Research has demonstrated success in reducing opioid dependence and abuse through use of buprenorphine plus psychological pain management skills. Widespread implementation of office-based buprenorphine treatment in Europe led to significant reductions in opioid use, improvements in the social and medical status of patients, reductions in overdose deaths, reduced medical morbidity, and reductions in infectious disease rates. (Barrau et al., 2001; Thirion et al., 2001; Vignau & Brunelle, 1998; Carrieri et al., 2003; Gueye et al., 2002). Buprenorphine is underutilized in the primary care management of opioid addiction due to organizational and provider barriers. Physicians other than addiction specialists in the U.S. have been reluctant to prescribe buprenorphine. They cite challenges including concerns about the complexity of the problem of pain and addiction, lack of referral sources for counseling, and lack of time to develop (Thomas et al., 2008). The prevalence of opioid abuse and dependence among primary care patients with chronic pain is unknown, and estimates vary widely. Various methods to identify aberrant drug-related behavior have been used in research and these methods could be used in clinical practice. Provider feedback regarding implementation of these methods as well as feasibility data regarding their use in a clinical setting may help address organizational and provider barriers to use of buprenorphine in primary care. A recent review concluded that the next step to address barriers to treatment of opioid use disorders in primary care will be to conduct "research and training to aid practitioners to determine the appropriate patient subpopulations and treatment protocols to achieve satisfactory outcomes" (Rosenblum et al., 2008).
This study was a two-site implementation intervention project designed to motivate primary care-based treatment of opioid abuse via buprenorphine. The study had two components: a focus group and questionnaires with primary care providers, and a medical records review aimed at estimating the prevalence of aberrant medication related behavior (AMRB) among patients taking opioids during a one year period.
The aims of the study were to:
1) Involve providers in selecting best approaches to identify AMRB, and to increase their motivation (readiness and confidence) to treat opioid abuse and dependence using buprenorphine.
2. Estimate the prevalence of patients with AMRB and compare methods for identifying AMRB likelihood. To estimate AMRB prevalence, the present study adapted methods that have been used in research studies to identify AMRB.
We hypothesized that:
1. One or more of the four AMRB identification methods will identify at least 15% of veterans
prescribed opioids as having AMRB.
2. Primary care providers will consider at least one of the AMRB identification methods as acceptable for efficient clinical us.
3. Primary care provider confidence and readiness to prescribe buprenorphine will increase after their involvement in the study focus groups.
This study was conducted at Primary Care clinics at two sites: The Michael E. Debakey Veterans Affairs Medical Center in Houston, Texas, and the VA Connecticut Healthcare System West Haven and Newington clinics.
1. A randomly selected sample of 400 total patient charts (200 per site) from primary care patients prescribed opioids for more than 60 out of 90 days between August 1, 2008 and July 31, 2009, at the Houston and Connecticut VA sites.
2. All primary care providers with or without buprenorphine prescribing certification at both sites. The Houston site had 5 PrimeCare teams with 5 physicians and 3 Nurse Practitioners or Physician Assistants per team for a total of 40 providers. The Connecticut sites combined (West Haven and Newington) have 38 MDs, 8 Advance Practice RNs, and 2 Physician Assistants for a total of 48 providers.
1. Medical and Pharmacy Records Reviews. Four previous research studies of aberrant medication related behaviors (AMRB) developed medical informatics methods to identify AMRB. We applied each of these schemes to attempt to estimate prevalence of AMRB and compare each schema's usefulness in clinical practice for identifying AMRB.
1.Method #1 (Adams, Plane, Fleming, Mundt, Saunders, & Stauffacher (2001). These authors flagged patients as being at risk for opioid use disorders if they had documented mental health disorders, documented substance abuse disorders, or had any Urine Toxicology (Utox) results documented as positive for use of illicit drugs or non-prescribed opioids.
2.Method #2 (Mahowald & Singh, 2005). These authors flagged charts for AMRB if they had escalations in daily opioid dose greater than two 30mg dose equivalents per day with no documented medical justification.
3.Method #3 (Reid, 2002). These authors flagged charts for AMRB if progress notes included any of the following: 1) One or more reports of lost or stolen opioids, 2) any other additional source to obtain more opioids (other physicians or clinics; street sales), 3) Requests for 2 or more early refills.
4.Method #4 (Manchikanti, Damron, McManus, & Barnhill, 2004). These authors flagged charts for AMRB if urine drug screens were positive for any non-prescribed opioid and/or any other illicit drug.
2. Questionnaires and focus groups with primary care providers.
Provider Focus Groups and Data
Participation was so limited that provider data are likely not generalizable. Most primary care providers did not respond to repeated emails, flyers in mailboxes, and in-person efforts to recruit participation or replied that they did not have spare time in their clinical schedules to participate. Data gathered from the few who participated did providesome perspective about what they believe could increase likelihood of buprenorphine prescribing and treatment of opioid addiction in primecare. Most PCPs in Houston still preferred to refer the cases to addiction specialists as they feel overwhelmed with patient loads and responsibilities. The situation in Connecticut was somewhat more complex. Their PCPs had been instructed not to prescribe buprenorphine, and they indicated confusion about which of two programs to refer patients with opioid abuse or dependence.
Estimates of AMRB
Method #1: Not all data related to this schematic were able to be collected. This method flags patients as having the potential for AMRB if either they have a mental health diagnosis or they have any Utox screens positive for illicit drug or non-prescribed opioid use. We were unable to pull Utox data from national databases. We were able to pull mental health and substance use disorder diagnoses for all 400 patients. We found that 210 (52.5%)of 400 patients had some MH diagnosis. If physicians used this method, up to 52% of their patients taking opioids could be flagged for further assessment of opioid misuse. This may be an inefficient method for identifying AMRB.
Method #2: We were unable to complete the process of converting opioid doses and to codeine equivalents and then identify patients with escalations in dose greater than two 30mg equivalents. As such we were also not able to identify specific dates of dose escalations for the purpose of investigating progress notes associated with that date. Even if we had the dose escalations calculated and the dates of escalations, it may not be possible to match progress notes to dates of dose escalations. One barrier is that progress notes and medication dose data come from two separate databases. We ran out of time to complete further work on this part of the study. In clinical practice, monitoring opioid dose escalations and the reasons for them is very important. Unjustified dose escalation may be a helpful indicator of AMRB.
Method #3: We were able to obtain progress notes and conduct text searches. Only two of all 400 charts were flagged as containing evidence of AMRB through this method. This could mean that in clinical practice, reasons for refill requests are not being obtained, or documentation of this type is not being completed on a regular basis. It could also be the case that documentation of this type is in pharmacist notes and not progress notes (we did not investigate pharmacist documentation). It is also possible that not many patients are requesting early opioid refills or extra prescriptions to replace lost or stolen medication.
Method #4: We were not able to estimate AMRB with this method because we were not able pull Utox data from national databases. We were not able to get help from local IT data personnel for pulling the data locally.
Although the study was not completed, the few primary care providers who participated did indicate verbally that participation had been informative and they valued being involved in the process. The chart reviews and efforts to estimate AMRB, although not complete, revealed that national data pulls of the type of information needed to flag patient charts for AMRB is likely not possible. Investigation of this data also revealed that improved documentation related to opioid prescribing must be implemented in order to monitor for AMRB.
External Links for this Project
None at this time.