however it did not include a comparator arm in which MTX was continued at the same dose. The two dose reduction strategies were associated with a lower rate of skin flare compared with sudden discontinuation.

Dose reduction

There was LOW certainty evidence that dose reduction of csDMARDs may increase the number of people experiencing a flare and increase disease activity but may have little or no effect on the number of people maintaining remission or low/minimal disease activity or on the number of people experiencing a serious adverse event.

There was VERY LOW certainty evidence for the effects of dose reduction of csDMARDs on serious adverse events or withdrawals due to adverse events as no events were reported for these.

No trials of csDMARD dose reduction measured or reported radiographic progression, and no trials reported function.

Upon review of the evidence from the one RCT that addressed this question, the panel agreed that they had VERY LOW confidence in the effect estimates for all outcomes.

Discontinuation

There was LOW certainty evidence that discontinuation of csDMARDs may have little to no effect on the number of people experiencing a flare, the number of people maintaining low/minimal disease activity, the number of withdrawals due to adverse events, the number of people experiencing a serious adverse event or any adverse events, mean disease activity and function.

No trials of csDMARD discontinuation measured or reported the number of people maintaining remission or progression of joint damage.

While there is considerable variability between individuals in the clinical manifestations and impacts of psoriatic arthritis, musculoskeletal symptoms (particularly pain) have been reported to be the most burdensome to people living with psoriatic arthritis ²⁸⁵. Among Australian patients with inflammatory arthritis (including psoriatic arthritis), the most important attributes of treatments include both clinical efficacy and the risk of mild-to-moderate adverse effects ²⁸⁶. The relative importance of these attributes, and preferences for treatments, may vary between individuals.

A study performed among British people with psoriatic arthritis investigated the relative importance of characteristics associated with both the disease (including function, participation, costs, pain and fatigue) and treatment (including gastrointestinal distress, headache, infections, hair loss, blood testing, and impact on pregnancy) ²⁸⁷. A discrete choice experiment was used to estimate the trade-off between the benefits of medication and risk of relapse for patients in a low disease activity state who might be considering tapering therapy. The majority of participants were taking methotrexate either alone or in combination with other csDMARDs or a TNF inhibitor. Participants attached the highest importance to eliminating severe nausea and were willing to accept an increase in the risk of relapse of more than 30% in exchange for improvements in this adverse effect.

The variation between individuals in their assessment of competing risks between disease and treatment, suggests that an individualised shared decision-making framework is of particular importance when considering treatment modification, including dose reduction or discontinuation of DMARDs.

The panel noted the importance of the full range of clinical manifestations of psoriatic arthritis, particularly cutaneous psoriasis, and that the variation between individuals (and within individuals over time) in the clinical features and their relative importance requires a thoughtful shared approach to treatment adjustment that takes all of these factors into account.

The panel unanimously agreed that there is likely to be substantial variability in preferences between individuals based on their unique circumstances, goals, disease course and impact, comorbidity, and experience with previous and current medications.

We did not identify any data regarding the cost-effectiveness of dose reduction or discontinuation of csDMARDs in people with psoriatic arthritis. In contrast to b/tsDMARDs, csDMARDs are relatively inexpensive medications. Therefore, the direct cost impact of dose reduction or discontinuation at both the individual and system level is likely to be relatively minor. The importance of any out-of-pocket cost difference may vary between individual patients. There is evidence that out-of-pocket healthcare costs may act as a barrier to accessing treatment for some people - as many as 25% of Australians living with arthritis (of any type) report skipping healthcare treatment due to cost ²⁴³. The proportion of Australians who delayed or did not fill a medication prescription due to cost increased to 7.6% in 2022-23, from 5.6% in 2021-22 ²⁴⁷.

While there are no data that specifically estimate the effect of csDMARD dose reduction or discontinuation on the environment, it is plausible that reduced prescription drug use may lower any negative impact on the environment attributable to manufacturing, packaging and transport. The magnitude of such an impact (if present) or other indirect effects is currently unknown.

While some panelists considered the resource implications of this recommendation to be relatively minor, the panel noted that at the time of this panel meeting, the relative cost of living in Australia is very high, and therefore even relatively small changes in out-of-pocket costs may have a large impact for some people. It was also noted that the cost implications of csDMARD dose reduction may be unpredictable and any putative cost savings attributable to lower medication use may potentially be offset by increased direct healthcare costs related to more frequent disease monitoring or management of flares, and indirect costs (e.g. an impact on work status) in the event of an increase in disease activity.

Many factors influence individual patient's health opportunities and outcomes. Such factors (based on the PROGRESS-Plus characteristics) might include poor health literacy; residence in rural, remote or relatively under-serviced locations; primary language other than English; low educational attainment; the presence of disability; or adverse socioeconomic or social circumstances. Some of these may affect csDMARD dose reduction, including an impact on access to urgent specialist care or advice in the event of a flare in disease activity following DMARD dose adjustment.

The current severe shortage of rheumatologists in Australia may increase the risk that access to urgent specialist care may be limited, and that this risk may be greater for disadvantaged groups or those living in rural or remote areas.

The acceptability of csDMARD dose reduction or discontinuation is likely to vary between patients. Although there is scant direct evidence in psoriatic arthritis, among people living with rheumatoid arthritis there is evidence that the perspective of patients regarding the acceptability of DMARD tapering differs from the perspective of clinicians ²⁴⁴²⁴⁶, highlighting the importance of shared decision-making tailored to the individual patient's preferences and context.

csDMARD tapering is likely to be broadly acceptable to other stakeholders.

While csDMARD dose reduction is likely to be feasible for many individuals, there is the potential for barriers to implementation in some settings, particularly where access to care and support may be limited during the period of dose modification.

In adults with rheumatoid arthritis and low disease activity, reducing the dose of a bDMARD or tsDMARD probably has little effect on control of disease activity, including mean DAS and the proportion who remain in remission (or a low disease activity state). There may be a small negative effect on radiographic progression and function. Based on clinical trials there appears to be little difference in safety or tolerability but adverse event rates in trials are low and therefore data are limited.

Discontinuing a bDMARD or tsDMARD without prior dose reduction probably reduces the proportion of participants with persistent remission, probably slightly worsens disease activity, probably increases the proportion of people who experience a flare, may slightly increase the proportion of people with minimal radiographic progression, may lead to a slight deterioration in function and may slightly worsen quality of life. Due to the small number of events reported, we are uncertain whether discontinuation results in fewer serious adverse events or whether adverse events differ between groups.

There are some limitations to the certainty of the evidence for dose reduction. First, most of the data are for TNF inhibitors (particularly adalimumab and etanercept), and therefore may not be directly applicable to other b/tsDMARDs. Second, for some outcomes there may not be sufficient precision in the data to demonstrate that dose reduction is an equivalent strategy to b/tsDMARD continuation. Third, the relatively short follow-up period in trials limits the number of important adverse events detected and may reduce the likelihood of detecting radiographic progression, thereby reducing our confidence in the estimates of these outcomes.

There is less certainty in the evidence regarding abrupt b/tsDMARD cessation. Data on drug abrupt cessation were available for TNF inhibitors only, and therefore may not be applicable to other b/tsDMARDs. Most of the evidence for cessation was considered to be low certainty, although there was moderate certainty for mean disease activity score. We have very low confidence in the evidence regarding adverse events due to low event rates.

We have no systematically collected information regarding patients’ preferences and values. It is likely that individual patient preferences for reduced exposure to medication versus the risk of flare will vary.

A qualitative study of patients with well-controlled RA ¹ indicated a general preference for reduction of medication burden, particularly in regard to concerns about the cumulative toxicity of DMARDs, but this was tempered by concern about the impact of a possible disease flare. Tolerance for the risk of flare was mediated by social circumstances, including the impact of a disease flare on employment and burden to family. There was variation between individuals in the assessment of competing risks between disease and treatment, suggesting that an individualised shared decision-making framework is of particular importance in this context.

b/tsDMARDs are generally considered to be expensive therapies and therefore strategies to reduce the total cumulative dose are likely to reduce costs for payers (governments and insurers) and may reduce out-of-pocket costs for patients. In the Australian context, the direct cost impact for individual patients is relatively small but this may differ in other contexts. The importance of direct out-of-pocket cost implications of dose reduction versus continuation is likely to vary between individual patients. There are no publicly-available data regarding the actual comparative costs of different bDMARDs to the payer (i.e. the Federal Government) in Australia.

Most trials that have assessed b/tsDMARD dose reduction have not assessed costs or cost-effectiveness. Compared with anti-TNF continuation, two trials found that anti-TNF activity-guided dose reduction significantly reduced costs and resulted in decremental cost-effectiveness ratios but the estimates were very broad ³.

The panel considered that many factors influence individual patient’s health opportunities and outcomes. Such factors (based on the PROGRESS-Plus characteristics) might include poor health literacy; residence in rural, remote or relatively under-serviced locations; primary language other than English; low educational attainment; the presence of disability; or adverse socioeconomic or social circumstances. Some of these may affect b/tsDMARD dose reduction, including an impact on access to urgent specialist care or advice in the event of a flare in disease activity following DMARD dose adjustment.

From a recent survey of ARAD participants (unpublished data):

The medication you have taken is known as a biologic or targeted DMARD: Evidence shows that some people with well-controlled disease can safely reduce the dose of their biologic or targeted DMARD or how frequently they take it, without any worsening of their condition. If your treating specialist suggested it, would you be willing to try taking your biologic or targeted DMARD less frequently, or at a reduced dose?

Responses as follows: Yes: 216 (53%) No: 66 (16%) Maybe: 127 (31%)

Implementation of the recommendation may be influenced in some cases by local prescribing rules or reimbursement conditions, and by the accessibility of care in the case of a disease flare. Current Australian prescribing and reimbursement rules for b/tsDMARDs are complex. For example, some reimbursement rules are predicated on stable disease control with full-dose medication. In this situation, some patients or prescribers may be concerned about the risk that, in the event of a disease flare with dose reduction, the resumption of a higher dose of the current b/tsDMARD may be prohibited.

In adults with axial spondyloarthritis and sustained low disease activity, reducing the dose of bDMARD probably results in fewer patients with persistent remission and may slightly increase the proportion of patients with a disease flare, but probably has little or no effect on disease activity as measured by BASDAI and may have little or no effect on the number of patients in ASAS partial remission. Dose reduction probably has little or no effect on function.

Abrupt cessation of bDMARD may result in fewer people with persistent remission or ASAS partial remission, probably worsens disease activity, may increase the number of people who experience a flare, and probably worsens function.

The effect of either dose reduction or discontinuation on the safety or tolerability of bDMARDs is difficult to estimate due to low rates of adverse events in the included clinical trials.

There are some limitations to our confidence in the evidence. There was moderate certainty for most of the evidence for the effect of dose reduction on disease activity, although certainty in the evidence was low for the number of patients with a disease flare and the proportion with ASAS partial remission. Confidence in the evidence for the effect of bDMARD discontinuation was moderate for disease activity (measured by BASDAI) and function, but was otherwise low. We have very low confidence in the evidence regarding adverse events following either dose reduction or discontinuation due to low event rates.

At the time of the initial panel meeting, all of the data were from trials of TNF inhibitors. Subsequently, a trial of discontinuation of an IL-17 inhibitor (ixekizumab) was published but this did not change the overall estimate of benefits and harms and no change was made to the recommendation. It remains unclear to what extent the current data may be extrapolated to dose reduction for bDMARDs other than TNF inhibitors or discontinuation for bDMARDs that do not target either TNF or IL-17.

We combined data from trials that included patients with either ankylosing spondylitis, non-radiographic axial spondyloarthritis or both, on the assumption that these represent aspects of the same disease spectrum, although we recognise that not all patients with non-radiographic axial spondyloarthritis will progress to radiographic disease (i.e., ankylosing spondylitis).

There were some differences between the trials in the definition of low disease activity or remission, the amount of time participants were in a low disease activity state before dose reduction, and the method used for dose reduction (such as increased dosing interval versus a lower dose of drug at the same interval). There are no data regarding the effect of further dose adjustments based on the response to an initial dose reduction (i.e., ‘stepwise’ dose reduction). All patients in the trials of bDMARD discontinuation stopped the drug without a prior dose taper.

We have no systematically collected information regarding patients’ preferences and values. It is likely that individual patient preferences for reduced exposure to medication versus the risk of flare will vary.

There are no data on the effect of dose reduction on important non-musculoskeletal features of spondyloarthritis (including inflammatory disorders of the eye, intestinal tract and skin). The presence of these clinical features may also influence individual patients' preferences regarding DMARD dose reduction.

A qualitative study ⁶ of British patients with inflammatory arthritis - including ankylosing spondylitis - found that patients who were using bDMARDs were interested in reducing their dose when the disease was under control in the hope that this would reduce inconvenience and potential toxicity while maintaining disease control, and reduce overall cost to society. However this was balanced by concerns that dose reduction might result in a disease flare and that resumption of the full dose, if required, might be limited due to the high cost of bDMARDs or might not achieve the previous level of disease control. Shared decision-making, including control over flexible dosing, was identified as an important theme.

Most trials that have assessed bDMARD dose reduction have not assessed costs or cost-effectiveness. bDMARDs are generally considered to be expensive therapies and therefore strategies to reduce the total cumulative dose are likely to reduce costs for payers (governments and insurers) and may reduce out-of-pocket costs for patients. In the Australian context, the direct cost impact for individual patients is relatively small but this may differ in other contexts. The importance of direct out-of-pocket cost implications of dose reduction versus continuation is likely to vary between individual patients. There are no publicly-available data regarding the actual comparative costs of different bDMARDs to the payer (i.e., the Federal Government) in Australia.

The panel considered that many factors influence individual patients' health opportunities and outcomes. Such factors (based on the PROGRESS-Plus characteristics) might include poor health literacy; residence in rural, remote or relatively under-serviced locations; primary language other than English; low educational attainment; the presence of disability; or adverse socioeconomic or social circumstances. Some of these may affect bDMARD dose reduction, including an impact on access to urgent specialist care or advice in the event of a flare in disease activity following DMARD dose adjustment.

Evidence for acceptability of the intervention to consumers comes from a recent survey of ARAD participants (unpublished data):

The medication you have taken is known as a biologic or targeted DMARD: Evidence shows that some people with well-controlled disease can safely reduce the dose of their biologic or targeted DMARD or how frequently they take it, without any worsening of their condition. If your treating specialist suggested it, would you be willing to try taking your biologic or targeted DMARD less frequently, or at a reduced dose?

Responses as follows: Yes: 95 (50%) No: 45 (23%) Maybe: 51 (27%)

There is little evidence regarding the acceptability of bDMARD dose reduction or discontinuation to clinicians, policymakers or other stakeholders, although the panel felt that consideration of reduction of medication burden in a shared decision-making framework would likely be acceptable to all stakeholders.

Implementation of the recommendation may be influenced in some cases by local prescribing rules or reimbursement conditions, and by the accessibility of care in the case of a disease flare. Current Australian prescribing and reimbursement rules for bDMARDs are complex. For example, some reimbursement rules are predicated on stable disease control with full-dose medication. In this situation, some patients or prescribers may be concerned about the risk that, in the event of a disease flare with dose reduction, the resumption of a higher dose of the current bDMARD may be prohibited.

There are no data on the effect of dose reduction of b/tsDMARDs in patients with psoriatic arthritis.

In patients with psoriatic arthritis in sustained minimal disease activity, discontinuation of b/ts DMARDs may reduce the proportion of participants with persistent remission, may increase the proportion of people who experience a flare, may result in little or no difference in function, and may or may not reduce the number of participants with adverse events.

We are uncertain about the effect of discontinuation of b/tsDMARDs on disease activity. We are also unable to estimate the risk of serious adverse events or withdrawals due to adverse events as there were too few events. Data on the effect of discontinuation of b/tsDMARDs are limited to patients using TNF inhibitors or the IL-17 inhibitor ixekizumab.

There are no data on the effect of dose reduction of b/tsDMARDs in patients with psoriatic arthritis.

We are uncertain about the effect of discontinuation of b/tsDMARDs in patients with psoriatic arthritis in sustained minimal disease activity. Data are available from two trials. One was a pilot RCT (of 17 patients on various treatment regimens including 10 participants on TNFi alone or in combination with csDMARDs). The other was an RCT of discontinuation versus continuation of the IL-17 inhibitor ixekizumab.

While the certainty of the evidence is low for some outcomes (proportion of participants who remain in remission, proportion who flare, function, and proportion who experience adverse events), it remains very low for several other important outcomes (disease activity and the proportion of participants with serious adverse events or withdrawals due to adverse events).

There are no trials that measure radiographic progression or switching to another b/tsDMARD. There are no trials of discontinuation of b/tsDMARDs other than ixekizumab (an IL-17 inhibitor) or TNF inhibitors.

Overall, our confidence in the estimates of benefits and harms of discontinuation of treatment in people with psoriatic arthritis who have achieved a low disease activity state with any b/tsDMARD remains very low.

We have no systematically collected information regarding patients’ preferences and values. It is likely that individual patient preferences for reduced exposure to medication versus the risk of flare will vary. The presence of non-musculoskeletal features of psoriatic arthritis (particularly psoriasis) may also influence individual patients' preferences regarding DMARD dose reduction.

A qualitative study ⁶ of British patients with inflammatory arthritis - including psoriatic arthritis - found that patients who were using bDMARDs were interested in reducing their dose when the disease was under control in the hope that this would reduce inconvenience and potential toxicity while maintaining disease control, and reduce overall cost to society. However this was balanced by concerns that dose reduction might result in a disease flare and that resumption of the full dose, if required, might be limited due to the high cost of bDMARDs or might not achieve the previous level of disease control. Shared decision-making, including control over flexible dosing, was identified as an important theme.

There are no data on the cost-effectiveness of b/tsDMARD dose reduction for psoriatic arthritis. b/tsDMARDs are generally considered to be expensive therapies and therefore strategies to reduce the total cumulative dose are likely to reduce costs for payers (governments and insurers) and may reduce out-of-pocket costs for patients. In the Australian context, the direct cost impact for individual patients is relatively small but this may differ in other contexts. The importance of direct out-of-pocket cost implications of dose reduction versus continuation is likely to vary between individual patients. There are no publicly-available data regarding the actual comparative costs of different b/tsDMARDs to the payer (i.e., the Federal Government) in Australia.

The panel considered that many factors influence individual patients' health opportunities and outcomes. Such factors (based on the PROGRESS-Plus characteristics) might include poor health literacy; residence in rural, remote or relatively under-serviced locations; primary language other than English; low educational attainment; the presence of disability; or adverse socioeconomic or social circumstances. Some of these may affect bDMARD dose reduction, including an impact on access to urgent specialist care or advice in the event of a flare in disease activity following DMARD dose adjustment.

Evidence for acceptability of the intervention to consumers comes from a recent survey of ARAD participants (unpublished data):

"The medication you have taken is known as a biologic or targeted DMARD: Evidence shows that some people with well-controlled disease can safely reduce the dose of their biologic or targeted DMARD or how frequently they take it, without any worsening of their condition. If your treating specialist suggested it, would you be willing to try taking your biologic or targeted DMARD less frequently, or at a reduced dose?"

Responses as follows: Yes: 74 (46%) No: 31 (19%) Maybe: 55 (34%)

There is little evidence regarding the acceptability of b/tsDMARD dose reduction or discontinuation to clinicians, policymakers or other stakeholders, although the panel felt that consideration of reduction of medication burden in a shared decision-making framework would likely be acceptable to all stakeholders.

Implementation of the recommendation may be influenced in some cases by local prescribing rules or reimbursement conditions, and by the accessibility of care in the case of a disease flare. Current Australian prescribing and reimbursement rules for b/tsDMARDs are complex. For example, some reimbursement rules are predicated on stable disease control with full-dose medication. In this situation, some patients or prescribers may be concerned about the risk that, in the event of a disease flare with dose reduction, the resumption of a higher dose of the current bDMARD may be prohibited.

Dose reduction

Compared with dose reduction of csDMARDs (methotrexate, sulfasalazine, hydroxychloroquine, leflunomide), dose reduction of b/tsDMARDs (etanercept, adalimumab, certolizumab or golimumab) in participants with RA in remission or with low disease activity at 12 months follow-up:

• May be comparable in terms of the number of people maintaining remission, the number of people experiencing flare, disease activity, function and quality of life, and in the number of people experiencing an adverse event.
• The effect on the number of people with radiographic joint progression or on the number of people experiencing serious adverse events is uncertain. The number of people who withdrew due to adverse events was not reported.

Discontinuation

Compared with discontinuation of csDMARDs (methotrexate), discontinuation of b/tsDMARDs (etanercept, abatacept, tofacitinib) in participants with RA in remission or with low disease activity at 12 months follow-up:

• May reduce the number of people maintaining remission
• May increase the number of people experiencing a flare
• May be comparable in terms of disease activity, function, quality of life, and the number of people experiencing an adverse event
• The effect on the number of people experiencing a serious adverse events or the number of withdrawals due to adverse events is uncertain

Radiographic progression was not measured.

The panel considered that the balance of benefits and harms differed for DMARD dose reduction strategies versus discontinuation strategies. While a discontinuation strategy may favour stopping csDMARDs rather than b/tsDMARDs, the panel noted that this guideline's existing recommendations regarding dose modification of both csDMARDs and b/tsDMARDs conditionally favour dose reduction but recommend against abrupt discontinuation of DMARDs. Therefore the panel elected to focus on the data regarding dose reduction and unanimously agreed on a rating of "small net benefit, or little difference between alternatives".

Dose reduction

Compared with dose reduction of csDMARDs, there was LOW certainty evidence that dose reduction of b/tsDMARDs may have little or no effect on the number of people maintaining remission, the number of people experiencing a flare, disease activity, function and quality of life.

Compared with dose reduction of csDMARDs, there was VERY LOW certainty evidence for the effects of b/ts dose reduction on radiographic progression and serious adverse events.

The number of people withdrawing due to adverse events was not reported.

Discontinuation

Compared to discontinuation of csDMARDs there was LOW certainty evidence that discontinuation of b/tsDMARDs may reduce the number of people maintaining remission, may increase the number of people experiencing a flare and may be comparable in terms of disease activity, function, quality of life, and the number of people experiencing an adverse event.

Compared to discontinuation of csDMARDs there was VERY LOW certainty evidence for the effects of discontinuation of b/tsDMARDs number of people experiencing serious adverse events or the number of withdrawals due to adverse events.

Radiographic progression was not reported.

The panel considered that the evidence regarding dose reduction strategies, derived from a single trial that was relatively underpowered, was low certainty at best.

While there are few data regarding patient preferences in the specific context of choosing between tapering csDMARDs and b/tsDMARDs, existing evidence regarding DMARD tapering in RA more broadly suggests that variation between individuals is likely.

A systematic review of patient preferences regarding DMARD treatment found that, in general, people with RA consider treatment benefits (particularly improvement in symptoms and function) to be more important than both serious and non-serious adverse events (AEs) ³⁵. However some studies found evidence of important risk aversion (particularly regarding serious AEs), which suggests that there is considerable variation in preferences between individuals.

A qualitative study of patients with well-controlled RA indicated a general preference for reduction of medication burden, particularly in regard to concerns about the cumulative toxicity of DMARDs, but this was tempered by concern about the impact of a possible disease flare ¹.

A scoping review of patient preferences for treatment modification (including DMARD tapering) in people with RA also identified variability in preferences between individuals ¹⁸³. The individual's satisfaction with the current treatment or health state, past treatment experiences, and perception of the consequences of change in the context of their unique life circumstances, were found to be important determinants of preferences regarding treatment change.

The variation between individuals in their assessment of competing risks between disease and treatment, suggests that an individualised shared decision-making framework is of particular importance when considering treatment modification, including dose reduction or discontinuation in those taking DMARDs in combination.

b/tsDMARDs are generally considered to be expensive therapies and therefore strategies that preference reduction in the total cumulative dose of these medications are likely to reduce costs for payers (governments and insurers) and may reduce out-of-pocket costs for patients. There are no publicly-available data regarding the actual comparative costs of different bDMARDs to the payer (i.e. the Federal Government) in Australia. In contrast, csDMARDs are relatively inexpensive medications. Therefore, dose reduction or discontinuation of csDMARDs in preference to b/tsDMARDs is likely to have less impact on direct cost reduction at the system level.

In the Australian context, the direct cost impact for individual patients is relatively small but this may differ in other contexts. The importance of any out-of-pocket cost difference may vary between individual patients. There is evidence that out-of-pocket healthcare costs may act as a barrier to accessing treatment for some people - as many as 25% of Australians living with arthritis (of any type) report skipping healthcare treatment due to cost ²⁴³. The proportion of Australians who delayed or did not fill a medication prescription due to cost increased to 7.6% in 2022-23, from 5.6% in 2021-22 ²⁴⁷.

A cost-effectiveness analysis of two-year data from the TARA trial ²⁴⁵, which compared csDMARD tapering with TNF inhibitor tapering in participants with RA who had achieved low disease activity with combination therapy, found that both strategies were cost-effective (in a Dutch healthcare setting). Total costs were similar regardless of which DMARD was tapered first; medication costs were lower in those who tapered TNFi first, but indirect costs were higher due to increased productivity loss.

While there are no data that specifically estimate the relative impact of csDMARD versus b/tsDMARD dose reduction or discontinuation on the environment, it is plausible that there may be differences attributable to differential manufacturing, packaging and transport. The magnitude of such an impact (if present) or other indirect effects is currently unknown.

The panel discussed the cost-effectiveness data, and the potential impact of disease flares at the individual level, including evidence from the included trials that flares may last for three months or longer. The panel considered the difficulty in weighing the relatively similar cost-impact of the two tapering strategies at the group or societal level against the potential impact of small differences in flare rates at the level of the individual experiencing a flare. The panel voted for a rating of “important issues”, noting that despite the difference in cost between csDMARDs and b/tsDMARDs, the potentially substantial impact on individual outcomes if there is a difference in the frequency or depth of flares following dose modification is likely to be an important factor in decision-making in the clinic.

Many factors influence individual patient's health opportunities and outcomes. Such factors (based on the PROGRESS-Plus characteristics) might include poor health literacy; residence in rural, remote or relatively under-serviced locations; primary language other than English; low educational attainment; the presence of disability; or adverse socioeconomic or social circumstances. Some of these may affect preferences for DMARD regimen modification, including the potential impact of access to medications or to urgent specialist care or advice in the event of a flare in disease activity following DMARD dose adjustment.

The panel considered that there were likely to be relatively few differential health equity consequences of the different dose reduction strategies, however there remained potential for such differences to exist.

The acceptability of csDMARD dose reduction or discontinuation is likely to vary between individual patients and other stakeholders. A qualitative study of patients and rheumatologists regarding tapering DMARDs in RA found that both groups were concerned about the risk of disease flare with dose reduction, particularly the risk of failure to recapture the previous low disease activity state, but that there was considerable variation in the degree of concern ²⁴⁶. Rheumatologists often favoured tapering of csDMARDs based on a perceived beneficial trade-off between risk of flare and potential reduction in drug-related harms.

Modification of the DMARD regimen is likely to be broadly feasible but may be influenced in some cases by local prescribing rules or reimbursement conditions, and by the accessibility of care in the case of a disease flare. Current Australian prescribing and reimbursement rules for b/tsDMARDs are complex. For example, some b/tsDMARD reimbursement rules are predicated on stable disease control with full-dose medication. In this situation, some patients or prescribers may be concerned about the risk that, in the event of a disease flare with dose reduction, the resumption of a higher dose of the current b/tsDMARD may be prohibited or delayed.

The panel also noted that some b/tsDMARDs currently require co-prescription of methotrexate under Australian prescribing restrictions, and that this may also impact the feasibility of some strategies for dose reduction or discontinuation.

For the purpose of this recommendation we considered inflammatory arthritis to include rheumatoid arthritis (RA), psoriatic arthritis (PsA) and spondyloarthritis (SpA). We identified three RCTs, all comparing perioperative discontinuation versus perioperative continuation of csDMARDs (methotrexate and leflunomide) in adults with RA undergoing elective orthopaedic surgery. The three RCTs provided outcome data on the occurrence of flare, post-operative infections, prosthetic joint infection, adverse events (including but not limited to wound dehiscence) and serious adverse events (including but not limited to need for revision surgery). None of the studies reported on mean disease activity, length of hospital stay, post-operative opioid use, function and quality of life.

Based on the RCT data, perioperative discontinuation of DMARDs, compared to perioperative continuation of DMARDs:

• may increase the risk of flare
• may have little or no effect on the number of people with post-operative infections, the number of people reporting adverse events and the number of people reporting serious adverse events

To supplement the evidence base, we included data from observational studies comparing perioperative discontinuation versus continuation of any DMARDs in adults with inflammatory diseases undergoing any type of elective surgery. There were eight eligible observational studies. Most of the participants had undergone elective orthopaedic surgery or various joint arthroplasties, with one study specifically looking at participants who had undergone spinal surgery. Four studies specifically investigated perioperative discontinuation versus continuation of bDMARDs, including TNF inhibitors ^113115117 and abatacept ¹¹⁸, two studies investigated MTX ¹¹¹¹¹⁶ and two studies assessed combination DMARD therapy ¹¹² or all DMARDs ¹¹⁴.

Evidence from observational cohort studies largely concurs with the findings from RCTs, indicating there may be an increased risk of flare if DMARDs are discontinued, with no apparent reduction in risk of postoperative infection. While we could not estimate the risk of prosthetic joint infections from trials, observational data suggests there may be no reduction in risk with discontinuation of DMARDs perioperatively.

The panel voted that there was “small net benefit or little difference” for all DMARDs combined but noted that the existing RCT evidence applies only to csDMARDs (methotrexate and leflunomide) and that there is an absence of evidence regarding tsDMARDs in the perioperative period. The panel also noted that flares may vary in severity, duration and impact, and that there may be various factors that impact on the risk of perioperative flare, including the preoperative disease activity, history of flares, use of combination DMARDs, use of glucocorticoids, and the duration of DMARD discontinuation. There are few data to permit accurate estimation of the absolute risk of a perioperative disease flare in the individual patient. The panel also noted that disease flares may be particularly important if they impair successful recovery or rehabilitation from surgery, or if they require additional use of glucocorticoids.

There was low certainty evidence for the following outcomes:

• risk of flare
• number of people with post-operative infections
• number of people reporting adverse events
• number of people reporting serious adverse events

No studies were found that looked at mean disease activity, length of hospital stay, post-operative opioid use, function and quality of life.

Evidence from the three RCTs is limited to patients with RA, discontinuation of csDMARDs (methotrexate and leflunomide) and elective orthopaedic surgery. The two RCTs that investigated methotrexate discontinuation included participants who were using slightly lower doses of methotrexate (average 10mg/week) than are used in most patients in the current era.

We did not identify any studies that directly investigated values and preferences of patients with RA on DMARDs undergoing surgical procedures. It is likely that the primary trade-off is between a potential increase in the risk of post-operative infection (particularly infection of a prosthesis) associated with the use of DMARDs in the perioperative period and the risk of a flare of the underlying inflammatory arthritis associated with interruption of disease-modifying therapy.

It is likely that individuals may vary in the relative importance afforded to these outcomes. Such variation is likely to have multiple contributors including the individual disease status and trajectory, treatment goals, history of disease flares and infections, current medications, the nature of the surgery, demographic factors (such as age) and individual risk tolerance. The American College of Rheumatology (ACR) panel that developed the current (2017) ACR guideline on the use of DMARDs in patients undergoing elective arthroplasty reported that the consumer panel considered the risk of infection to be "much more important" than the risk of flare, and estimated the relative importance of infection compared to flares as "greater than 10:1" ¹⁰⁹. This was noted to be a strong driver of that guideline's cautious approach to the use of b/tsDMARDs in the perioperative period. In contrast, a study performed in the UK that included patients, rheumatologists and orthopaedic surgeons in focus groups reported that patients prioritised avoidance of disease flare and "would be willing to accept a risk of post-operative infection as high as 3-7% to prevent a flare of their disease" ¹¹⁹.

The panel agreed unanimously that individual patients are likely to vary in their weighting of outcomes in this setting. The consumer member of the panel noted that infection is likely to be a greater concern for many people with inflammatory arthritis, particularly infections that may have severe or irreversible consequences (including those affecting prosthetic joints), but that this would be likely to vary according to the type of surgery, the individual’s prior experience (including infections and disease flares), and the potential individual impact of a surgical site infection or other hospital-acquired infection.

While b/tsDMARDs are expensive medications, temporary interruption of therapy is unlikely to have an important impact on overall costs. Major adverse outcomes of surgery in this population, including infection or disease flare, may carry an additional resource burden to the individual patient and to the healthcare system, however there are no available cost-effectiveness data regarding the decision to continue or withhold DMARDs in the perioperative period.

We do not anticipate any environmental impact of temporary discontinuation of DMARDs in the perioperative period.

The panel considered that this recommendation is unlikely to have a major impact on health equity. Some of the risks associated with this recommendation, including infection and disease flare, may have a disproportionately high impact in populations or geographic areas with limited access to specialist care (particularly rheumatologists) in the post-operative period.

Adjustment of DMARD regimens in the perioperative period with the goal of optimising the balance between the risk of flare and the risk of perioperative complications, particularly infection, is likely to be acceptable to all stakeholders including consumers, surgeons, rheumatologists, hospital administrators, and funders. The panel considered that there may be some variation in acceptability between different stakeholders.

In general, adjustment of DMARD regimens in the perioperative period for people with RA is likely to be feasible to implement. However, some evidence suggests the existence of variation in practice in other settings in which society guideline recommendations on this topic currently exist. For example, a national survey of rheumatologists and orthopaedic surgeons in the UK found that 79% of rheumatologists follow BSR guidelines whereas only 39% of surgeons follow BSR or local guidelines, and there was frequently disagreement about which specialty was responsible for perioperative dose adjustment ¹¹⁹. It is therefore plausible that a similar variation in implementation of recommendations on this topic may also occur in an Australian context.

There may be some other barriers to implementation depending on the particular healthcare setting. For example, implementation may be complex in hospital settings in which junior medical staff may be responsible for adjusting treatment arrangements in the perioperative period, particularly in situations where there is a high turnover of staff. In addition, implementation may be more challenging for some bDMARDs with infrequent dosing schedules, particularly in settings where surgery may be commonly delayed or rescheduled, or in the case of urgent surgery.

We did not identify any randomised controlled trials that addressed this question directly.

We included data from two observational studies that assessed reduced periodic monitoring of liver enzymes, renal function and full blood count compared with routine monitoring.

Evidence from a small single centre retrospective cohort study of temporarily relaxed blood test monitoring from 3-monthly tests to 6-monthly tests for people with stable disease taking methotrexate during COVID restrictions in the UK, indicates that a small proportion of people having blood tests at 6-monthly intervals had abnormal test results (6/67, 9%), but this did not lead to any harms ³⁰⁶.

Evidence from a large long-term cohort study (published as an abstract) indicated that the incidence of highly abnormal test results over a 12-year period in a large number of people with rheumatoid arthritis taking disease modifying anti-rheumatic drugs (DMARDs) and undergoing recommended 3- to 6- monthly routine testing (the Netherlands) (4911 people) was very low, and did not differ between those taking DMARDs for which regular blood monitoring is recommended versus those using other DMARDs ³⁰⁴.

The panel noted that the evidence base to inform the optimal monitoring strategy is extremely limited, and that other factors, including historical practice conventions, mechanistic considerations, and individual risk tolerance, have contributed to current practice, which is known to vary between practitioners ²⁹⁷. For this reason, the panel unanimously agreed on a rating of “small net benefit, or little difference between alternatives”.

There is no evidence from randomised controlled trials. The evidence from observational studies regarding the effect of different safety monitoring methods is very low certainty.

We did not identify any studies that directly investigated the values and preferences of people living with inflammatory arthritis (IA) regarding safety monitoring for csDMARDs. Specifically, we did not find any data regarding the relative weight that patients afford to the potential risks of more frequent blood monitoring (such as inconvenience, cost, anxiety, opportunity cost, and the risk of subsequent additional investigations) versus the putative benefits of earlier detection and potential prevention of serious adverse effects of disease-modifying therapy.

Data from the Australian Rheumatology Association Database (ARAD; manuscript in preparation) investigating the experiences and preferences of 451 Australians using DMARDs for the treatment of rheumatoid or psoriatic arthritis indicates variation in the frequency of routine blood monitoring, although 12 weeks was the most common interval (in 30% of respondents). Most respondents were adherent to the monitoring schedule recommended by their prescriber, and reported perceiving more benefits than harms or costs associated with regular blood testing, although 10% reported opportunity costs related to employment and 45% travelled for more than 30 minutes to have their blood tests. The perceived benefits related primarily to detection of hepatotoxicity and measurement of disease activity rather than detection of myelosuppression.

In general, there is evidence that people with IA consider the risk of adverse events to be less important than treatment benefits, but there is considerable variation in preferences and risk aversion between individuals ⁸⁴. Similarly, the impact of safety monitoring requirements on patients' preferences regarding the choice of DMARD ranks lower than other factors such as route of administration, dosing frequency, efficacy, and out-of-pocket cost ²⁹⁵. In a British study of methotrexate monitoring in primary care, more than half of people with RA were undertaking blood monitoring more frequently than BSR recommendations, due to either risk aversion or inaccurate information on monitoring frequency ²⁹⁶.

There is likely to be variation between individual patients in their preferences, influenced by a variety of factors including social circumstances, concomitant medications, alcohol consumption, prior adverse events, comorbidity, and individual risk tolerance.

The panel discussed the likelihood of variation in preferences, and the multiple potential contributors to this variation. One of the consumer panellists noted that many patients are concerned about the long-term risks associated with csDMARDs, particularly methotrexate, and that, for many, participation in regular blood testing may allow a better sense of control over mitigation of this perceived risk.

We did not find any evidence regarding the cost-effectiveness of current csDMARD laboratory safety monitoring practices.

The frequency of blood monitoring is likely to have cost implications at both the societal and individual level. While the cost of routine blood monitoring for Australians taking csDMARDs is typically covered by Medicare, there may be other associated costs including travel and parking, and opportunity costs associated with the time spent attending for blood tests. There may also be indirect or additional costs, including additional testing if abnormalities are detected on routine monitoring.

The panel noted that the total cost to the healthcare system of regular blood testing in this population is likely to be substantial, but the degree to which this is offset by prevention of major harmful outcomes is unknown. Similarly, the potential cost implications of unwarranted discontinuation of effective and inexpensive medications in those with minor abnormalities on blood testing, including progression to expensive b/tsDMARDs, remains unknown.

The importance of any out-of-pocket cost difference may vary between individual patients. There is evidence that out-of-pocket healthcare costs may act as a barrier to accessing treatment for some people - as many as 25% of Australians living with arthritis (of any type) report skipping healthcare treatment due to cost ²⁴³.

While there are no data that specifically estimate the relative impact of the frequency of blood monitoring on the environment, it is plausible that there may be differences attributable to the increased volume of healthcare activity and associated factors, including travel. It is also plausible that this may be offset to some extent if serious adverse events are prevented by monitoring. The magnitude of such an impact (if present) or other indirect effects is currently unknown.

Many factors influence individual patient's health opportunities and outcomes. Such factors (based on the PROGRESS-Plus characteristics) might include poor health literacy; residence in rural, remote or relatively under-serviced locations; primary language other than English; low educational attainment; the presence of disability; or adverse socioeconomic or social circumstances. Some of these may affect preferences and capability regarding csDMARD safety monitoring.

The panel noted that monitoring regimens are often complex and place a burden on patients. The extent and consequences of this burden is likely to vary between individuals. Further, the imposition of a long-term monitoring regimen may offer a sense of reassurance and control for some, while for others it may reinforce a misperception that DMARDs are inherently toxic, particularly in those with lower health literacy. Seeking opportunities to provide additional information and education for patients is also crucial and may help to prevent some health inequity.

The acceptability of different csDMARD safety monitoring schedules is likely to vary between individual patients and among other stakeholders. In the absence of clear evidence regarding the most effective method for safety monitoring, acceptability is likely to be highly preference-dependent.

The panel discussed at some length the potential for acceptability of any recommendation that may vary from current practice norms given the paucity of data to inform best practice, and the likelihood of aversion to rare but major risks (particularly organ failure) rather than more common but less tangible competing risks. Further, the panel noted that while there is variation in monitoring practice, individual approaches to monitoring are often entrenched and may be unlikely to change in the absence of evidence to support an alternative approach.

Routine blood monitoring is an established practice for people taking csDMARDs for the treatment of inflammatory arthritis. The specific method (including frequency of testing) varies considerably between practitioners ²⁹⁷. It is plausible that this variation in practice is relatively entrenched and, in the absence of clear evidence to support a particular monitoring method, change in practice at the level of the individual may be difficult to achieve.

For the purpose of this recommendation we considered inflammatory arthritis to include rheumatoid arthritis (RA), psoriatic arthritis (PsA) and spondyloarthritis (SpA). We identified one randomised controlled trial (RCT), one non-randomised controlled study and four observational studies that addressed our primary question of whether people with inflammatory arthritis treated with glucocorticoids who are undergoing elective surgery should continue their glucocorticoids at their current dose or receive a higher (stress) dose during the perioperative period. The RCT included participants with secondary adrenal insufficiency due to the use of glucocorticoids at a dose of at least 7.5mg prednisone per day for several months for a variety of diseases, including eight participants with RA.

Based on the RCT data, we remain uncertain about the benefits and harms of additional or increased doses of glucocorticoids compared to continuation of the usual dose of glucocorticoids, because the certainty of the evidence is very low. Clinical adrenal insufficiency, post-operative infections and other adverse events were rare and only reported in one of the treatment groups of the trial, and therefore we could not estimate the risks of these events with additional doses of GCs. The RCT did not report on several outcomes of interest, including arthritis flare, prosthetic joint infection, mean disease activity or function.

Evidence from one non-randomised controlled study ²⁰⁵ and one observational study ²⁰⁸ of adults with RA undergoing elective orthopaedic surgery did not suggest that the risk of perioperative adrenal insufficiency varies according to the dose of glucocorticoids used perioperatively, although this evidence is very low certainty.

Evidence from a further three observational studies ^206207209 suggests higher perioperative glucocorticoid exposure may be associated with an increased risk of postoperative infection (reported in one study) and prosthetic joint infection (consistent across the three larger cohort studies), and possibly long term mortality (reported in one large cohort). However evidence from these studies is also very low certainty and did not directly assess the effect of glucocorticoid stress dosing versus continuation of the usual glucocorticoid dose.

The panel agreed that there was insufficient evidence to reliably estimate the relative benefits and harms of continuing the usual glucocorticoid dose versus the use of an additional dose or doses of glucocorticoids in the perioperative period.

Evidence was from a single small randomised controlled trial and was very low certainty due to potential selection, detection and reporting biases, imprecision from very small event rates, and indirectness as the study included people with various conditions; only four participants per group had inflammatory arthritis (specifically, RA).

Evidence from the observational studies was also very low certainty overall. Three of the four studies were large population-based cohorts from administrative datasets but the generalisability was limited as it is unclear if participants in these studies received additional ("stress") glucocorticoid doses perioperatively and all cohorts included RA participants (but no participants with other forms of IA).

The applicability of the evidence is limited. The RCT included people with various conditions and only four participants per group (8 of 18 total participants) had IA. The evidence from the observational studies was largely limited to participants with RA undergoing elective joint replacements, and while the studies assessed prior glucocorticoid exposure, it is unclear if participants received stress glucocorticoid doses perioperatively in three of the four studies.

We did not identify any studies that directly investigated values and preferences of patients with IA on glucocorticoids undergoing surgical procedures. It is likely that the primary trade-off is between a potential risk of clinically-relevant intra- or post-operative adrenal insufficiency (including serious hypotension) and the potential risk of surgical complications related to a transient increase in the glucocorticoid dose, including infection and poor wound healing.

The panel considered that there is likely to be variability between individual patients in their preferences, influenced by a variety of factors including the duration and dose of glucocorticoids, the nature and complexity of the surgery, concomitant medications (particularly b/tsDMARDs or other immunomodulatory drugs), the risk and potential consequences of infection or altered wound healing, comorbidity, history of adrenal insufficiency, and individual risk tolerance.

The panel also noted that the existing evidence base is insufficient to permit a reliable estimate of competing risks, but that this should not discourage discussion of perioperative glucocorticoid management with patients who are planning elective surgery. Indeed, the panel considered it important that decisions regarding perioperative glucocorticoid management ought to form part of the routine pre-operative counselling and consent process.The conditional nature of this recommendation and the lack of certainty in the evidence base implies that the choice of intervention is dependent on the individual clinical circumstances and patient preferences, and heightens the importance of clear communication between members of the clinical team (including surgeon, anaesthetist, perioperative physician, pharmacist, and GP) and the patient prior to surgery.

Glucocorticoids are a low-cost intervention and short-term adjustment of dosing is unlikely to have any direct cost implications. Major adverse outcomes of surgery in this population, including infection, altered wound healing or intra- or post-operative haemodynamic compromise, may carry an additional resource burden to the individual patient and to the healthcare system, however there are no available cost-effectiveness data regarding the decision to adjust glucocorticoid dosing in the perioperative period.

We do not anticipate any environmental impact of glucocorticoid stress dosing in the perioperative period.

Individuals with limited access to specialists might face challenges in making informed decisions about their perioperative glucocorticoid management. Variation in access to care might lead to inappropriate use of glucocorticoids, either underuse or overuse, and result in possible health inequities. However, this specific recommendation — whether to continue glucocorticoids at the usual dose or provide an increased "stress" dose — is not expected to have substantial health equity implications.

Adjustment of glucocorticoid dosing in the perioperative period with the goal of optimising the balance between the risk of adrenal insufficiency and the risk of adverse effects related to glucocorticoids, is likely to be acceptable to all stakeholders including consumers, surgeons, rheumatologists, hospital administrators, and funders.

The panel noted there are a number of clinicians who may be involved in perioperative care, including anaesthetists, surgeons, pharmacists, and perioperative physicians, in addition to the patient’s usual care team (including the GP and rheumatologist), and that the acceptability of the recommendation may vary between team members. It was also noted that the panel for this recommendation did not include a surgeon or anaesthetist but did include one member with expertise and experience in perioperative management, and two pharmacists.

In general, any adjustment of glucocorticoid dosing in the perioperative period for people with IA is likely to be feasible to implement. However it is plausible that there may be uncertainty regarding the specialist with the primary responsibility for advising the patient and making clinical decisions in the perioperative setting (e.g., rheumatologist, anaesthetist, surgeon, perioperative physician, general practitioner), and that the approach to the use of stress dosing and the use of clinical guidelines on this topic may vary between specialties, which might lead to variation in implementation of this recommendation.

While the majority of panellists considered that there were important or potential feasibility issues, some considered there were no major issues while others felt that the recommendation might be difficult to implement. After discussion, there was consensus that there were some potential feasibility issues. One such potential barrier to implementation of this recommendation may be the number of clinicians involved in perioperative and intra-operative clinical decision-making, particularly if recommendations on this topic developed by different specialty groups do not align. The panel highlighted that this again underscores the importance of clear communication between clinicians and with the patient, and a deliberate and clearly-documented approach to the conditional nature of this recommendation, including the individual clinical context and the patient’s values and preferences.

There was considerable debate among the panel about whether the potential benefits and harms of opioids are balanced or whether there is net harm on average. There was a majority (but not unanimous) panel vote for “Small net benefit, or little difference between the alternatives”, based on the data from trials of short-term use of opioids in people with rheumatoid arthritis. Some panellists were of the opinion that, on balance, there were important harms that outweighed the potential benefits.

Evidence regarding opioid analgesics in patients with rheumatoid arthritis (RA) comes mostly from short-term trials comparing various opioids with placebo: opioids may improve mean pain and increase the number of people who report treatment success but may have little or no effect on function. Opioids may have a small or no effect on the number of people reporting adverse events in short-term clinical trials (up to 6 weeks). We are uncertain of the effect of opioids on serious adverse events or withdrawals from treatment due to either inefficacy or adverse effects.

There is very little evidence regarding the efficacy and safety of opioids compared with other types of analgesic medications in patients with RA.

The Cochrane review of opioids for RA pain ⁷ used a ‘net efficacy adjusted for risk’ (NEAR) analysis to estimate the balance of benefits and harms of opioids in RA. The NEAR calculation combines efficacy and adverse event outcomes into a single estimate of the proportion of participants in each treatment group who achieve the clinically desirable state of both analgesic response and avoidance of important adverse effects. By this measure, there was no evidence of a difference between opioids and placebo: the NEAR relative risk (that is, the chance of achieving benefit without harm for those treated with opioids versus placebo) was 1.20 (95% CI, 0.89–1.61).

Although there was relatively little high-quality data specific to RA, the widespread use of opioids for chronic non-cancer pain, including in patients with RA, has been increasingly recognised in recent years and use is increasing. The prevalence of opioid use among Australians with RA has been estimated at 32.9% (95% CI 31.6, 34.2) ¹⁵ . The use of high-potency opioids in this group has increased over time and now represents approximately a third of all opioid use. Across all indications, over 1.9 million Australians initiate treatment with opioids each year, the majority for non-cancer pain ¹³. 2.6% of people who initiate opioids in Australia for non-cancer pain will go on to become persistent users over a 12-month period ¹⁶. In a US emergency department setting, for every 48 patients given an initial opioid prescription, one will become a long-term opioid user ²³. Rheumatological conditions may be a risk factor for transition from first use to long-term opioid use for non-cancer pain ¹⁷.

Use of low-potency or low-dose opioids is also associated with a risk of transition to high-dose therapy. Tolerance with long-term use of opioids often leads to a requirement for dose escalation. Almost one percent of Australians who commence weak opioids may escalate to high-dose opioid therapy (equivalent to at least 90mg oral morphine per day) and approximately one in 13 will transition to a strong opioid during a 12-month period ²¹. Older people are at increased risk of transition to higher doses or more potent opioids. Higher doses are associated with an increased risk of opioid-related morbidity and mortality.

All of the trials included in the evidence summary for this recommendation evaluated short-term use (up to 6 weeks); the effectiveness of opioids for RA pain beyond 6 weeks is unknown. Potential harms from chronic use of opioids are well recognised but are unlikely to be adequately captured in short-term trials. Evidence is available regarding harms in trials of medium-term use of opioids (2 weeks to 13 months) for chronic non-cancer pain ¹⁸: opioid use is associated with an increased risk of any adverse event (RR versus placebo 1.42, 95% CI 1.22, 1.66) and serious adverse events (RR 2.75, 95% CI 2.06, 3.67). Adverse effects that occurred at significantly increased rates in opioid users included nausea and vomiting, constipation, dizziness, drowsiness and fatigue, pruritus, hot flushes and sweating.

Evidence for opioid harms in the longer term comes from observational studies. Data from a study of almost 100,000 UK patients treated with opioids for musculoskeletal conditions and followed for a median of 3.4 years indicate that long-term opioid use is associated with a dose-dependent increase in the risk of fractures and other major trauma, overdose, falls, gastrointestinal pathology and addiction ¹⁹. Episodes of major trauma increased by 84 for every 10,000 person-years of use of low-dose opioids (equivalent to <20mg morphine per day) and by 139/10,000 for doses equivalent to at least 50mg morphine per day. While the risk of dependence, addiction and overdose is generally well recognised, the potential for other important harms associated with opioid use, including fractures, endocrine dysfunction, and worsening of chronic pain (opioid-induced hyperalgesia), is frequently underestimated.

Opioid prescription is also associated with an increased risk of mortality. Treatment with opioids for chronic non-cancer pain is associated with a 58% increase in the risk of all-cause mortality compared with other analgesic therapies (HR 1.58, 95% CI 1.38 to 1.82), equivalent to 148 excess deaths per 10,000 person-years of treatment ²⁰.

Evidence from trials regarding benefits and harms of opioids in patients with RA was generally of low certainty. Important outcomes for which there was low certainty evidence included pain, function and the number of participants reporting treatment success. There was very low certainty evidence for serious adverse events and withdrawals due to either adverse events or inadequate analgesia.

We included trials of any opioid, including medications with other mechanisms of action in addition to opioid receptor effects (i.e. tramadol) and compound analgesics (e.g. an opioid plus paracetamol). Some of the opioids included in these trials are not currently available in Australia. Importantly, most of the evidence comes from trials that pre-date the modern era of treat-to-target management of RA, and none of the participants in the included trials were using biologic or targeted synthetic DMARDs for the treatment of their RA, so the results may not necessarily generalise to people with RA who have been treated with an intensive disease-modifying strategy. No trials were longer than 6 weeks in duration.

Patients with rheumatoid arthritis report relief of pain to be their highest priority ⁹. A 2018 systematic review of patient values and preferences regarding chronic non-cancer pain ⁸ included 6 studies involving patients with various forms of chronic non-cancer pain (CNCP). While none of the included studies specifically recruited patients with RA, one included patients with other forms of musculoskeletal pain, including osteoarthritis and fibromyalgia. It is likely that the preferences and values of patients with various forms of CNCP are broadly applicable to people with RA pain.

Across the included studies, pain relief was consistently reported as being of high importance to patients. The most important adverse effects were nausea/vomiting and alteration of mental state. Other adverse effects, including risk of addiction, were of lesser importance.

Included studies that evaluated the trade-off between benefits and harms indicated that substantial pain relief (at least 2 points on a 10-point numerical scale) would be required for a net benefit in the presence of adverse effects (particularly nausea or vomiting).

No studies considered the risk of overdose or death, although these outcomes and deliberate misuse or diversion of prescription opioids are common concerns at a societal level. Similarly, while opioid prescribers also view CNCP as an important problem, they are more likely to report concerns about the risk of opioid misuse or addiction ¹⁰.

Direct costs of prescription opioid analgesics to individuals are generally relatively low, although prescribing rules in Australia require frequent visits to healthcare providers for ongoing prescriptions which may result in higher out-of-pocket costs or may have an impact on work.

The societal costs of chronic non-cancer pain are significant. Approximately 3 million Australian adults use prescription opioids ¹³. The majority of these are people with non-cancer pain, although it is likely that opioid use for rheumatoid arthritis represents only a small fraction of these cases. Potential costs to society of widespread use of opioids for chronic non-cancer pain include direct and indirect costs relating to overdose, misuse, dependence and altered productivity.

Socioeconomic factors are important determinants of chronic pain, opioid use and opioid-related adverse outcomes ¹¹²². Variation in access to rheumatologists for people with early inflammatory arthritis may lead to delayed treatment with DMARDs in some settings, which may increase the risk of persistent pain throughout the disease course. Access to comprehensive and multidisciplinary chronic pain management services varies within Australia. In particular, access may be limited for socially-disadvanted people and those in regional and remote areas ¹². This may affect pain management strategies in general and, specifically, may alter the balance of potential benefits and harms of opioid analgesics.

It is likely that the option of short-term use of opioids guided by an explicit management plan in a shared decision-making framework would be acceptable to both patients and prescribers. Evidence from qualitative studies suggests that while patients strongly value relief of pain, this is balanced by concerns about adverse effects (particularly nausea and vomiting) and that patients feel that opioids should be used with caution ¹⁴. Given the potential net harms of opioids use at the population level, widespread use of opioids for RA pain may be less acceptable to policy-makers.

Opioids are a widely-available and feasible treatment option.

Benefits: In adults with axial spondyloarthritis (including ankylosing spondylitis), we are uncertain of the benefits of opioid use compared with other analgesics (i.e. NSAIDs) or in addition to NSAIDs versus NSAIDs alone, with regards to treatment success, pain or function, due to very low certainty evidence.

Harms: In adults with axial spondyloarthritis (including ankylosing spondylitis), we are uncertain of the harms of opioid use compared with other analgesics (i.e. NSAIDs) or in addition to NSAIDs versus NSAIDs alone, with regards to the number of adverse events, ineffective analgesia and withdrawals due to either or both of these outcomes. No serious adverse events were reported in two randomised controlled trials.

There was considerable debate among the panel regarding the balance of potential benefits and harms of opioid therapy in people with axial spondyloarthritis pain. Only two randomised controlled trials were available, which provided very low certainty evidence regarding efficacy or short-term harms. Given the lack of evidence regarding efficacy and the known risk of long-term harm associated with opioid use in the broader population of people with chronic non-cancer pain, the panel formed the consensus view that the known risks outweigh the potential benefits in this setting.

There are few data regarding the prevalence of opioid use among people with axial spondyloarthritis, particularly in the Australian setting. Observational data from the US suggests that opioid use for spondyloarthritis pain is common. Data from a claims database indicated that patients with various forms of inflammatory arthritis are at higher risk of receiving long-term opioid prescriptions, and that the risk was highest for patients with ankylosing spondylitis ²⁸. In a cohort of US patients with ankylosing spondylitis, 9.5% were taking chronic opioids (defined as taking opioids daily for at least 6 months) and 21.7% were using intermittent or “on demand” opioids ²⁹. Opioid use was higher among those with multiple comorbidities, polypharmacy (including anxiolytics and hypnotics), higher functional impairment and depression.

Across all indications, over 1.9 million Australians initiate treatment with opioids each year, the majority for non-cancer pain ¹³. 2.6% of people who initiate opioids in Australia for non-cancer pain will go on to become persistent users over a 12-month period ¹⁶. In a US emergency department setting, for every 48 patients given an initial opioid prescription, one will become a long-term opioid user ²³. Rheumatological conditions may be a risk factor for transition from first use to long-term opioid use for non-cancer pain ¹⁷.

Use of low-potency or low-dose opioids is also associated with a risk of transition to high-dose therapy. Tolerance with long-term use of opioids often leads to a requirement for dose escalation. Almost one percent of Australians who commence weak opioids may escalate to high-dose opioid therapy (equivalent to at least 90mg oral morphine per day) and approximately one in 13 will transition to a strong opioid during a 12-month period ²¹. Older people are at increased risk of transition to higher doses or more potent opioids. Higher doses are associated with an increased risk of opioid-related morbidity and mortality.

Although there is little trial evidence regarding harms of opioids that is specific to spondyloarthritis, data are available regarding harms in patients with chronic non-cancer pain from trials of medium-term use of opioids (2 weeks to 13 months) ¹⁸: opioid use is associated with an increased risk of any adverse event (RR versus placebo 1.42, 95% CI 1.22, 1.66) and serious adverse events (RR 2.75, 95% CI 2.06, 3.67). Adverse effects that occurred at significantly increased rates in opioid users included nausea and vomiting, constipation, dizziness, drowsiness and fatigue, pruritus, hot flushes and sweating.

Evidence for opioid harms in the longer term comes from observational studies. Data from a study of almost 100,000 UK patients treated with opioids for musculoskeletal conditions and followed for a median of 3.4 years indicate that long-term opioid use is associated with a dose-dependent increase in the risk of fractures and other major trauma, overdose, falls, gastrointestinal pathology and addiction ¹⁹. Episodes of major trauma increased by 84 for every 10,000 person-years of use of low-dose opioids (equivalent to <20mg morphine per day) and by 139/10,000 for doses equivalent to at least 50mg morphine per day. While the risk of dependence, addiction and overdose is generally well recognised, the potential for other important harms associated with opioid use, including fractures, endocrine dysfunction, and worsening of chronic pain (opioid-induced hyperalgesia), is frequently underestimated.

Opioid prescription is also associated with an increased risk of mortality. Treatment with opioids for chronic non-cancer pain is associated with a 58% increase in the risk of all-cause mortality compared with other analgesic therapies (HR 1.58, 95% CI 1.38 to 1.82), equivalent to 148 excess deaths per 10,000 person-years of treatment ²⁰.

There is very low certainty evidence for the effect of opioid use compared with other analgesics (i.e. NSAIDs) or in addition to NSAIDs versus NSAIDs alone, for:

• participant-reported treatment success
• total withdrawals due to adverse events or inadequate analgesia
• withdrawals due to adverse events
• withdrawals due to inadequate analgesia
• number of participants reporting adverse events
• number of serious adverse events
• mean change in pain from baseline
• mean change in function from baseline

A 2018 systematic review of patient values and preferences regarding chronic non-cancer pain (CNCP) ⁸ included 6 studies involving patients with various forms of CNCP. While none of the included studies specifically recruited patients with axial spondyloarthritis, one included patients with other forms of musculoskeletal pain, including osteoarthritis and fibromyalgia. It is likely that the preferences and values of patients with various forms of CNCP are broadly applicable to people with spondyloarthritis pain.

Across the included studies, pain relief was consistently reported as being of high importance to patients. The most important adverse effects were nausea/vomiting and alteration of mental state. Other adverse effects, including risk of addiction, were of lesser importance.

Included studies that evaluated the trade-off between benefits and harms indicated that substantial pain relief (at least 2 points on a 10-point numerical scale) would be required for a net benefit in the presence of adverse effects (particularly nausea or vomiting).

No studies considered the risk of overdose or death, although these outcomes and deliberate misuse or diversion of prescription opioids are common concerns at a societal level. Similarly, while opioid prescribers also view CNCP as an important problem, they are more likely to report concerns about the risk of opioid misuse or addiction ¹⁰.

Direct costs of prescription opioid analgesics to individuals are generally relatively low, although prescribing rules in Australia require frequent visits to healthcare providers for ongoing prescriptions which may result in higher out-of-pocket costs or may have an impact on work.

The societal costs of chronic non-cancer pain are significant. Approximately 3 million Australian adults use prescription opioids ¹³. The majority of these are people with non-cancer pain, although it is likely that opioid use for axial spondyloarthritis represents only a small fraction of these cases. Potential costs to society of widespread use of opioids for chronic non-cancer pain include direct and indirect costs relating to overdose, misuse, dependence and altered productivity. There was concern among some panellists that the societal costs of opioid therapy may be disproportionately high compared with other analgesic strategies in axial spondyloarthritis, and that this may therefore represent an important argument against the use of opioids in this setting.

Socioeconomic factors are important determinants of chronic pain, opioid use and opioid-related adverse outcomes ¹¹²². Variation in access to rheumatologists for people with inflammatory back pain or other features of axial spondyloarthritis may lead to delayed diagnosis and treatment in some settings, which may increase the risk of persistent pain throughout the disease course.

Access to comprehensive and multidisciplinary chronic pain management services varies within Australia. In particular, access may be limited for socially-disadvantaged people and those in regional and remote areas ¹². This may affect pain management strategies in general and, specifically, may alter the balance of potential benefits and harms of opioid analgesics.

It is likely that the option of short-term use of opioids guided by an explicit management plan in a shared decision-making framework would be acceptable to both patients and prescribers. Evidence from qualitative studies in patients with musculoskeletal disease suggests that while patients strongly value relief of pain, this is balanced by concerns about adverse effects (particularly nausea and vomiting) and that patients feel that opioids should be used with caution ¹⁴. Given the potential net harms of opioids use at the population level, widespread use of opioids for axial spondyloarthritis pain may be less acceptable to policy-makers.

Opioids are generally a widely-available and feasible treatment option. In some settings, alternatives to opioids, particularly for chronic pain, including multidisciplinary and multimodal pain management services, may be difficult to access or implement. This may particularly be the case in rural or remote areas ¹², among socially-disadvantaged individuals, or in primary care settings where resources or access to multidisciplinary or specialist services are limited. In such cases the barriers to opioid avoidance or deprescribing may make a recommendation against opioid use difficult to implement without additional resources. The panel strongly supported the development of implementation initiatives that seek to facilitate either avoidance of opioid initiation or deprescribing in those already using opioids.

In people with psoriatic arthritis (PsA), we are uncertain whether opioid use is safe or effective as no studies addressing the question were identified.

There are few data regarding the prevalence of opioid use among people with psoriatic arthritis, however across all indications over 1.9 million Australians initiate treatment with opioids each year, the majority for non-cancer pain ¹³. 2.6% of people who initiate opioids in Australia for non-cancer pain will go on to become persistent users over a 12-month period ¹⁶. In a US emergency department setting, for every 48 patients given an initial opioid prescription, one will become a long-term opioid user ²³. Rheumatological conditions may be a risk factor for transition from first use to long-term opioid use for non-cancer pain ¹⁷.

Use of low-potency or low-dose opioids is also associated with a risk of transition to high-dose therapy. Tolerance with long-term use of opioids often leads to a requirement for dose escalation. Almost one percent of Australians who commence weak opioids may escalate to high-dose opioid therapy (equivalent to at least 90mg oral morphine per day) and approximately one in 13 will transition to a strong opioid during a 12-month period ²¹. Older people are at increased risk of transition to higher doses or more potent opioids. Higher doses are associated with an increased risk of opioid-related morbidity and mortality.

Although there is little trial evidence regarding harms of opioids that is specific to psoriatic arthritis, data are available regarding harms in patients with chronic non-cancer pain from trials of medium-term use of opioids (2 weeks to 13 months) ¹⁸: opioid use is associated with an increased risk of any adverse event (RR versus placebo 1.42, 95% CI 1.22, 1.66) and serious adverse events (RR 2.75, 95% CI 2.06, 3.67). Adverse effects that occurred at significantly increased rates in opioid users included nausea and vomiting, constipation, dizziness, drowsiness and fatigue, pruritus, hot flushes and sweating.

Evidence for opioid harms in the longer term comes from observational studies. Data from a study of almost 100,000 UK patients treated with opioids for musculoskeletal conditions and followed for a median of 3.4 years indicate that long-term opioid use is associated with a dose-dependent increase in the risk of fractures and other major trauma, overdose, falls, gastrointestinal pathology and addiction ¹⁹. Episodes of major trauma increased by 84 for every 10,000 person-years of use of low-dose opioids (equivalent to <20mg morphine per day) and by 139/10,000 for doses equivalent to at least 50mg morphine per day. While the risk of dependence, addiction and overdose is generally well recognised, the potential for other important harms associated with opioid use, including fractures, endocrine dysfunction, and worsening of chronic pain (opioid-induced hyperalgesia), is frequently underestimated.

Opioid prescription is also associated with an increased risk of mortality. Treatment with opioids for chronic non-cancer pain is associated with a 58% increase in the risk of all-cause mortality compared with other analgesic therapies (HR 1.58, 95% CI 1.38 to 1.82), equivalent to 148 excess deaths per 10,000 person-years of treatment ²⁰.

No studies addressing this question were identified.

A 2018 systematic review of patient values and preferences regarding chronic non-cancer pain ⁸ included 6 studies involving patients with various forms of chronic non-cancer pain (CNCP). While none of the included studies specifically recruited patients with psoriatic arthritis, one included patients with other forms of musculoskeletal pain, including osteoarthritis and fibromyalgia. It is likely that the preferences and values of patients with various forms of CNCP are broadly applicable to people with psoriatic arthritis pain.

Across the included studies, pain relief was consistently reported as being of high importance to patients. The most important adverse effects were nausea/vomiting and alteration of mental state. Other adverse effects, including risk of addiction, were of lesser importance.

Included studies that evaluated the trade-off between benefits and harms indicated that substantial pain relief (at least 2 points on a 10-point numerical scale) would be required for a net benefit in the presence of adverse effects (particularly nausea or vomiting).

No studies considered the risk of overdose or death, although these outcomes and deliberate misuse or diversion of prescription opioids are common concerns at a societal level. Similarly, while opioid prescribers also view CNCP as an important problem, they are more likely to report concerns about the risk of opioid misuse or addiction ¹⁰.

Direct costs of prescription opioid analgesics to individuals are generally relatively low, although prescribing rules in Australia require frequent visits to healthcare providers for ongoing prescriptions which may result in higher out-of-pocket costs or may have an impact on work.

The societal costs of chronic non-cancer pain are significant. Approximately 3 million Australian adults use prescription opioids ¹³. The majority of these are people with non-cancer pain, although it is likely that opioid use for psoriatic arthritis represents only a small fraction of these cases. Potential costs to society of widespread use of opioids for chronic non-cancer pain include direct and indirect costs relating to overdose, misuse, dependence and altered productivity. There was concern among some panellists that the societal costs of opioid therapy may be disproportionately high compared with other analgesic strategies in psoriatic arthritis, and that this may therefore represent an important argument against the use of opioids in this setting.

Socioeconomic factors are important determinants of chronic pain, opioid use and opioid-related adverse outcomes^{11 22}. Variation in access to rheumatologists for people with psoriatic arthritis may lead to delayed diagnosis and treatment in some settings, which may increase the risk of persistent pain throughout the disease course.

Access to comprehensive and multidisciplinary chronic pain management services varies within Australia. In particular, access may be limited for socially-disadvantaged people and those in regional and remote areas ¹². This may affect pain management strategies in general and, specifically, may alter the balance of potential benefits and harms of opioid analgesics.

It is likely that the option of short-term use of opioids guided by an explicit management plan in a shared decision-making framework would be acceptable to both patients and prescribers. Evidence from qualitative studies in patients with musculoskeletal disease suggests that while patients strongly value relief of pain, this is balanced by concerns about adverse effects (particularly nausea and vomiting) and that patients feel that opioids should be used with caution ¹⁴. Given the potential net harms of opioids use at the population level, widespread use of opioids for psoriatic arthritis pain may be less acceptable to policy-makers.

Opioids are generally a widely-available and feasible treatment option. In some settings, alternatives to opioids, particularly for chronic pain, including multidisciplinary and multimodal pain management services, may be difficult to access or implement. This may particularly be the case in rural or remote areas ¹², among socially-disadvantaged individuals, or in primary care settings where resources or access to multidisciplinary or specialist services are limited. In such cases the barriers to opioid avoidance or deprescribing may make a recommendation against opioid use difficult to implement without additional resources. The panel strongly supported the development of implementation initiatives that seek to facilitate either avoidance of opioid initiation or deprescribing in those already using opioids.

In people with rheumatoid arthritis (RA), we are uncertain whether the short-term use of glucocorticoids for disease flare is safe or effective as no studies addressing the primary question were identified. We did not include trials of short-term glucocorticoids for the treatment of persistent synovitis in one or more joints, or for the induction of remission in recent-onset RA. The latter will be addressed in a separate living recommendation in this guideline.

Anecdotal evidence supports the efficacy of short-term or single doses of glucocorticoids, administered either systemically (orally, intramuscularly or intravenously) or via intra-articular injection, but there are no reliable estimates of the relative benefits and harms. Many of the recognised adverse effects of long-term systemic glucocorticoid therapy (including endocrine disturbance, weight gain, thin skin, easy bruising, osteoporosis and atherosclerotic vascular disease) are unlikely to be associated with very short-term glucocorticoid use however some adverse effects may plausibly occur even with short-term use. The effect of different doses and routes of administration of glucocorticoids on the short-term risk of adverse effects is unknown.

There is little observational evidence available regarding the specific risks of short-term glucocorticoids in people with RA, however data from the general population are consistent with a small increase in the risk of serious adverse events. In a study using data from the Taiwanese National Health Insurance Research Database (n = 15,859,129 adult participants), brief courses (14 days or fewer) of oral glucocorticoids were associated with an approximate doubling of the risk of GI bleeding, sepsis or heart failure in the subsequent month ⁹⁴. Similarly, a study using a US insurance database found that a short course (less than 30 days) of systemic glucocorticoids was associated with an increased risk of sepsis, venous thromboembolism and fracture in the subsequent 30 days ⁹⁵. The estimated absolute risk of any of these serious adverse events is approximately one per 1,000 short courses of glucocorticoid ⁹³.

No studies addressing the primary question were identified.

Among patients with RA, outcomes ranked as being of high importance include pain, joint damage, fatigue, daily functioning and mobility ⁷⁷. A systematic review of 36 studies of patient preferences for RA outcomes with disease-modifying therapies found that, in general, treatment benefits were considered more important than risks (including serious adverse events). However there was heterogeneity among included studies and variation between patients, suggesting that individual values and preferences are likely to be an important determinant of treatment decisions in people with rheumatoid arthritis ³⁵.

For glucocorticoid therapy in particular, tradeoffs between benefits and risks may vary between patients and within the same person over time, depending on a variety of factors including individual preferences and values, age, comorbidities, disease activity, concurrent medications, social, work and family roles, and risk tolerance.

Beneficial and harmful effects often coexist in patients who use glucocorticoids for RA. Data on this topic primarily relate to long-term users of glucocorticoids and it is unclear to what extent the findings can be extrapolated to the short-term use of glucocorticoids for the treatment of flares.

Among Australian patients with rheumatoid arthritis who responded to an online survey regarding glucocorticoid use (n=804), adequate disease control and adverse effects were both important considerations for patients using prednisolone. The most common adverse effects reported by those who had discontinued glucocorticoid therapy were weight gain (27.5%), osteoporosis (14.7%) and neuropsychiatric symptoms (13.8%) ⁸³.

In another survey of 124 patients with RA, 62% felt that glucocorticoids “helped their disease a lot”, but 86% reported at least one adverse effect. Two-thirds of participants reported that the benefits of glucocorticoids were greater than the adverse effects ⁸². The adverse effects identified as “worst” were symptomatic experiences such weight gain, sleep disturbance, gastrointestinal upset and muscle weakness, rather than other physiological effects such as osteoporosis, hypertension and hyperglycaemia. Similarly, in an online survey of glucocorticoid users in a social network, the most important adverse effects were weight gain, insomnia and change in facial appearance ⁷⁸.

Glucocorticoids are widely available and low-cost, particularly for short-term use.

Glucocorticoids are a low-cost, familiar and accessible intervention. Most panellists felt that there were no important equity concerns related to the use of glucocorticoids as a treatment for RA flare in Australia. However, some panellists noted that there remains some potential for variation in the use of glucocorticoids between different groups, particularly in regard to availability of intra-articular therapy for people who may have difficulty accessing specialist care (e.g., those living in rural or remote locations, or those with lower socioeconomic status). Similarly, it is plausible that people who are less able to access timely specialist care may be at a higher risk of over-treatment with glucocorticoids rather than adjustment of DMARD therapy. Overall the panel felt that for this intervention, these potential differences in access were unlikely to result in important inequity in health outcomes.

The short-term use of glucocorticoids to treat RA flares is likely to be generally acceptable to patients, clinicians and other stakeholders. However, there is a risk of adverse effects in the short-term (e.g., disordered glycaemic control in people with diabetes, or neuropsychiatric effects) that may reduce acceptability in people who are susceptible to such effects. In addition, some people with RA - particularly those who have previously been treated with glucocorticoids for longer periods or who have experienced adverse effects related to glucocorticoids - may be unwilling to accept any use of glucocorticoids, even brief courses.

There are likely to be no important feasibility issues as glucocorticoids have been in common use for the treatment of rheumatoid arthritis for several decades.

Based on data from eight randomised controlled trials (RCTs), in adults with rheumatoid arthritis (RA) receiving DMARD therapy, compared to a placebo (or DMARD-only control), use of low dose glucocorticoids for longer than six months:

• Probably results in a small improvement in mean disease activity and mean pain
• Probably results in a very small reduction in radiographic disease progression (mean x-ray changes)
• May result in more people reporting glucocorticoid-specific adverse events
• May have little or no effect on the number of people achieving low disease activity, or on mean function
• May have little or no effect on the number of people withdrawing due to adverse events, the number of people reporting serious adverse events, or the number of people reporting any adverse event

Adverse effects related to glucocorticoid use have long been recognised, however the absolute risk of these effects may be difficult to estimate. RCTs are typically not powered to estimate differences in adverse events and may not have long enough follow-up periods to detect important long-term harms. Observational studies may supplement data from RCTs but are at risk of bias, particularly confounding by indication, which may overestimate the risk of adverse events that is attributable to glucocorticoids. There is continued debate in the literature regarding the relative safety of low-dose glucocorticoids in RA, particularly whether there might be a 'safe' threshold dose below which the risks of long-term glucocorticoid therapy become negligible and the putative benefits might be retained ⁷¹.

Observational data suggest that long-term glucocorticoid use for RA may be associated with an increased risk of major adverse effects, including death. One observational study of 779 people with RA followed for an average of nine years found an increased risk of death in those who were exposed to glucocorticoids (IRR 1.77, 95% CI 1.36 - 2.32), equivalent to a 7% increase in the risk of death for each one milligram increase in the daily dose of prednisolone ⁸⁰. In this study, a threshold dose was detected: daily doses below 8mg and a cumulative dose below 40g were not associated with an increased mortality risk, and a cumulative dose below 9g was associated with a reduced mortality risk (adjusted HR 0.59, 95% CI 0.36 - 0.95). Similarly, one cohort study of RA patients that included a stricter adjustment for potential confounding by indication found an increase in all-cause mortality risk with glucocorticoid use (HR 1.64, 95% CI 1.50–1.79), and a threshold daily prednisolone dose of 5mg below which there was no increase in mortality ⁷⁶. Another demonstrated an increased mortality risk in RA patients who had received a cumulative dose of at least 700mg of prednisolone ⁷².

Data from observational studies of glucocorticoid use in people with RA are also consistent with a dose-dependent increased risk of major adverse effects, including osteoporosis and fracture ⁷²⁷³, type 2 diabetes mellitus ⁷²⁷⁴, serious infection ⁷², cataracts ⁷⁰ and cardiovascular events ⁷²⁷⁵. 10-year follow-up of participants in an RCT that compared prednisolone 7.5mg/day versus placebo in addition to DMARDs for the first two years of RA demonstrated an increased risk of cerebrovascular events in those who had been randomised to prednisolone (HR 3.7, 95% CI 1.2 - 11.4) but did not find a difference in coronary events (HR 0.98, 95% CI 0.4 - 2.6) ⁶⁹.

There is some evidence that the risk of some adverse effects that are important to consumers, including thin skin, easy bruising, and change in appearance, may occur even at very low daily doses of glucocorticoids (i.e., less than 5mg prednisolone per day) ⁷⁹. The risk of incident type 2 diabetes has also been reported to be elevated (HR 1.66, 95% CI 1.37 - 2.02) in people with RA treated with prednisolone at doses below 5mg per day ⁷⁴.

Adrenal suppression can be demonstrated in at least one-third of RA patients treated with low-dose prednisolone ⁶⁸, although estimates of the rate of clinically-important episodes of adrenal insufficiency vary ⁶⁷.

The panel debated the balance of benefits and harms at length. Some panellists were of the view that the known harms of glucocorticoids outweigh the benefits, particularly as the benefits demonstrated in RCTs of long-term use were of very low magnitude (generally less than the minimal clinically-important difference for each of the important outcomes). However, the panel also recognised that while the potential harmful effects of glucocorticoids are well-known, the absolute magnitude of these effects is difficult to estimate from either RCTs (which may not adequately detect uncommon or long-term adverse events) or observational studies (which are subject to various confounding factors) and, as such, the balance of benefits and harms may vary between individual patients. Therefore, the panel agreed on a judgment of “small net benefit, or little difference between alternatives” but noted that for many patients, the important potential harmful effects of long-term low-dose glucocorticoids may outweigh the marginal potential clinical benefits.

There was MODERATE certainty evidence for mean pain, disease activity and radiographic disease progression.

There was LOW certainty evidence for the number of people achieving low disease activity, the number of adverse events of special interest, the number of people withdrawing due to adverse events, the number of serious adverse events, and the number of people reporting adverse events and mean function.

Among patients with RA, outcomes ranked as being of high importance include pain, joint damage, fatigue, daily functioning and mobility ⁷⁷. A systematic review of 36 studies of patient preferences for RA outcomes with disease-modifying therapies found that, in general, treatment benefits were considered more important than risks (including serious adverse events). However there was heterogeneity among included studies and variation between patients, suggesting that individual values and preferences are likely to be an important determinant of treatment decisions in people with rheumatoid arthritis ³⁵.

For glucocorticoid therapy in particular, tradeoffs between benefits and risks may vary between patients and within the same person over time, depending on a variety of factors including individual preferences and values, age, comorbidities, disease activity, concurrent medications, social, work and family roles, and risk tolerance.

Beneficial and harmful effects often coexist in patients who use glucocorticoids for RA. Among Australian patients with rheumatoid arthritis who responded to an online survey regarding glucocorticoid use (n=804), adequate disease control and adverse effects were both important considerations for patients using prednisolone. The most common adverse effects reported by those who had discontinued glucocorticoid therapy were weight gain (27.5%), osteoporosis (14.7%) and neuropsychiatric symptoms (13.8%) ⁸³.

In another survey of 124 patients with RA, 62% felt that glucocorticoids “helped their disease a lot”, but 86% reported at least one adverse effect. Two-thirds of participants reported that the benefits of glucocorticoids were greater than the adverse effects ⁶³. The adverse effects identified as “worst” were symptomatic experiences such weight gain, sleep disturbance, gastrointestinal upset and muscle weakness, rather than other physiological effects such as osteoporosis, hypertension and hyperglycaemia. Similarly, in an online survey of glucocorticoid users in a social network, the most important adverse effects were weight gain, insomnia and change in facial appearance ⁷⁸.

Direct costs to patients due to glucocorticoids are low. The direct cost to the healthcare system is also lower than some other treatments for RA (particularly biological and targeted synthetic DMARDs), however there are also potential additional costs to the individual and society related to some of the recognised adverse effects of long-term glucocorticoid use, including osteoporosis, cardiovascular disease and cataracts.

The panel debated this issue at some length. While some felt that there are likely to be few resource implications of treatment with glucocorticoids, which is a low-cost intervention at the individual level, there was some concern within the panel that the long-term adverse effects of glucocorticoids (e.g., cataracts and fractures) may result in an additional resource burden on the healthcare system as a whole. Therefore, the panel agreed on a judgment of “important issues, or potential issues not investigated”.

Glucocorticoids are a low-cost, familiar and accessible intervention. In contrast, in Australia csDMARDs are typically prescribed under the supervision of a rheumatologist and high-cost interventions (i.e., b/tsDMARDs) must be prescribed by a specialist. Therefore it is plausible that glucocorticoid therapy may be offered in preference to other DMARDs to some people who may have difficulty accessing specialist care (e.g., those living in rural or remote locations, or those with lower socioeconomic status), or in some other groups of patients, such as those who are older or who have lower health literacy.

Most panellists felt that there were no important equity concerns related to the use of glucocorticoids in Australians with RA. However, some panellists expressed concern that there remains some potential for variation in the use of glucocorticoids between different groups that may result in inequity in health outcomes. After some discussion, the panel agreed on a judgment of “important issues, or potential issues not investigated”.

A study of treatment preferences among patients with RA commencing one of four different initial DMARD regimens in a clinical trial setting found that while over a third of participants had hoped to avoid a treatment strategy that included a tapering high-dose glucocorticoid component, this preference was much stronger in those who had no experience of using glucocorticoids for their RA ⁶⁶. This suggests that for a proportion of patients, the benefits of glucocorticoids may be sufficient to increase acceptability despite concerns about adverse effects, although this may vary according to dose and duration of treatment.

In current practice, glucocorticoids are often used for longer periods or at higher doses than generally recommended, suggesting that prescribers and patients may make judgments about glucocorticoid risk-benefit trade-offs in daily practice that indicate relatively higher acceptability than anticipated ⁸¹.

There are likely to be no important feasibility issues as glucocorticoids have been in common use for the treatment of rheumatoid arthritis for several decades.

In adults with active rheumatoid arthritis who are initiating DMARD therapy, the addition of glucocorticoids (compared to a glucocorticoid placebo):

• improves mean function and slightly improves mean disease activity
• probably slightly improves the number of people achieving remission and probably slightly improves mean overall pain
• probably results in little or no difference in the number of people withdrawing due to adverse events
• may result in little or no difference in the number of people reporting serious adverse events or the number of serious adverse events reported
• no placebo-controlled studies were found that looked at the number of people achieving low disease activity

No placebo-controlled studies were found that included people adding or switching DMARD therapy.

In adults with active rheumatoid arthritis who are initiating, adding or switching DMARD therapy, the addition of glucocorticoids compared to DMARD therapy alone (i.e. without a placebo):

• may increase the number of people achieving remission, and may improve mean function and mean disease activity
• may have little or no effect on the number of people achieving low disease activity and the number of people withdrawing due to adverse events
• we are uncertain whether bridging glucocorticoid therapy results in more people with serious adverse events as the certainty of the evidence is very low

No studies were found that looked at mean overall pain.

In adults with active rheumatoid arthritis who are initiating, adding or switching DMARD therapy, compared to oral bridging glucocorticoid therapy, parenteral bridging glucocorticoid therapy:

• may have little or no effect on mean pain, function, disease activity and the number of people achieving remission and low disease activity
• we are uncertain whether parenteral bridging glucocorticoid therapy results in more people reporting serious adverse events and withdrawing due to adverse events compared with oral bridging glucocorticoid therapy, as the certainty of the evidence is very low

We are uncertain of the benefits and harms of additional bridging glucocorticoid therapy in adults with active rheumatoid arthritis who are on background glucocorticoid therapy and are adding or switching DMARD therapy, due to a lack of RCT evidence.

Adverse effects related to glucocorticoid use have long been recognised, however the absolute risk of these effects may be difficult to estimate. RCTs are typically not powered to estimate differences in adverse events and may not have long enough follow-up periods to detect important long-term harms. Observational studies may supplement data from RCTs but are at risk of bias, particularly confounding by indication, which may overestimate the risk of adverse events that is attributable to glucocorticoids. There is continued debate in the literature regarding the relative safety of low-dose glucocorticoids in RA, particularly whether there might be a 'safe' threshold dose below which the risks of long-term glucocorticoid therapy become negligible and the putative benefits might be retained ⁷¹.

Observational data suggest that long-term glucocorticoid use for RA may be associated with an increased risk of major adverse effects, including death. One observational study of 779 people with RA followed for an average of nine years found an increased risk of death in those who were exposed to glucocorticoids (IRR 1.77, 95% CI 1.36 - 2.32), equivalent to a 7% increase in the risk of death for each one milligram increase in the daily dose of prednisolone ⁸⁰. In this study, a threshold dose was detected: daily doses below 8mg and a cumulative dose below 40g were not associated with an increased mortality risk, and a cumulative dose below 9g was associated with a reduced mortality risk (adjusted HR 0.59, 95% CI 0.36 - 0.95). Similarly, one cohort study of RA patients that included a stricter adjustment for potential confounding by indication found an increase in all-cause mortality risk with glucocorticoid use (HR 1.64, 95% CI 1.50–1.79), and a threshold daily prednisolone dose of 5mg below which there was no increase in mortality ⁷⁶. Another demonstrated an increased mortality risk in RA patients who had received a cumulative dose of at least 700mg of prednisolone ⁷².

Data from observational studies of glucocorticoid use in people with RA are also consistent with a dose-dependent increased risk of major adverse effects, including osteoporosis and fracture ⁷²⁷³, type 2 diabetes mellitus ⁷²⁷⁴, serious infection ⁷², cataracts ⁷⁰ and cardiovascular events ⁷²⁷⁵. 10-year follow-up of participants in an RCT that compared prednisolone 7.5mg/day versus placebo in addition to DMARDs for the first two years of RA demonstrated an increased risk of cerebrovascular events in those who had been randomised to prednisolone (HR 3.7, 95% CI 1.2 - 11.4) but did not find a difference in coronary events (HR 0.98, 95% CI 0.4 - 2.6) ⁶⁹.

There is some evidence that the risk of some adverse effects that are important to consumers, including thin skin, easy bruising, and change in appearance, may occur even at very low daily doses of glucocorticoids (i.e., less than 5mg prednisolone per day) ⁷⁹. The risk of incident type 2 diabetes has also been reported to be elevated (HR 1.66, 95% CI 1.37 - 2.02) in people with RA treated with prednisolone at doses below 5mg per day ⁷⁴.

Adrenal suppression can be demonstrated in at least one-third of RA patients treated with low-dose prednisolone ⁶⁸, although estimates of the rate of clinically-important episodes of adrenal insufficiency vary ⁶⁷.

There is little observational evidence available regarding the specific risks of short-term glucocorticoids in people with RA, however data from the general population are consistent with a small increase in the risk of serious adverse events. In a study using data from the Taiwanese National Health Insurance Research Database (n = 15,859,129 adult participants), brief courses (14 days or fewer) of oral glucocorticoids were associated with an approximate doubling of the risk of GI bleeding, sepsis or heart failure in the subsequent month ⁹⁴. Similarly, a study using a US insurance database found that a short course (less than 30 days) of systemic glucocorticoids was associated with an increased risk of sepsis, venous thromboembolism and fracture in the subsequent 30 days ⁹⁵. The estimated absolute risk of any of these serious adverse events is approximately one per 1,000 short courses of glucocorticoid ⁹³.

The panel also noted the important current concern during the COVID-19 pandemic regarding the potential harmful effects of both glucocorticoids and uncontrolled disease activity on the risk of COVID-19 (including severe outcomes) and the negative impact of glucocorticoids on the immunogenicity of COVID-19 vaccines ^33159162. While COVID-19 infection rates remain high, the balance of risks and harms of induction glucocorticoid regimens may be altered, particularly in people who are at increased risk of exposure to COVID-19 or of severe COVID-19 disease outcomes.

Given the above, the panel agreed that the likely benefits of induction glucocorticoids may be offset by the risk of harms, but that this balance is likely to vary importantly between individuals.

1. For adults with active RA who are initiating DMARD therapy, addition of glucocorticoids versus placebo:

• There was HIGH certainty evidence that bridging glucocorticoid therapy improves mean function (i.e. lower HAQ score) and slightly improves mean disease activity.
• There was MODERATE certainty evidence that bridging glucocorticoid therapy probably slightly improves the number of people achieving remission and mean overall pain (i.e. lower VAS score), and probably has little or no difference on the number of people withdrawing from adverse events.
• There was LOW certainty evidence that bridging glucocorticoid therapy may have little or no difference on either the number of people reporting serious adverse events or the number of serious adverse events reported.
• No studies reported data on the number of people with low disease activity.

2. For adults with active RA who are initiating, adding or switching DMARDs, addition of glucocorticoids (versus no additional glucocorticoids):

• There was LOW certainty evidence the bridging glucocorticoid therapy may improve the number of people achieving remission, mean function and mean disease activity; but may have little or no effect on the number of people achieving low disease activity or the number of people withdrawing due to adverse events.
• There was VERY LOW certainty evidence regarding whether or not bridging glucocorticoid therapy results in more people with serious adverse events.
• No studies reported data on mean pain.

3. For adults with active RA who are initiating, adding or switching DMARDs, parenteral (IV or IM) bridging glucocorticoids versus oral bridging glucocorticoids:

• There was LOW certainty evidence for the number of people achieving remission, the number of people achieving low disease activity, mean overall pain, function and mean disease activity.
• There was VERY LOW certainty evidence for the number of people withdrawing due to adverse events and the number of people reporting serious adverse events.

Among patients with RA, outcomes ranked as being of high importance include pain, joint damage, fatigue, daily functioning and mobility ⁷⁷. A systematic review of 36 studies of patient preferences for RA outcomes with disease-modifying therapies found that, in general, treatment benefits were considered more important than risks (including serious adverse events). However there was heterogeneity among included studies and variation between patients, suggesting that individual values and preferences are likely to be an important determinant of treatment decisions in people with rheumatoid arthritis ³⁵. Similarly, people with early RA prioritise the rapid relief of symptoms and restoration of function, however there is evidence of variation in priorities and preferences, both between individuals and within individuals over time ¹⁵⁷¹⁵⁸.

For glucocorticoid therapy in particular, tradeoffs between benefits and risks may vary between patients and within the same person over time, depending on a variety of factors including individual preferences and values, age, comorbidities, disease activity, concurrent medications, social, work and family roles, and risk tolerance.

The consumer panellist noted that values and preferences are likely to differ between people with recent onset RA versus those who have had RA for some time. For example, people who have had prior experience with glucocorticoids for RA (particularly those who have previously experienced adverse effects) may weigh the potential benefits and harms differently from those with early RA.

Among Australian patients with RA who responded to an online survey regarding glucocorticoid use (n=804), adequate disease control and adverse effects were both important considerations for patients using prednisolone. The most common adverse effects reported by those who had discontinued glucocorticoid therapy were weight gain (27.5%), osteoporosis (14.7%) and neuropsychiatric symptoms (13.8%) ⁸³.

In another survey of 124 patients with RA, 62% felt that glucocorticoids “helped their disease a lot”, but 86% reported at least one adverse effect. Two-thirds of participants reported that the benefits of glucocorticoids were greater than the adverse effects ⁸². The adverse effects identified as “worst” were symptomatic experiences such weight gain, sleep disturbance, gastrointestinal upset and muscle weakness, rather than other physiological effects such as osteoporosis, hypertension and hyperglycaemia. Similarly, in an online survey of glucocorticoid users in a social network, the most important adverse effects were weight gain, insomnia and change in facial appearance ⁷⁸.

The panel agreed unanimously that there is likely to be substantial variability in values and preferences, and that there remains an important need for further research on the impact of different glucocorticoid regimens on the outcomes that are of highest importance to consumers, and how these may vary in the setting of short-term bridging therapy versus long-term use.

Glucocorticoids are widely available and low-cost, particularly for short-term use. However the panel noted that there might be some indirect costs (to individuals and society) related to the recognised adverse effects of glucocorticoid use, although these are likely to be more important with long-term use. We are not aware of any data regarding the impact of short-term glucocorticoid therapy on the environment, however a major negative impact is unlikely.

The panel debated this issue at some length. While some felt that there are likely to be few resource implications of treatment with glucocorticoids, which is a low-cost intervention at the individual level, there was some concern within the panel that the potential adverse effects of glucocorticoids may result in an additional resource burden on the healthcare system as a whole. Therefore, the panel agreed on a judgment of “important issues, or potential issues not investigated".

Glucocorticoids are a low-cost, familiar and accessible intervention. There is a theoretical risk of some variation in the use of glucocorticoids between different groups, particularly people who may have difficulty accessing specialist care (e.g., those living in rural or remote locations, or those with lower socioeconomic status). It is plausible that people who are less able to access timely specialist care may be at a higher risk of over-treatment with glucocorticoids rather than adjustment of DMARD therapy.

The panel discussed this issue at length and reiterated that a recommendation in favour of bridging glucocorticoids should not permit or encourage the use of glucocorticoids in place of the appropriate use of DMARDs. The panel also noted that in Australia there is variation in timely access to specialist rheumatology care. While glucocorticoids may help to reduce health inequity in people with RA by enabling suppression of disease activity in those who face delayed access to specialist care, there is also a risk of prolonged use of glucocorticoids or delayed use of adequate DMARD therapy in this group, which may create inequity in long-term outcomes.

Overall, the panel agreed that there are likely to be effects on health equity related to glucocorticoid use that are complex and poorly understood, and for this reason the panel voted for 'important issues, or potential issues not investigated'.

There is likely to be some variation among patients, clinicians and other stakeholders in the acceptability of the addition of glucocorticoids to the initial DMARD regimen in people with RA.

A study of treatment preferences among patients with RA commencing one of four different initial DMARD regimens in a clinical trial setting found that while over a third of participants had hoped to avoid a treatment strategy that included a tapering high-dose glucocorticoid component, this preference was much stronger in those who had no experience of using glucocorticoids for their RA ⁶⁶. Similarly, a qualitative study of participants in a clinical trial of combination csDMARD and glucocorticoid therapy in early RA found that concerns about glucocorticoid adverse effects were common at the start of the trial, however these concerns diminished over the course of the trial as beneficial effects were experienced or expected side-effects did not occur ¹⁵⁵. This suggests that for a proportion of patients, the benefits of glucocorticoids may be sufficient to increase acceptability despite concerns about adverse effects, although this may vary according to dose and duration of treatment.

A study of rheumatologists in Belgium found variation in the acceptability of a regimen of combination DMARDs and glucocorticoids for the treatment of early RA ¹⁵⁶. The most important barriers to the rheumatologists' implementation of such a regimen were the increased risk of adverse effects (particularly in patients with comorbidity) and perceived reluctance of their patients to take combination therapy. Notably, two of the six highest-ranked barriers to implementation related to the use of glucocorticoids (specifically, concerns about use of prednisolone for longer than 28 weeks or at a dose higher than 10mg/day).

In current practice, glucocorticoids are often used for longer periods or at higher doses than generally recommended, suggesting that prescribers and patients may make judgments about glucocorticoid risk-benefit trade-offs in daily practice that indicate relatively higher acceptability than anticipated ⁸¹.

There are likely to be no important feasibility issues as glucocorticoids have been in common use for the treatment of rheumatoid arthritis for several decades.