Commentary on the KDIGO Clinical Practice Guideline for Glomerulonephritis

Not so long ago, a paper published in Kidney International explored the reasons for the paucity of high-quality clinical trials in glomerulonephritis (GN). The authors identified several factors, including the low prevalence of disease, variability in clinical presentation, variability in treatment responses, lack of consensus about definitions, difficulty in recruiting patients, the high cost of randomized controlled trials, and the lack of collaborative research efforts. To this list, I would also add our poor understanding of the pathogenesis of most if not all of the glomerulonephritides commonly encountered in clinical practice. This point is relevant because it limits the variety of new therapies that we can put up for clinical trials, and ultimately our ability to punch through the therapeutic envelope, so to speak. This is not to say that progress has not been made, but simply that we do not have enough hard data to provide precise advice on all aspects of therapy. Hence, there is a need for clinical practice guidelines. Guidelines serve many purposes, the most important of which is to set down boundaries to define what is reasonable to do, and in so doing, either directly or indirectly, also set down what is not reasonable to do. The KDIGO Clinical Practice Guideline for Glomerulonephritis published in 2012 aims to do this and much more. It attempts to present a series of treatment recommendations and suggestions based on the available evidence for a number of glomerulonephritides. It also grades the quality of evidence behind these recommendations. Notably, several conditions are not covered, such as the management of diabetic nephropathy, dysproteinaemias, fibrillary GN and haemolyticuraemic syndrome. Arguably, these latter conditions are not strictly inflammatory glomerulonephritides, but nonetheless, they can affect the glomerulus predominantly. In a series of chapters, each dedicated to a specific GN, a set format is used to tackle various aspects of clinical management such as induction therapy, maintenance therapy, management of relapse, and management of resistant disease. At the start of these subsections, recommendations and suggestions are made (annotated Level 1 and 2, respectively), followed by grading of the quality of evidence behind these recommendations (annotated A, B, C and D for high, moderate, low and very low quality of evidence). A brief background is provided, the rationale for these recommendations/suggestions is discussed, including a summary of the most relevant clinical trials, and a number of recommendations are made for future research. The references to publications are listed together in one location at the end of the document. To achieve this mammoth task, an Evidence Review Team of physician-methodologists worked closely with a Work Group of clinicians to develop the relevant clinical questions, search the literature up to November 2011, grade the quality of evidence for each topic, and finally arrive at a series of recommendations and suggestions. There seems to be an operational assumption that every minute aspect of our clinical decision-making process has to be based on a randomized controlled trial. This position may be scientifically sound, but it is not achievable in reality or even desirable. Incidentally, the panel of clinicians which made up the Work Group is international in representation, and I note with interest the inclusion of an Australian nephrologist in the group. In total, 47 Level 1 recommendations and 120 Level 2 suggestions were made. The major problem, however, lies with the fact that the vast majority (77%) of these recommendations and suggestions were made based on what was judged to be either low (class C) or very low (class D) quality evidence. It is important to note that the available quality of evidence varies widely between the different conditions. For example, the quality of evidence for some aspects of management in lupus nephritis and idiopathic membranous nephritis was moderate, but very low for others such as focal and segmental glomerulosclerosis. Only four of 47 Level 1 recommendations were based on high quality (class A) evidence, and 24 of 47 Level 1 recommendations were based on moderate quality (class B) evidence. That left 19 Level 1 recommendations based on class C or D evidence. On the other hand, 110 of the 120 Level 2 suggestions were based on class C or D evidence. One would think that Level 1 recommendations should only be made in the context of class A and B evidence, and that no recommendations or suggestions should be made based on class C or D evidence. But this is apparently not the case. In the ‘Methods for guideline development’ section, we are told that besides the quality of evidence, the Work Group took into account other factors such as the balance of benefits and harm, cost, perception of what the majority of nephrologists would do in a particular situation, and on top of these, I suspect perhaps a dose of ‘expert opinion’. In relation to cost, there is regular mention at the start of each chapter that ‘the cost implications for global application of this guideline’ have been addressed in an earlier chapter. However, cost-benefit analysis is nowhere to be found in this document. For relatively cheap medicines such as prednisolone, this is probably not a significant issue. But this is a relevant issue for expensive items such as mycophenolate mofetil or rituximab, and it is bs_bs_banner