Seventy-two studies were included (1969 to 1996). These included over 454 patients in CRPS trials. The number of patients in PNP trials is not stated. There were 47 placebo-controlled trials and 3 comparing one drug to another in the treatment of PNP. There were 17 placebo-controlled trials, 4 trials comparing one drug to another and 1 trial comparing a drug to no drug in the treatment of CRPS.
1) Determine which diagnostic procedures and treatments were consistently supported by controlled trial data.
PNP: There was consistent support (defined as two or more trials with positive results) for the use of tricyclic antidepressants, intravenous and topical lidocaine, intravenous ketamine, carbamazepine and topical aspirin. There was limited evidence for the effectiveness of oral, topical or epidural clonidine. Conflicting evidence was found on the effectiveness of phenytoin, mexiletine, capsaicin, oral NSAIDs, and intravenous morphine. Treatments found to be ineffective were magnesium, propranolol, lorazepam, phentolamine and oral codeine.
CRPS: The only treatment that showed consistent effect was oral corticosteroids. There was limited support for the effectiveness of topical DMSO, epidural clonidine and intravenous regional blocks (IVRBs) using bretylium and ketanserin. Mixed results were found for intranasal calcitonin and intravenous phentolamine. Treatments found to be ineffective were guanethidine and reserpine IVRBs, and limited evidence that droperidol and atropine IVRBs are ineffective was found.
2) Evaluate studies which had conflicting results for the same treatment to determine if there were large differences in methods scores or other significant differences in the trial designs to explain the differences.
PNP: Contradictory trial results were found with mexiletine, capsaicin, oral NSAIDs and intravenous morphine. Mexiletine was found to be effective only after post-hoc analysis of diabetic patients in a large trial. The capsaicin studies that showed an effect all suffered from a lack of blinding. Oral NSAIDs are probably ineffective as two large, well-designed studies found no effect, while intravenous morphine probably is effective as two large, well-designed studies did show an effect.
CRPS: Trial data were contradictory for intranasal calcitonin and intravenous phentolamine. The two calcitonin studies had high methods scores, similar designs, but different results. The differences may be due to different outcome measures, with the study using continuous objective outcome data showing no effect. The phentolamine studies showed mixed effects, but the two with the highest methods scores both showed no effect. Issues that need to be addressed include the optimal dose, duration and onset of treatment.
3) Perform a meta-analysis of any treatment which was evaluated by three or more controlled trials with similar methodology and adequate data, but with contradictory results.
PNP: meta-analysis of data from 5 capsaicin studies showed improvement in the GESs in 167/247 (68%) of capsaicin versus 122/257 (47%) of placebo treated patients. The pooled odds ratio was 2.35 (95% CI 1.48 to 3.22).
CRPS: not enough studies met the criteria for meta-analysis.
4) Identify potential flaws in design and differences in methods between PNP and CRPS trials.
Mean methods scores were higher for PNP trials than for CRPS trials : 66.2+/-1.5 vs 57.6 +/-2.9, p<0.01. CRPS trials had fewer subjects than the PNP trials, 18.6 +/-3 vs 28.8 +/-3.4, p<0.01. CRPS trials were less likely to use placebo controls (71% CRPS vs 94% of PNP trials, p<0.01). CRPS trials used fewer double-blind designs 57% vs 94%, p<0.001. Statistical tests were used less often in CRPS trials 67% vs 90%, p<0.05. More subjects dropped out during the PNP studies (30% vs 15%, p<0.05), due primarily to drop-outs in tricyclic antidepressant studies. Neither CRPS nor PNP trials tested for adequacy of blinding. Other design flaws found were the lack of long-term outcome measures and the small number of outcomes measures used per trial.
5) Determine what treatments were used in both PNP and CRPS trials and if there were treatment outcome differences specific to these diagnoses.
The only two treatments used in both diagnoses were intravenous phentolamine and epidural clonidine. There was evidence to support the effectiveness of clonidine and the ineffectiveness of phentolamine in both diagnoses.