A recent trial that found amitriptyline had only minimal benefit for low back pain still led its authors to call for its use, a likely byproduct of the opioid crisis but a correct conclusion nonetheless.
The authors noted that the drug may be worth considering, especially if the alternative is opioids, despite the nearly uniformly negative outcomes in the randomized, controlled trial. (JAMA Intern Med 2018 Oct 1; doi: 10.1001/jamainternmed.2018.4222; http://bit.ly/2RlaGYY.)
The Australian authors randomized 146 patients ages 18-75 with chronic low back pain present for more than three months to 25 mg of amitriptyline or 1 mg of benztropine, an active comparator selected to mimic the adverse events of amitriptyline without any effect on pain. A total of 118 participants (81%) completed the study, with 12 percent withdrawing due to medication adverse events. The primary outcome measure was level of pain intensity measured at six months using a 100 mm visual analog scale, with secondary outcomes of disability, global improvement, general health status, depression, and fear of movement and (re)injury.
Treatment with amitriptyline did not result in a greater pain reduction at three or six months, nor was there a statistically significant difference in any outcome except for a mild improvement in disability at three months. Stunningly, in reporting their results, the authors argued for the drug's effectiveness: “Amitriptyline may be an effective treatment for chronic low back pain,” particularly in an opioid-sparing strategy, and they called for larger trials with dose-finding and escalation.
Despite the negative results, these authors were almost certainly correct despite assertions that in no way were backed by the data they presented. Amitriptyline, relegated to the dark corners of emergency medicine after decades of use and misuse, has a larger role to play in managing pain in the emergency department.
Relief v. Side Effects
The use of tricyclic antidepressants (TCAs) in treating pain is not new, with one systematic review in 1996 identifying a number-needed-to-treat as low as 2.9 for patients with postherpetic neuralgia or atypical facial pain. The same study extensively argued the heterogeneity of pain, suggesting that compared with placebo, 30 percent of patients with painful conditions would have significant relief—at the cost of an equal number of patients suffering medication side effects ranging from minor to severe. (Pain 1996;68[2-3]:217.) Controversy continues about whether the analgesic effect of TCAs can be separated from their effects on mood, and the literature continues to emerge supporting their use across a wide range of painful conditions.
A 2015 Cochrane Review supported previous data, finding significant pain relief in 25 percent of neuropathic pain patients at the cost of a similar proportion reporting at least one adverse event. (Cochrane Database Syst Rev 2015 Jul 6;; https://bit.ly/2qkjvH1.) A separate meta-analysis investigating amitriptyline's effectiveness in fibromyalgia identified an essentially identical response. (Cochrane Database Syst Rev 2015 Jul 6;; https://bit.ly/2ETSC75.) In beautiful Bayesian prose, the authors wrote, “Amitriptyline probably does give really good pain relief to some people...but will not work for most people.” They went on to argue that “amitriptyline should continue to be used...but we should be cognizant of the fact that only a small number of people will achieve satisfactory pain relief.”
The incredible perceptiveness of these statements strikes at the heart of the dichotomy of our Australian investigators' recommendations. The trials we perform, study, and interpret are designed to evaluate effectiveness across a wide population and by necessity gloss over the granularity and individuality that can make a profound difference at the bedside. Recognition of these limitations and synthesis of emerging data with clinical experience, patient values, and collated literature allows a more mature application of the information—the evidence-based, patient-centered approach that we laud and toward which we strive.
Consider the clinical condition of chronic low back pain, the population investigated by the Australian researchers. These patients are frequently seen in the ED, and can suffer from a catalog of agonizing precipitants—sprains, strains, disc pathology, radiculopathy, and stenosis, to name a few. It is unreasonable and unfair to call upon a single agent to beat back a horde of painful pestilence, and trial upon trial has proven that the diverse causes of back pain are refractory to intervention, with orphenadrine, methocarbamol, diazepam, cyclobenzaprine, and even oxycodone/acetaminophen all falling short in large investigations. (Ann Emerg Med 2018;71:348; Ann Emerg Med 2017;70:169; http://bit.ly/2JsH2hT; JAMA 2015;314:1572; http://bit.ly/2JsH2hT.) But interventions often succeed when confined to a narrow and directed population. (Spine 1997;22:2323; http://bit.ly/2qlDkOw.)
Pain is a heterogenous, complex, and dynamic clinical condition that requires an informed but tailored approach. Caution and skepticism remain the watchwords of safe implementation, but acknowledgement of the limitations in the questions we ask fosters a mature synthesis of the literature with experience, and allows the best outcomes for patient and physician.
The role of amitriptyline in treating pain remains unchanged with the publication of this latest RCT, but to borrow from our pragmatic Cochrane authors from 2015: “Amitriptyline probably does give really good pain relief to some people,” but won't work for everyone. It merits a place in every emergency physician's analgesic armamentarium, but we should remain aware of its limitations and the likelihood that only a small number of our patients given this drug will achieve satisfactory pain relief.