Controlled Release Morphine for Chronic Non-Cancer Pain—Multiple N of 1 Trials

Abstract

Purpose of the study: Use of controlled release morphine for chronic non-cancer pain has risen greatly during the last decade. The literature provides evidence of its efficacy in the short term but is more equivocal for long-term use. In the light of the limited evidence base, the potential problems associated with long term opioid use, the placebo responsiveness of pain, and the difficulty of generalizing randomised controlled trials on heterogenous populations to individual patients, the authors felt that single subject (N of 1) trials may be a useful means by which patients and health professionals could make clinical judgements. We also felt that conducting N of 1 trials with multiple crossovers on a medication with frequent side effects and abstinence symptoms presented specific methodological challenges worthy of review. Methods: Patients taking regular opioids prior to the trial were withdrawn in an open label fashion and were opioid free for a minimum of 2 weeks prior to a single crossover open label trial. This was followed by a titration phase in which patients were commenced at 10 mg twice daily of controlled release morphine (MS Contin) and were allowed to increase the dose to as high as 50 mg twice daily, to a level at which clinical improvements were noted without significant side effects. The open and double blind trials were then conducted at this dose. Pain, sleep, an overall measure of helpfulness, depression, anxiety, stress and activity interference were all measured during the second of each two-week period. In those with significant improvements on the open label trial, a four crossover, double blind randomized controlled trial, lasting a total of 16 weeks, was undertaken, with measures taken in the second week of each period. Patients, naive to regular opioid use prior to the study, moved straight to the open label trial. Results: Seventeen patients were enrolled in the study, and ten completed the randomized controlled trial. Three patients, one of whom was taking regular opioids prior to the commencement of the study, withdrew from the trial after the open label crossover because of a lack of benefits on the controlled release morphine and side effects. A further four patients withdrew during the double blind component of the single subject trial, two because of recurrent abstinence symptoms with each rotation to placebo, one because of increased pain levels on placebo, and one because of side effects as well as unrecognised recurrent withdrawals. A further three, of the ten patients who completed the trial suffered from significant withdrawal symptoms with each rotation placebo. In each case of recurrent withdrawals the patients had been taking regular opioids for prolonged periods prior to entering the study. Four of the six patients did not recognize their withdrawal symptoms as such, including abstinence related pain. All results of the double blind trials were recorded at significant levels of greater than .01 because of the multiple outcomes and comparisons. In spite of the universal use of stool softeners, seven of the ten patients who completed the trial had more side effects on the active rotations compared to placebo. The combination of withdrawal symptoms and side effects meant that nine of the ten (90%) subjects who completed the trial were effectively unblinded. Seven of the ten (70%) patients had significant reductions in pain, of whom six chose to continue on opioids. One chose to withdraw because of significant side effects demonstrated in the trial, although the difference only reached the .05 level. Five of the ten (50%) showed significant improvements in sleep. In all cases those patients with improved sleep also showed significant reductions in pain. Seven out of the ten patients had significant improvements in overall measures of helpfulness (the same patients who had significant reductions in pain). Conclusions: Whilst the general principles of N of 1 trials can usefully be applied to opioid use in chronic non-cancer pain a number of limitations have been demonstrated. Firstly, in the absence of an “active” placebo blinding is limited because of side effects. Secondly, abstinence symptoms were common in spite of the multiple crossovers and the limited time on the opioid using this study design. Both issues would potentially favour positive results in the trial. Despite explanation of the possibility of abstinence symptoms, as part of the consenting process to the study, subjects commonly attributed them to unrelated illness or underlying pain. It is our opinion that multiple crossover placebo N of 1 trials are inappropriate for controlled release morphine because of the frequency and severity of abstinence symptoms.