Outpatient Off-Label Gabapentin Use for Psychiatric Indications Among U.S. Adults, 2011–2016

Gabapentin is an anticonvulsant medication that received approval in 1993. As of 2019, gabapentin was ranked the fourth most prescribed medication in the United States (1, 2). Gabapentin was initially approved by the U.S. Food and Drug Administration (FDA) as adjuvant therapy for partial-onset seizures with and without secondary generalization, first for adults and subsequently for pediatric patients (3). It was later approved for adults for postherpetic neuralgia (PHN). Notwithstanding these FDA-approved indications, there are more than 20 documented off-label indications for which gabapentin is also prescribed, and this is a key factor in its recent surge in use (4, 5). Upon initial FDA approval, off-label experimentation with gabapentin commenced for neuropathic pain, bipolar disorder, and migraine prophylaxis (6). The sustained upward trend in off-label prescribing of gabapentin is concerning, considering the minimal high-quality evidence for many off-label uses.

Recent attention to off-label use of gabapentin is largely attributable to a renewed interest in this medication as a chronic pain treatment in the wake of the opioid epidemic. However, recent case reports suggest that gabapentin is increasingly being prescribed for psychiatric indications, predominantly anxiety and depression (7, 8). Conflicting evidence suggests that gabapentin’s use in psychiatry has decreased over time, despite renewed interest in its potential for alcohol use disorder, posttraumatic stress disorder (PTSD), and obsessive-compulsive disorder (OCD) (9–12). The pharmacological evidence cited for gabapentin’s use in various psychiatric disorders is based on an incomplete understanding of its mechanism of action. It was once believed that gabapentin had anxiolytic properties and increased gamma-aminobutyric acid levels in the brain, but evidence remains inconclusive (13). However, gabapentin appears to affect calcium channels, and its mechanism is hypothesized to reduce neurotransmitter release in the brain and spinal cord (13).

In December 2019, the FDA issued a safety warning advising against the combined use of gabapentin with central nervous system depressant (CNS-D) drugs, many of which are used in psychiatric treatment regimens (14). A range of short-term and long-term adverse outcomes can result from concomitant use of CNS-D drugs and gabapentin that jeopardize the previously touted favorable safety profile of gabapentin (e.g., breathing problems, sedation, and respiratory arrest) (15, 16). There is a clear need to better understand off-label gabapentin use for psychiatric conditions in order to inform prescribing and improve patient safety via care coordination between primary care and psychiatry providers. Therefore, this study aimed to examine off-label use of gabapentin for psychiatric indications in the outpatient setting and to identify concomitant use of CNS-D prescription drugs among users of prescribed gabapentin for a psychiatric indication.

Methods

Study Design and Data Source

This was a descriptive cross-sectional study using 2011–2016 data from the National Ambulatory Medical Care Survey (NAMCS), a publicly available deidentified data source provided by the National Center for Health Statistics (NCHS). The NAMCS is a survey of visits to nonfederally employed, office-based physicians primarily engaged in direct patient care (17). Data are collected by a random sample of approximately 30 visits annually from a 1-week reporting period and constitute a nationally representative sample of ambulatory care office visits. Medication data represent documentation of drugs that were ordered, supplied, administered, or continued at the visit (17, 18). Additional information on sampling methodology, details on the survey questionnaire, and data documentation are available online (17).

Sample Selection

The NAMCS captures physician-patient encounters, and thus the unit of observation is an office visit record (18). All office visit records with gabapentin documented were included, herein referred to as a “gabapentin visit.” All generic and brand-name forms of gabapentin were included. Combination drugs containing gabapentin (e.g., amitriptyline-baclofen-gabapentin) were excluded, because they are compounded topical preparations without an on-label indication and would be inappropriate for the off-label indications of interest. Office visit records that documented gabapentin enacarbil were excluded because of its differing extended-release formulation and on-label indication for restless legs syndrome, which is not approved for the original formulation.

The analysis was restricted to gabapentin visits of adults ages 18 and older. Although gabapentin’s approval for partial-onset seizures includes patients ages 3 and older, there were fewer than five pediatric patients with this diagnosis, and thus the age restriction was applied in accordance with NCHS standards (3, 19). Finally, records were excluded when the primary diagnosis code was missing or blank.

Gabapentin Use

The 2011 data have up to eight variables available that document the number of different medications that were ordered, supplied, administered, or continued by a patient attending the office visit. Data for 2012 and 2013 increased to a total of 10 medication variables, and the 2014–2016 data further increased to 30. To ensure consistency across all survey years, gabapentin use was examined if it was included in the first eight medications listed for a visit for all survey years. Drug codes are based on the Lexicon Plus, a proprietary database of all drugs available in the United States (20). (A list of specific drug codes is provided in an online supplement to this article.)

On-Label and Off-Label Indications

Data for 2011 to 2013 have three variables available that document a patient’s medical diagnoses at the office visit and are recorded as ICD-9-CM diagnosis codes. Data for 2014 to 2016 have five variables available for diagnosis codes, and the 2016 data use the ICD-10-CM diagnosis codes. To ensure consistency across all survey years, indications were examined with the first three diagnosis code variables listed for all survey years. ICD-9-CM and ICD-10-CM codes in any position were used to determine on-label and off-label indications.

On-label indications were defined as gabapentin’s FDA-approved indications for partial-onset seizures and PHN, which were determined by ICD-9-CM and ICD-10-CM codes based on past studies that assessed these conditions (21–23). Collectively, these are referred to as “on-label gabapentin visits.”

The remaining gabapentin visits that did not list an on-label indication were examined, and the records are referred to as “off-label gabapentin visits.” To further determine which of the off-label gabapentin visits listed a psychiatric condition for an off-label indication, indicator variables were created for prominent off-label psychiatric uses on the basis of the literature and recent case reports. The psychiatric indications of interest were anxiety, depression, bipolar disorder, alcohol dependence or withdrawal, OCD, PTSD, and schizophrenia spectrum disorders (12, 24). (The specific ICD-9-CM and ICD-10-CM codes used for on-label and off-label indications are provided in the online supplement.)

CNS-D Prescription Drugs

Concomitant use of CNS-D drugs was defined as at least one CNS-D medication documented in the record as a current or new prescription at the same gabapentin visit. The selected drug classes examined and the specific drugs included within each class were guided by the 2019 FDA Drug Safety Communication as well as a 2016 communication regarding risks with CNS-D drugs (14, 25). The drug classes included were antidepressants, antipsychotics, first-generation antihistamines, benzodiazepines, sedatives-hypnotics, opioids (including opioid-combination products), and skeletal muscle relaxants.

Demographic Characteristics and Covariates

The NAMCS includes patient demographic information, insurance coverage type, total number of chronic conditions, and other clinical characteristics collected at the office visit, such as vital sign measurements. Information on the health care providers surveyed is characterized by provider type, provider specialty, the practice’s geographic region, and whether the provider is the patient’s primary care physician (PCP). For this study, we described patient characteristics with age, sex, race, geographic region (i.e., Northwest, Midwest, South, or West), insurance coverage, and number of chronic conditions (separate from diagnosis codes). The NAMCS imputes values for certain variables that have a high percentage of missingness; for this analysis, only race was used with imputed values. Providers were described by specialty (i.e., primary care, neurology, psychiatry, or other specialty), PCP status of the provider, and the provider type (i.e., physician or other provider); more than one provider can be seen during a single visit.

Statistical Analyses

The NAMCS utilizes a complex survey design, in which strata are determined first by physicians and second by number of visits; strata were accounted for in statistical analyses by using survey procedures and sampling weights to obtain national-level estimates. Data were analyzed from 2011 to 2016 for each survey year and in aggregate across all years. Descriptive statistics were used to examine gabapentin utilization, on-label and off-label indications, and patient and provider characteristics with frequency counts, sample means, and proportions. Proportions were used to calculate the number of gabapentin visit records overall, with the available number of NAMCS records that met inclusion criteria for each survey year as the denominator. The proportions of on-label gabapentin visits and off-label gabapentin visits were calculated, with the total number of gabapentin visits (2011–2016) as the denominator. Frequency counts were used to describe the psychiatric off-label indications of interest. A sensitivity analysis was conducted to include all medications and diagnoses collected (i.e., 10 medications for 2012–2013, 30 medications for 2014–2016, and five diagnoses for 2014–2016). This study was exempt from institutional review board approval, because NAMCS data are publicly available and deidentified. All analyses were conducted with SAS, version 9.4.

Results

Between 2011 and 2016, a total of 249,282 office visit records were available in NAMCS. Of these, 205,417 records initially met the age and primary diagnosis code inclusion criteria. After applying the gabapentin utilization criteria, a total of 5,732 records (nationally representative weighted estimate of 129.6 million visits), or gabapentin visits, were available for analysis. Characteristics of NAMCS visit records were compared between nongabapentin visits and gabapentin visits (Table 1). The gabapentin visits differed from nongabapentin visits in age, insurance type, number of chronic conditions, and provider specialty. For the gabapentin visits, the patient was likely to be older, to have Medicare insurance coverage, to have more chronic conditions, and to visit a provider specializing in neurology.

Characteristics of the gabapentin visits were similar across years (Table 1). For these visits, patients’ mean age was 59.8 years, with the greatest proportion between 45 and 64 years (43.2%). Most patients were White (86.7%) and female (61.5%). The largest proportion lived in the South (35.4%), followed by nearly equal proportions in the Midwest (26.2%) and West (26.7%). In terms of insurance, the largest proportion were Medicare beneficiaries (43.3%), followed by private insurance (33.3%). Most (61.4%) had two or fewer chronic conditions, and nearly half (47.9%) saw a provider who specialized in another specialty (i.e., not primary care, neurology, or psychiatry), followed by primary care (34.9%). For 30.9% of the visits, the provider was the patient’s PCP.

For the gabapentin visits, the proportion of on-label indications for gabapentin was exceptionally low for all 6 years, with <1% having an on-label diagnosis of either partial-onset seizures or PHN (Table 2). Survey year 2014 accounted for the lowest proportion with only 0.4% of visits having an on-label indication for gabapentin, and 2012 and 2016 had the highest proportion, with 0.8% of visits. Across all years, the proportion of gabapentin visits with any off-label indication was 99.4%. This proportion remained unchanged in the sensitivity analysis for 2011–2016, with only survey year 2016 decreasing slightly to 98.8% (see online supplement).

For the off-label psychiatric indications of interest, the most frequent diagnoses across all years were depressive disorders (5.3%), followed by anxiety disorders (3.5%) and bipolar disorders (1.8%) (Table 3). By year, anxiety disorders increased in frequency from 2011 to 2016 with the exception of 2015. For depressive disorders, the frequency decreased from 2011 to 2014 but then increased slightly from 2014 to 2016. The frequency for bipolar disorders remained relatively stable, with 2015 having the highest (2.4%) and 2016 the lowest (1.2%). Other psychiatric indications of interest were less frequent, with 0.7% of patients in off-label gabapentin visits having a diagnosis of alcohol dependence or withdrawal; all remaining indications (schizophrenia spectrum disorders, PTSD, and OCD) accounted for 1.3% of visits.

Provider characteristics for the off-label gabapentin visits were similar to those in the overall sample (Table 3). In total, off-label gabapentin visits were mostly attended by a provider who specialized in another specialty (48.1%), followed by primary care (34.9%)—the latter were primarily physicians (98.1%), but more than one provider could be seen at a single visit.

Concomitant use of gabapentin and CNS-D drugs was similar for the total sample of gabapentin visits and for the off-label gabapentin visits (Table 4). For the off-label gabapentin visits, only 3.6% of visits listed gabapentin as the only prescription medication. Most off-label gabapentin visits (90.6%) listed two or more prescription medications. For nearly one-third (33.1%) of the off-label gabapentin visits, one concomitant CNS-D prescription drug was listed, and one-fourth (25.3%) of the visits listed two or more. The most frequently utilized CNS-D drug classes were antidepressants (24.3%), opioids (including opioid-combination products) (22.9%), and benzodiazepines (17.3%). In contrast, for the on-label gabapentin visits, a smaller proportion of visits listed concomitant prescription medications (82.9% for two or more such medications) and CNS-D prescription drugs (28.6% for one CNS-D drug and 17.1% for two or more), and the most frequently utilized CNS-D drug class was benzodiazepines (25.7%), followed by antidepressants (17.1%) and opioids (14.3%).

Discussion

In this study assessing outpatient use of off-label gabapentin for psychiatric indications, we found a remarkably low proportion of visits with on-label utilization, consistent with earlier literature in the United States and internationally (26–29). The most common diagnoses in this study were depressive disorders, followed by anxiety disorders and bipolar disorders, which supports case reports suggesting an increase in gabapentin use for these disorders despite limited high-quality evidence. The evidence cited for use of gabapentin for anxiety disorders is limited to randomized controlled trials (RCTs) for situational anxiety, including perioperative anxiety, but there have been no RCTs to date examining gabapentin’s efficacy as monotherapy or adjunctive treatment for generalized anxiety disorder or for depressive disorders (12). Although the NAMCS data did not allow us to ascertain the psychiatric diagnosis for which gabapentin was specifically prescribed, the results suggest that gabapentin was in fact prescribed at least adjunctively in the presence of a depression or anxiety diagnosis, consistent with our findings of high proportions of concomitant antidepressants and benzodiazepines. Patients with chronic pain often have secondary anxiety and depressive symptoms, and thus adjuvant use may fall into this category, particularly in light of gabapentin’s approval for PHN and its frequent off-label use for other neuropathic pain conditions (30, 31).

The relatively low frequency of bipolar disorder diagnoses in the sample of off-label gabapentin visits suggests that use of gabapentin as a mood stabilizer has declined, which corresponds with more recent psychopharmacology literature concluding that gabapentin’s mood-stabilizing effects are minimal to negligible (13, 32). We found few instances of gabapentin use for other psychiatric indications, which concurs with a greater likelihood of off-label experimentation being initiated by a psychiatrist rather than by a PCP for these conditions. Furthermore, the time frame of the data analyzed may explain why few gabapentin visits for conditions other than anxiety and depressive disorders were observed—particularly for alcohol use disorder—because the American Psychiatric Association (APA) did not include gabapentin in its guidelines until 2018 and graded the evidence as low (33). A recent systematic review concluded that gabapentin for alcohol use disorder may be safe only as an adjuvant but not as a monotherapy and that there is insufficient evidence to support its use as a treatment for other substance dependencies or for depression, OCD, PTSD, and schizoaffective disorder (24). The largest proportion of off-label gabapentin prescribing was initiated by a PCP, as evident in our findings. PCPs are the first point of contact for many patients, prior to referral to a psychiatrist or other mental health specialist. Compared with psychiatrists, PCPs may initiate gabapentin more often because of the widespread perception among PCPs that it is relatively safe (34). During the 6 years covered by the data, use of outpatient mental health services increased overall, whereas the number of practicing psychiatrists declined (35–37). Furthermore, disparities in provider reimbursement by Medicare and Medicaid suggest reduced access to specialty care from psychiatrists, which may have particularly affected beneficiaries of these governmental programs, who were prominent in our sample (38, 39). Thus, in some situations, PCPs may remain responsible for caring for a patient’s mental health care needs.

For the gabapentin visits, we found a high proportion of concomitant use of CNS-D prescription drugs, despite the FDA warning (14). Although this warning arose primarily because of concerns about acute adverse effects of serious breathing problems and respiratory depression, the FDA also warned that elderly patients are at an increased risk when coprescribed gabapentin and CNS-D drugs. Our sample had a large representation of older patients, which highlights the need for providers to be aware of this risk when considering whether to continue gabapentin when a patient is also prescribed a CNS-D medication. Additionally, the FDA warned of potential increased risks of misuse and abuse associated with gabapentin. As gabapentin has risen to the top five prescribed medications in the United States, a parallel rise in reports of misuse, abuse, and suicidality associated with gabapentin has occurred (40–42). Curtailing high-risk use could entail a Schedule V classification to limit prescribing quantities and to require reporting to prescription drug monitoring programs, which has been initiated in some states (43).

Furthermore, it is worthwhile to note that much of the literature on gabapentin’s use in psychiatry fails to mention that gabapentin has carried a warning for suicidal behavior and ideation in its package insert dating back to 2008 (3, 44, 45). Publications by the APA on prescribing psychiatric medications do not mention this serious adverse effect associated with gabapentin, despite its particular relevance in psychiatry practice (13, 46). Although a direct link between gabapentin and suicidality is not definitive, it is known that patients with psychiatric conditions have a higher risk of suicidal behavior, compared with the general population. These added risks warrant a reevaluation of gabapentin’s risk-benefit profile in psychiatry, specifically to prevent unintended consequences of exacerbating a psychiatric disorder. The complexity of managing psychiatric symptoms treated with gabapentin in light of these safety and efficacy concerns highlight the need for further research to identify cost-effective strategies to improve care management.

We recognize that the largest proportion of gabapentin prescribers fell into an “other disciplines” category (i.e., not primary care, neurology, or psychiatry), and these providers are likely responsible for adjuvant prescribing for anxiety or depression secondary to a chronic condition. It is not clear whether these providers have the proper training and resources to adequately treat secondary mental health conditions. However, if providers see benefit in gabapentin’s adjuvant uses, they should have the tools to manage the patient’s need, particularly when gabapentin is used concomitantly with other medications that may increase risks. Training for various practitioners providing mental health care may be warranted.

This study was not without limitations. First, the record-level nature of the data limited interpretation of findings in a patient-level context. Sampling was conducted over a 1-week reporting period and thus limited examination of visits by the same patient, although underestimation of gabapentin prescriptions was still possible. Second, we restricted analysis to the first eight medication variables and first three diagnosis variables because of variations in the data collection process for each survey year. This may have underestimated both gabapentin use and off-label psychiatric indications, although results did not change in a sensitivity analysis. Third, NAMCS data lack details on dosage, frequency, and refills, which limited the understanding of therapeutic uses. Fourth, attribution of gabapentin to the specific off-label indications was not definitive because clinical notes were unavailable. Finally, “central nervous system depressants” is a broad term, and the selected drug classes may not be exhaustive. The strengths of this study were a nationally representative sample, and the study is the first to investigate off-label gabapentin use for psychiatric indications in the primary care setting and the first to comprehensively assess concomitant use with CNS-D drugs.

Conclusions

This study examined off-label use of gabapentin for psychiatric indications and its concomitant use with CNS-D prescription drugs in a nationally representative sample of ambulatory care office visits. Less than 1% of outpatient gabapentin use was for FDA-approved indications. Depression and anxiety disorders were the most frequent off-label psychiatric diagnoses. In light of recent safety warnings from the FDA, the high proportions of concomitant CNS-D use found in this study underline the need for future research on the efficacy and safety of gabapentin prescribed off label for the treatment of patients with psychiatric symptoms or conditions.