Benzodiazepine Prescription Patterns After Mild Traumatic Brain Injury in U.S. Military Service Members

ABSTRACT

Introduction

Clinical practice guidelines (CPGs) and clinical recommendations (CRs) are developed to aide and guide providers in treating a variety of conditions, including traumatic brain injury (TBI). There is little knowledge on the impact that CPGs and CRs have on provider practice. One TBI recommendation that was able to be tracked in medical record codes was the use of benzodiazepines (BZD). Because of potential for misuse, diversion, addiction, cognitive impairment, and brain healing interference, the DoD and Department of Veterans Affairs (VA) jointly discourage prescribing BZD after TBI. As part of an effort to look at translation of CPG guidance into clinical practice, our objective was to examine the issuance of BZD prescriptions, including dose, type, and prescribing provider, prescribing setting, and primary diagnosis at issuance among U.S. service members with mild Traumatic Brian Injury (mTBI).

Materials and Methods

Using DoD data sourced from the Comprehensive Ambulatory/Professional Encounter Record (CAPER) databases of the Military Health System (MHS) Medical Data Repository (MDR), we identified all U.S. service members with a first lifetime diagnosis of mTBI from October 1, 2015 to September 30, 2016. Data on prescriptions issued to this group during a period of active treatment for a mTBI were obtained from the Pharmacy Detail Transaction Service (PDTS) databases of the MDR and identified BZD prescriptions based on the American Hospital Formulary Service (AHFS) therapeutic classification system. We validated coding assumptions through structured review of the clinical record contained within the Armed Forces Health Longitudinal Technology Application (AHLTA) of 30 randomly selected cohort members.

Results

Among U.S. service members, 4.5% filled a BZD prescription while under active medical treatment for a recent mTBI. These service members were more likely female and older when compared to their counterparts not prescribed BZD. Among service members under active treatment for mTBI during the study period, 52.6% (n = 7,935) filled a prescription; of these, 8.5% (n = 676) filled a BZD prescription. Of U.S. service members filing a BZD prescription while undergoing active treatment for an mTBI, 64.6% (n = 437) filled prescriptions for BZD and antidepressants, 54.9% (n = 371) filled prescriptions for both BZD and NSAIDS, and 42.2% (n = 285) concurrently filled prescriptions for BZD and opioids.

Conclusions

This effort to examine the translation of CPG recommendations into practice through evaluation of medical record data indicates that providers are prescribing BZD to patients under active treatment for an acute mTBI. The mTBI CPG recommends that the BZD class of medications be avoided in patients healing from brain injury. However, the team recognizes there are confounding factors that may impact the medications that are prescribed for patients with mTBI. Additional work to understand how CPGs and CRs are received and utilized by providers may elucidate opportunities to close the gap between clinical practice guidance and clinical practice.

INTRODUCTION

In the United States, Traumatic Brain Injury (TBI) has long been considered a significant public health issue.¹ TBI, defined as “a bump, blow or jolt to the head or a penetrating head injury that disrupts the normal function of the brain,” contributes to deaths and is a major cause of long-term disability.² In 2010, approximately 2.5 million TBIs occurred among civilians, leading to 280,000 hospitalizations and the death of more than 50,000 people.³ Among hospitalized TBI survivors, more than 40% developed long-term disabilities, including functional limitations, cognitive complaints, and decreased physical and psychosocial health.⁴ TBI is also a health issue among U.S. Service members. Since 2000, an estimated 472,785 service members have been diagnosed with at least one TBI while participating in the Tricare system.⁵

In 2016, a DoD and Department of Veterans Affairs (VA) working group published guidelines for clinical practice (CPGs) for the treatment of mTBI.⁶ The guidelines grouped twenty-three recommendations into four categories: diagnosis and assessment of mTBI, co-occurring conditions, mTBI treatment, and mTBI setting of care. While treating mild TBI, the joint VA-DoD guidelines recommend a general medication review and to avoid medications that can cause confusion, such as “benzodiazepines, lithium and anticholinergic agents,” or dizziness, such as “stimulants, benzodiazepines, tricyclics, monoamine oxidase inhibitors, tetracyclics, neuroleptics, anticonvulsants, selective serotonergic agonists, beta blockers and cholinesterase inhibitors.” Under the category of co-occurring conditions, the management of sleep disorders co-occurring with mTBI, the guidelines instruct clinicians to avoid benzodiazepines (BZD) after an mTBI and recommend non-pharmacologic alternatives. The guidelines cite risks for dependency, worsening of other persistent symptoms such as cognitive changes and decision-making ability, and increased risk to worsen comorbid conditions when recommending clinicians avoid the administration, either by initiation or continuation, of BZD in patients being treated for a mild TBI.

After sustaining an mTBI, the most common post-TBI diagnoses were headache, substance use disorder, post-traumatic stress disorder, and sleep disorder.⁷ In absence of TBI, treatment plans for these diagnoses could include a BZD prescription. Given the potential for addiction and brain damage risks, BZDs are recommended for short-term use only.⁸ However, among BZD-users surveyed in a past study, the most common use duration at time of survey is often greater than 1 year.⁹

Although evidence against BZD pharmacotherapy after a TBI is somewhat limited in human models, existing evidence and physiological modeling suggests the avoidance of BZD use in the TBI population. Rationale for recommending avoidance of BZD after TBI includes the drug class’s potential harmful effect on neuronal recovery.^10–12 The intent of this study was to determine brain injuries if clinician behavior and translation of CPG information into clinical practice could be evaluated by medical record data. Given the risks associated with use of BZDs among those under active recovery from brain injury, we sought to examine its use among service members newly diagnosed with mTBI.

METHODS

We identified U.S. service members who sustained a diagnosis of mTBI from the Comprehensive Ambulatory/Professional Encounter Record (CAPER) databases of the Military Health System (MHS) Medical Data Repository (MDR), using the International Classification of Diseases 10th Revision, Clinical Modification (ICD-10 CM) codes for mTBI found in any diagnosis position (Supplementary Appendix 1). We qualified the first mTBI-coded medical encounter that prompted the service member’s inclusion in the study as the “index encounter.” In order to ensure that service member’s first mTBI encounter occurred during the study period (October 1st, 2015, and September 30th, 2016), we included only those without a medical encounter containing an ICD-10 CM diagnostic TBI code in the previous 90 days leading to the index encounter. The use of BZD in cases of resolved mTBI is not contraindicated; thus, we included only instances of mTBI under active treatment. We defined a period of active treatment as an instance of a minimum of two TBI-related medical encounters no greater than 60 days apart after a washout period of 6 months. We defined our period of interest as the time in which the service member was actively receiving treatment for a newly sustained mTBI.

We obtained data on the prescriptions filled during active treatment for an mTBI from the Pharmacy Detail Transaction Service (PDTS) databases of the MDR and defined BZD prescriptions based on the American Hospital Formulary Service (AHFS) therapeutic classification system, including Anxiolytics, Sedatives, and Hypnotics (282,408) and Anticonvulsants (281,208). See supplementary materials for full list of diagnostic ICD-10 codes used for case identification and for a diagramed flow of the study cohort identification process (Supplementary Materials, Figure 1). We obtained prescription data for the period of active TBI treatment. Prescription data within the PDTS does identify the encounter in which the prescription originated. Benzodiazepines are a Schedule IV substance. Their prescriptions must be renewed at least every 6 months and cannot be refilled more than five times. Providers may elect to monitor more frequently than every 6 months. PDTS data identifying the encounter at which the prescription was written were used in conjunction with MDR data to reference the diagnostic codes assigned at the encounter at which the benzodiazepine prescription was written.

VALIDATION

After cohort determination, we undertook a structured review of 30 randomly selected cohort members to explore three validation questions. The first question confirmed the case definition, the second confirmed cohort member exposure, and the third served as a proxy for clinical documentation integrity. Results from the structured review confirmed all 30 patients were diagnosed with an mTBI and received medical care during the study period. Likewise, we confirmed patients identified as having filled a BZD prescription via pharmaceutical transaction data had records of BZD prescription in their clinical notes. Only one of 30 individuals (3.3%) had a first-time diagnosis for mTBI without documentation of a date of injury. See supplementary materials for a tabled validation result (Supplemental Materials Table S1).

RESULTS

Table I shows the demographic distribution of the study surveillance population. U.S. service members who filled a BZD prescription while undergoing active treatment for an mTBI were more likely to be female (16.9% versus 12.1%, P < 0.05) and older than those who filled a prescription for any medication other than a BZD (median age range 35–44 versus 25–34; P < 0.05). We did not find a statistically significant difference among branches between BZD and non-BZD prescription patterns during a service member’s mTBI treatment.

The most common principal diagnoses assigned to BZD prescribing encounter are reflected in Table II. Cumulatively, neuromusculoskeletal disorders including TBI (13.2%), skull and face fractures (5.2%), and neuromusculoskeletal complaints related to the back and spine (10.5%) were the primary diagnosis on record for 28.9% of medical encounters in which the clinician issued a BZD prescription. Behavioral health diagnoses cumulatively accounted for the primary diagnosis of 33.3% of the same measure.

Table III lists prescriptions issued to acute mTBI patients within a 60-day period from their identifying “index mTBI encounter.” Over half (52.6%) of the surveillance population filled any class of prescription, of which 676 (4.5%) filled a BZD prescription. Among patients filling a BZD prescription, 64.6% (n = 437) filled multiple prescriptions for BZDs and antidepressants, 54.9% (n = 371) filled prescriptions for both BZDs and NSAIDS, and 42.2% (n = 285) concurrently filled prescriptions for BZDs and opioids. Among cohort patients filling a BZD prescription (n = 676), we matched 606 of these cohort patients to the prescribing provider, as well as clinic and appointment classification of the care setting where the BZD prescription was issued. Of these 606 matched individuals, 42.8% received their BZD prescription from a mental health designated clinical facility during the acute treatment phase after sustaining their mild TBI.

For the purposes of this study, the acute treatment phase after sustaining mild TBI phase was defined as either the first 60 days after sustaining the mTBI or the timeframe between the first care appointment and the last care appointment for the service member’s acute mTBI, if no further follow up was received for their injury. Mental health care practitioners made up 40.4% of the prescribing providers identified (Table IV).

DISCUSSION

From 2013–2014 data, the CDC reported that 36.5% of Americans aged 18 to 44 used at least one prescription drug in the preceding 30 days.¹³ Among U.S. service members diagnosed with mTBI, the percent receiving a prescription medication was 52.6%. The rate of prescription drug fills in the surveillance population may be explained by potential differences in access to care and care utilization habits of U.S. service members versus the general U.S. population. Through military health care, service members can be prescribed over the counter (OTC) medications and have them filled free of charge, which would also expand prescription count in the surveillance population, especially in pharmaceutical categories for which OTC equivalents are common, such as anti-inflammatories. Additionally, the potential for exposure to occupational hazards ultimately leading to a clinical condition requiring pharmaceutical treatment may be higher in a military population. A more apt comparison would be the prescription utilization patterns of civilians who had sustained an mTBI within 60 days after their injury.

There is no identified safe dosage of a BZD for individuals under active treatment for TBI. We observed receipt and fill of a BZD prescription in 4.5% of service members under active treatment for mTBI in the 12-month study period. This observation represents a departure from current DoD/VA CPGs on the medical management of mTBI.¹⁴ Of all U.S. service members who sustained an mTBI in our study, it’s notable that the majority were on at least one pharmaceutical medication, and especially notable that 31.1% were on an antidepressant of any type. Additional exploration of prescription medications already initiated before TBI, especially centrally acting medications and medications prescribed to treat behavioral health conditions, may be of interest to those seeking to understand the role of pre-existing behavioral health conditions in medications prescribed. Further study on health care utilization patterns, including pharmaceutical and non-pharmaceutical interventions, which support the function of the service member with behavioral health and TBI diagnoses, in conjunction with measures of that service member’s level of functioning, is needed.

When demographic characteristics of service members who were prescribed and filled a prescription for BZD while under active treatment for an mTBI were compared to those who did not fill a prescription, statistically significant differences between the groups were found on age and sex, but not for branch, indicating that no one branch had a higher rate of BZD prescribing. The relationship of sex to provider–patient interaction and the proclivity of a patient to feel, recognize, or express symptoms that may warrant a provider to prescribe a benzodiazepine are complex and deserve further exploration. Research on sex contribution to the fill rate of inappropriate prescriptions is lacking; however, a study of older adults in British Columbia found women were at greater risk of filling an inappropriate prescription than were male counterparts, even after adjusting for other socioeconomic factors.¹⁵

We found neuromusculoskeletal conditions to be the primary diagnosis in 28.9% of appointments where a BZD script was issued to a service member in active treatment for mTBI. While a portion of clinical providers may be electing to treat a neuromusculoskeletal injury with benzodiazepines, CPGs issued jointly by the DoD and the VA for lower back pain, for all types of chronic pain, for chronic insomnia and obstructive sleep apnea, for management of patients at risk for suicide, and for posttraumatic stress disorder and acute stress disorder recommend against the use of BZDs for the treatment of these conditions.^16–20 The joint VA-DoD CPGs on the management of substance use disorders cover the treatment of SUD for multiple agents. Therefore, it advises on how to treat benzodiazepine misuse, cautions against the combination of benzodiazepine and other CNS-sedating agents, and advises against the utilization of BZD in the treatment of mild and moderate alcohol withdrawal in patients for whom risk, like the potential abuse liability, outweigh benefits. This document does endorse BZD for one specific application: the medically monitored treatment of moderate to severe alcohol withdrawal.²¹ Patients at risk for seizures may also be prescribed with a BZD as an anticonvulsant agent.²² Likewise, BZDs are currently part of common hospital management of Alcohol Withdrawal Syndrome.^23,24 However, only 3.7% of those found in receipt of a BZD after a TBI had an alcohol-related disorder at prescription originating encounter, and none had a seizure diagnosis. Inpatient prescribing for any reason at all made up only 1.0% of BZD prescriptions to mTBI patients.

The most common provider specialty prescribing a BZD to a service member with recent TBI was a mental health provider (42.8%) and the most common first position diagnosis was an anxiety disorder (18.3%) of any type. It is typically part of a prescriber’s practice to discuss recent care utilization and/or screen for recent events that might inform the provider’s treatment plan, so it’s possible that providers are aware of the recent TBI sustained while filling or refilling prescriptions. The medical records across the MHS are digitalized and are designed to allow providers full visibility of records. However, it is possible that a service member may have gaps in their digital record for unknown reasons or through a potential lag in the repopulation of the legacy electronic health record (EHR) system records to the Joint Legacy Viewer associated with the sundowning of the legacy system and adoption of MHS Genesis. Twenty-four hour, acute, or emergency room type encounters comprised only 13.9% of encounters from which a BZD prescription was issued during active mTBI treatment. Conversely, encounters that were future set, scheduled seven or more days in advance and telephone consults made up 58.8% of such encounters. These might represent a portion of U.S. service members who had an established clinical relationship with a provider. Providers may be unaware of the service member’s recent TBI or be unaware of the guidelines that guide providers not to prescribe BZD after mTBI. Both such situations could be addressed through decision support solutions at the point of care and provider training.

Defense Health Management operations develop evidenced-based clinical practice recommendations and guidelines for multiple conditions; the dissemination and operationalization of these recommendations presents a challenge in the expansive and diverse spectrum of clinical settings across military health systems. An approach that incorporates the Defense Health Agency-wide practice guidelines into clinical workflows that auto-populate a patient’s record within a chosen EHR software system at the appropriate time given a set of pre-determined, a-b test-validated criteria addresses both cumbersome tasks of dissemination and operationalization of clinical recommendations (CRs) and practice guidelines. Civilian EHR software vendors incorporate workflow modules into clinical documentation to increase care quality, care standardization, and efficiency of care delivery. Historically, MHS approximates the EHR clinical workflow feature with an after-market edition of Tri-Service Workflow (TSWF) forms available to clinicians charting within Armed Forces Health Longitudinal Technology Application (AHLTA). The TWSF forms are a valuable workaround but differ from current civilian technology in that TSWF forms work like flat Word or PDF files; smart logic and skip logic are not capabilities of the TSWF forms.

In the cases where the provider issues a BZD prescription in light of a known recent mTBI, further exploration of patterns of BZD prescribing to U.S. service members independent of a TBI might be helpful in understanding a provider’s difficulty in discontinuing a BZD when a service member sustains a TBI while under their treatment. Future studies may be conducted to explore both TBI and behavioral health outcomes in service members with matched diagnoses receiving either a BZD, a BZD-pharmaceutical alternative, or a non-pharmaceutical intervention for their diagnoses. Because of treatment crossover patterns common in behavioral health care in a non-randomized control trial environment, finding clean groups for each treatment arm could be difficult.

LIMITATIONS

Data for this surveillance study were obtained from the MDR and PDTS, systems designed for the tracking of administrative coding and billing. Study methodology relied on correct assignments of ICD-10 coding, which can vary by provider, clinic, and area. Limitations on pharmaceutical benzodiazepine usage apply to the acute healing period after a TBI. Although a washout period and a record review to check methodology were employed, the possibility remains that a portion of the sample of acute mTBI patients either did not truly sustain an mTBI or sustained one remotely enough that the guideline would no longer apply. The study did not explore benzodiazepine fill rates of the population before sustainment of the mTBI as the guidance on benzodiazepines does not make an exception for pre-existing prescriptions. However, this information may shed light on provider hesitancy to discontinue such a prescription and would be of interest in future research.

CONCLUSION

We used available military health data to compare recent medical care delivered in the military health care system to DoD/VA CPGs. We confirmed that the practice of prescribing BZDs to recently diagnosed service members diagnosed with mTBI and undergoing medical treatment occurred in 4.5% of the surveillance population despite CPGs discouraging this practice. The service members receiving a BZD prescription while undergoing treatment for TBI tend to be older and more likely to be female, like estimated prescribing practices in the general U.S. population.²⁵ The relatively high percentage of BZD prescriptions generated from mental health clinics and mental health practitioners after TBI diagnosis may reflect the prevalence of mental health disorders among U.S. service members. Further study on the tendency of military mental health care clinicians to prescribe BZDs to treat these conditions, even when such prescribing patterns are discouraged, is warranted. For the TBI community, these findings demonstrate continued use of BZDs in TBI patients with co-morbid mental health diagnoses that could be treated with alternate pharmaceutical and non-pharmaceutical treatments. Future research might address MHS-wide measurement of quality clinical practice and later, perhaps, measurement of impact of CRs and CPGs on patient outcomes. As standardization and unification of health care delivery across the MHS materializes, CRs and guidelines endorsed by the DoD could be built into upcoming DoD EHR replacement systems through diagnoses-appropriate workflows and pharmaceutical alerts.

ACKNOWLEDGMENTS

None declared.

CLINICAL TRIAL REGISTRATION

None declared.

Study makes use of no human subjects

 

INSTITUTIONAL ANIMAL CARE AND USE COMMITTEE (IACUC)

Not applicable.

INDIVIDUAL AUTHOR CONTRIBUTION STATEMENT

YA pulled and analyzed data from the MDR. LE collected data from medical records, analyzed the data and drafted the original manuscript. YA, LE, and KS designed this research, reviewed and edited the manuscript. All authors read and approved the final manuscript.

INSTITUTIONAL CLEARANCE

Institutional clearance approved.

SUPPLEMENTARY MATERIAL

Supplementary material is available at Military Medicine online.

FUNDING

Traumatic Brain Injury Center of Excellence, Silver Spring, MD

CONFLICT OF INTEREST STATEMENT

None declared.

DATA AVAILABILITY

Data available upon request.

REFERENCES

Author notes