In January 2007, the American Academy of Neurology (AAN), the American Association of Neuromuscular and Electrodiagnostic Medicine, and the American Academy of Physical Medicine and Rehabilitation convened an expert panel from the United States and Canada, selected to represent a broad range of relevant expertise. In August 2008, a literature search of MEDLINE and EMBASE was performed in all languages using the MeSH term diabetic neuropathies and its text word synonyms and key words for the therapeutic interventions of interest (see appendix e-1 on the Neurology®
Web site at www.neurology.org
for a full list of search terms). The search identified 2,234 citations, the titles and abstracts of which were reviewed by at least 2 authors for relevance, resulting in 463 articles. All of these articles were reviewed in their entirety, and of these, the panel identified 79 relevant articles. Each of these articles was rated by at least 2 authors according to the AAN criteria for the classification of therapeutic articles (appendix e-2), and recommendations were linked to the strength of evidence (appendix e-3) and to effect size of the intervention. Disagreements regarding classification were arbitrated by a third reviewer.
Articles were included if they dealt with the treatment of PDN, described the intervention clearly, reported the completion rate of the study, and defined the outcome measures clearly. The panel also considered the side effects of the treatment and measures of function and QOL, if any. Case reports and review articles were excluded.
We anticipated that studies would use varying measures for quantifying pain reduction. For the purposes of this guideline we preferred the following outcome measures, listed in order of preference:
- The difference in the proportion of patients reporting a greater than 30 to 50% change from baseline on a Likert or visual analog pain scale (VAS) as compared to no treatment (placebo) or the comparative treatment. The Likert scale is an 11-point linear scale ranging from 0 (no pain) to 10 (maximum pain), and the patient rates his or her pain level on this scale.4–6
- The percent change from baseline on a Likert or VAS as compared to no treatment (placebo) or the comparative treatment.6
- Any other quantitative measure of pain reduction provided by the investigators.
For studies reporting the difference in the proportion of patients reporting a greater than 30% to 50% reduction in pain, we considered a risk difference of >20% a large effect (number needed to treat [NNT] <5), a risk difference of >10% to 20% (NNT >5 to 10) a moderate effect, and a risk difference of ≤10% (NNT >10) a small effect, where risk difference is the reduction in pain in the active treatment group minus the reduction in the control group. For studies using a mean reduction from baseline on a Likert scale or VAS as compared to no treatment (placebo) or a comparative treatment, we considered a reduction difference of >30% a large effect, >15% to 30% a moderate effect, and ≤15% a small effect. For any other quantitative measure of pain reduction, we considered a reduction of >30% a large effect, >15% to 30% a moderate effect, and ≤15% a small effect.
The panel recognized that older studies generally lacked measures of QOL and function compared to more recent studies. Furthermore, the panel was aware that a standardized QOL measure for PDN or a standardized assessment of function is not available, and multiple instruments were used to measure QOL, such as the SF-36® Health Survey, subsections of the SF-36, and function (such as sleep interference).
Studies with the highest levels of evidence for each intervention are discussed in the text, and data from other studies are shown in the tables. Details of Class I, II, and III studies are presented in the evidence tables.