We found no first or second tier evidence of efficacy. Evidence was downgraded primarily because of the short duration of the studies, small numbers of participants in comparisons, reporting of completer analyses in cross-over studies, and lack of desirable primary outcomes.
Four studies reported at least one pain-related outcome indicating some improvement with imipramine compared with placebo. Three studies (Kvinesdal 1984; Sindrup 1990; Sindrup 1992) reported improvement on a set of neuropathy symptoms, which included pain, but only one (Sindrup 1990) reported the pain item separately (as group median scores). Details of data from individual studies are shown in Appendix 5.
Third tier evidence
Kvinesdal 1984 assessed six symptoms (pain, dysaesthesia, paraesthesia, numbness, nightly aggravation, and sleep disturbances) which were combined to give a neuropathy symptom score; pain was not reported separately. The authors reported that 8/12 participants who completed both phases of the cross-over (ITT 8/15) assessed their condition as "improved" (undefined) with imipramine 100 mg daily, compared with 1/12 (ITT 1/15) with placebo.
In Sindrup 1990, pain was not reported separately but only as part of a set of neuropathy symptoms.
Participants' median VAS scores for pain shown on a graph indicated about an 80% reduction in pain rating at the end of the treatment period from baseline or placebo for those who completed all phases of the cross-over. Participants with at least 50% reduction, compared with placebo, in combined VAS scores for five symptoms (pain, paraesthesia, dysaesthesia, nightly aggravation, and sleep disturbances) with imipramine 25 mg to 350 mg daily, numbered 12/19 for completers (ITT 12/29). For at least 30% reduction, the number was 14/19 for completers (ITT 14/29).
Sindrup 1992 also reported a combined set of neuropathy symptoms (pain, paraesthesiae, dysaesthesiae, numbness, nightly exacerbation, and sleep disturbance) assessed by both participants and observers. Participants reporting at least 50% reduction in symptoms with imipramine 25 mg to 350 mg (median 150 mg) daily totaled 3/18 for those who completed all phases of the cross-over (ITT 3/22), with 9/18 completers (ITT 9/22) reporting at least 30% reduction, both figures relative to placebo. The relevant figures for mianserin were 3/18 and 5/18, respectively.
Sindrup 2003 reported participants' ratings for several types of pain (pain paroxysms, constant pain, touch-evoked pain, pressure-evoked pain). These were reduced by 29%, 29%, 4%, and 20%, respectively, for completers with imipramine up to 150 mg daily. An overall reduction of 23% from baseline is quoted.
Participants' global evaluation of pain relief (complete, good, moderate, slight, none, worse) was complete or good for 9/29 with imipramine in those who completed all phases of the cross-over (ITT 9/40), 7/30 with venlafaxine (ITT 7/40), and 2/29 with placebo (ITT 2/40). For an evaluation of complete, good, or moderate, the numbers were 14/29 with imipramine (ITT 14/40), 8/30 with venlafaxine (ITT 8/40), and 2/29 with placebo (ITT 2/40). A subgroup analysis indicated that the subgroup of participants with diabetic neuropathy were more likely to obtain clinically relevant pain relief.
The study authors report an NNT of 2.7 for > 50% pain relief, but do not report the raw data on which this calculation was based, or any confidence intervals. We believe it may be based on a completer analysis of participants giving complete, good, or moderate global evaluations of pain relief, which we would consider to be closer to ≥ 30% pain relief.
Turkington 1980 reported complete (100%) pain relief with imipramine 100 mg daily (20/20 completers and ITT) and no relief with placebo (0/20 completers and ITT).
Details of adverse events reported in individual studies are in Appendix 6. Four studies (Kvinesdal 1984; Sindrup 1990; Sindrup 1992; Sindrup 2003) reported some information about adverse events, but reporting was inconsistent and fragmented; the remaining study (Turkington 1980) did not mention adverse events.
Participants experiencing any adverse event
No study adequately reported the number of participants who experienced one or more adverse events for all those who were treated (ITT).
Kvinesdal 1984 reported that "side effects were generally few". Dry mouth was reported by 9/12 completers with imipramine and 1/12 with placebo, and impaired micturition was reported in two participants with imipramine, and dizziness in two participants, probably with imipramine.
Sindrup 1990 asked participants to score the intensity of a range of adverse events using a VAS. Mean ratings for dry mouth, sweating, dizziness and fatigue were higher for imipramine than for placebo, but only dry mouth was notably higher. Frequencies of adverse events were not given except that 4/18 participants showed withdrawal symptoms (nausea, tremor, vomiting) after cessation of imipramine treatment.
Sindrup 1992 asked participants to score the intensity of a number of adverse events on a scale of 0 to 2.0. The most common events were dry mouth, orthostatic dizziness, and fatigue. The total adverse events score was not significantly different between imipramine and mianserin, but scores for both active drugs were significantly higher than for placebo, with the total adverse event score for imipramine approximately four times that for placebo. Almost half (10/22) of participants experienced one or more adverse events during treatment with placebo.
Sindrup 2003 reported that of 29 participants completing all three treatment phases with valid results, 20 experienced an adverse event (of any intensity) with imipramine, 20 with venlafaxine, and 14 with placebo. At least one additional adverse event leading to withdrawal was reported with imipramine, four with venlafaxine, and two with placebo.
There was a tendency for more adverse events with active treatment, particularly dry mouth with imipramine and tiredness with venlafaxine. Specific adverse events that were more common with imipramine than with placebo were as follows (in completers): dry mouth (12, 3); nausea (5, 1); sweating (5, 0); dizziness (3, 1); and blurred vision (1, 0).
Turkington 1980 did not report any adverse events.
Participants experiencing any serious adverse event
No study reported serious adverse events as such, although Sindrup 2003 reported that one participant had to drop out due to hospitalisation for a "pronounced" urinary tract infection (report does not state during which phase).
Details of withdrawals reported in individual studies are provided in Appendix 6. Turkington 1980 did not report any withdrawals, so withdrawals are discussed only for participants in the other four studies.
Withdrawals due to adverse events
The remaining four studies reported withdrawals due to adverse events.
Seven participants withdrew during treatment with imipramine, three during a dose-finding phase (nausea, fatigue, dry mouth), and four (of 109) during a treatment phase. All four cited dizziness, in combination with, variously, orthostatic hypotension (1), tiredness (3), and vomiting (2).
Three (of 109) dropped out during placebo treatment (gastritis, nausea/diarrhoea, vomiting), and four withdrew during venlafaxine treatment.
Withdrawals due to lack of efficacy
There was only one withdrawal due to lack of efficacy reported during imipramine treatment (Sindrup 2003) and none during placebo treatment.
Withdrawals for other reasons
Three studies (69 participants) reported eight withdrawals for other reasons (Kvinesdal 1984; Sindrup 1990; Sindrup 1992).
During imipramine treatment, three participants withdrew for lack of compliance, two for personal reasons, and two because they needed analgesia for other pain. One from each of placebo and mianserin treatment groups withdrew for personal problems.
Two participants were lost to follow-up in Sindrup 2003 (treatment group was not specified).