PMS51 GABAPENTIN, OPIOID, AND BENZODIAZEPINE PRESCRIBING IN EARLY CARE OF INDIVIDUALS WITH CHRONIC BACK PAIN: A SECULAR TRENDS ANALYSIS

Evidence suggests gabapentin does not work better than placebo for chronic back pain; however, the extent to which these medications are used for this diagnosis is unknown and the subject of this study. A secular trends analysis using claims data from a commercially insured population (2008–2017) was conducted. Individuals with chronic back pain indicated by two diagnoses at least 30 days apart were included. Six months of continuous enrollment before (baseline) and following (follow-up) the first diagnosis was required. Individuals with diagnosis for seizure, post-herpetic neuralgia, restless leg syndrome, fibromyalgia, diabetic neuropathy and neuropathic pain with spinal cord injury or opioid, benzodiazepine, or gabapentin use in the baseline were excluded. Cochran-Armitage test was used to identify trends. 308,207 individuals with chronic back pain were included. Among them, 7,998 were prescribed gabapentin, 72,734 were prescribed opioids, and 808 were prescribed benzodiazepines within six months of diagnosis. The percentage of individuals prescribed gabapentin increased from 1.2% in 2008 to 4.9% in 2017, with a peak of 5.6% in 2016. Cochran-Armitage test was statistically significant (p<0.001) and the test statistic indicated an increasing trend. Opioids were prescribed to 21.8% of individuals in 2008 and 15.1% in 2017; however, the trend was not significant (p=0.256). Opioid use increased from 2008 through 2013 (p<0.001) and decreased from 2013 to 2017 (p<0.001). Benzodiazepine use had a small decreasing trend, with 0.3% and 0.2% use in 2008 and 2017, respectively (p=0.003). There was no consistent trend for concomitant use of gabapentin with opioids or benzodiazepines. A notable increase in gabapentin use was observed in a population of chronic back pain patients. Given the lack of effectiveness of gabapentin, this trend is concerning. Future study adjusting for patient-level covariates is warranted.