There’s this expectation that guidelines put perspective into the scheme of things as new evidence is obtained on the use of both novel and established agents. Personally, a well made algorithm is the core of guidelines as it organizes the tools available to a physician after weighing in concerns such as efficacy, safety, tolerability and, hopefully, costs (biologics and targeted synthetic DMARDs are far from being cheap!).
The following are my opinions on parts of the ACR 2015 Guidelines on Rheumatoid arthritis that I feel more time should have been devoted to in clarifying issues. Others may call these missed opportunities (see the editorial by van Vollenhoven R. Nature Rev Rheum 2016; 12. doi: 10.1038/nrrheum.2015.181). I’d refer to them as conflicts as these added more confusion rather than organization to how we manage the disease. Feel free to discuss them with me as I’d appreciate more ideas on these matters.
The GRADE approach is a system of interrogating the evidence and making recommendations used by groups such as the Cochrane Collaboration and relevant in making meta-analyses, health technology assessments (HTA) and guidelines. Traditionally, the level of evidence is classified based on whether metas/ systematic reviews (highest), RCTs, observational studies or expert opinion (lowest) are available to answer a concern. GRADE adds to this layer by looking across the studies and evaluating for study limitations, inconsistency, directness, imprecision, publication bias, dose response, magnitude of effect and presence of confounders. These would either increase or decrease the eventual quality of the evidence. It has implications on whether newer studies will be able to change the confidence in the estimate of the effect. I suggest referring to the series of Guyatt, et al published in the Journal of Clinical Epidemiology (2011) for better appreciation of this. Following these, the panel would determine which outcomes were critical in making decisions and balance benefits versus harms. These would all be considered in evaluating the strength of recommendation which has been simplified to two choices – strongly recommend or conditionally recommend – or as was taught to us, “Do it” or “Probably Do It”. A strong recommendation means the panel is confident the adherence would result in more benefit than harm. While a conditional recommendation indicates that the panel is less confident about the impact of following the recommendation and that adherence would PROBABLY result in benefit rather than harm. The PRA used GRADE when it made local CPGs on gout (which I was part of) and knee osteoarthritis. I’ll admit familiarity to GRADE but not claim expertise.
The use of GRADE in ACR 2015 RA Guidelines probably contributed most to the confusion about the recommendations. I’ve seen ACR successfully use GRADE in making recommendations in the past (check out their 2012 Osteoarthritis Guide!) but here I felt more at a loss. Twelve of 19 instructions on treatment were conditionally recommended. While most of these had low to very low quality of evidence, there were some where quality was high and yet did not merit a strong recommendation. It would be interesting to review how the panel evaluated outcomes and weighed benefits versus harms in these instances that lead to the translation of “good” evidence to “weak” recommendations. Cases in point are the the use of combination DMARDs in MTX-naive Early RA with moderate to high disease activity (the 2008 version allowed for its use but the 2015 preferred monotherapy rather than combination in this setting) and the addition of steroids in established RA following DMARDs and biologics and disease activity remains moderate to high.
Another source of confusion is the appreciation of TNFi, non-TNFi biologics and tofacitinib in the hierarchy of treatment following conventional DMARD use. Recommendation No. 4 for both early and established RA indicate that if disease activity remains moderate-high following DMARD (MTX) monotherapy either combination DMARDs, TNFi, non-TNFi biologics or tofacitinib may be added WITHOUT any preference. However, subsequent statements (No. 5 in Early RA and No. 6, 7, 8 and 10 in Established RA) indicate preference of either TNFi or non-TNF biologics over tofacitinib in various scenarios when disease activity remains moderate-high. The only setting where tofacitinib is preferred is when disease activity is moderate to high despite multiple TNFi therapies and non-TNF biologics are contraindicated. There are several studies demonstrating benefit with tofacitinib in RA (DMARD-naive RA, inadequate response following DMARDs or biologics). But the often cited reasons in the papers is the lack of long term safety data for tofacitinib and the long clinical experience doctors have with biologics. Tofacitinib is available as an oral preparation with costs comparable to biologics. Again, it would be interesting to find out how the panel evaluated critical outcomes and weighed benefits versus harms in their final analysis of the evidence.
Surprisingly, the panel had a different take on the role of steroids in RA treatment. The guidelines suggests using low dose steroids for treating flares and as add on when disease activity remains moderate-high following DMARDs, TNFi and non-TNF biologics. Aside from steroids being potent anti-inflammatory agents, it has been shown to exhibit disease modifying activity when used for >6 weeks. There have been an increase in safety concerns over long term use of moderate-high doses of steroids in the past years. But most of the studies of steroids in RA use prednisone <=7.5mg/day which many believe has an acceptable balance of benefits versus harms. While there is no doubt about its use in treating flares, it’s surprising that ACR recommends adding it only after DMARDs, TNFi and non-TNFI biologics (makes me wonder – which would ACR prefer more steroids or tofacitinib as a drug of last resort). This is contrary to what other guidelines advise – that’s the use of low dose steroids during the first 6 months of RA treatment to serve as both as bridge until DMARDs fully take effect and for its low level of disease modifying properties.
In general, I feel that this iteration of RA guidelines lost impact with the way it presented its recommendations. I understand that these are really just guides and are not intended to be rules carved in stone. But readers often turn to these in the hopes of gaining clarity on how best to maximize medications available in the market for the benefit of our patients. I still await for subsequent papers to help in this endeavor.