Setting
The Comprehensive Care Practice is a primary care clinic on the Johns Hopkins Bayview Medical Center campus which is staffed by 5 internists, a nurse practitioner, and 3 internal medicine residents who share a panel of patients. The practitioners provide general primary care, with a focus of serving patients with HIV infection and/or substance use disorders.
Visits for opioid dependence occur as routine primary care visits. A more detailed description of this clinic's treatment practices has been published previously [19]. Briefly, there is no uniform protocol or dosing regimen, with buprenorphine doses ranging from 2 to 32 mg, with most patients on 8-16 mg daily. Induction occurs at home or in the office and follow-up occurs weekly to monthly, based on the provider's discretion; patients are usually seen more frequently early in treatment or when there is continued substance use. Treatment is continued or discontinued based on the provider's discretion. The practice does not provide any additional onsite psychosocial services and patients are referred to community resources.
Subjects
The study included all patients who had received at least one prescription for sublingual formulation of buprenorphine from August 2003 to September 1, 2007.
Data collection
As part of a previously performed study [19], a database of all patients who received at least one prescription for buprenorphine during this period of time had already been created. Data were collected retrospectively from the patient medical records. Demographic information recorded included age, gender, type of insurance and employment status. Substance abuse history collected included substances used and history of injection drug use. We also collected data on recent (within 30 days) drug treatment, including methadone and buprenorphine maintenance. Also recorded were relevant comorbidities (hepatitis C, HIV, chronic pain and chronic psychiatric illness).
The time period after receiving the first prescription was divided into twelve monthly blocks for the purposes of outcomes analysis. Patients were considered to be in treatment for each block in which they were prescribed buprenorphine at any point. There was no fixed protocol for collection of urine drug tests, so for each month in which the patient was receiving treatment, patients were classified as "opioid-positive" or "opioid-negative". Patients were classified as "opioid-positive" if any of their urine drug tests during that month were positive for opioids (other than those prescribed), if they reported using other non-prescribed opioids, or if a urine drug test was not collected and their most recent test was positive. Patients were classified as "opioid-negative" if all urine drug tests collected during that month were negative for opioids (other than those prescribed), or if the provider decided not to collect a test and their most recent one was negative.
We utilized the Maryland Judiciary Case Search website http://casesearch.courts.state.md.us/inquiry/inquiry-index.jsp to determine whether subjects had any criminal charges and the number of cases filed against each subject. This database includes data on all criminal charges in the state of Maryland since 1991. The database provides the defendant name, city and state, date of birth, trial date, charges, and case disposition. We searched this database by each patient's name and birthdate, and recorded whether the subject had ever had criminal charges prior to initiating treatment. We tabulated the total number and types of criminal cases two years prior to the date of the first prescription for buprenorphine, and for the two-year period of time after that date. We counted cases as listed separately on the website, not individual charges; we decided to do this because many cases included multiple related charges (for example, multiple charges of drug possession with separate charges for possession of drug paraphernalia). We did not count cases that were limited to motor vehicle charges, with the exception of driving under the influence of alcohol. We classified criminal cases as "drug cases" or "other"; "drug cases" included any cases with charges of possession or distribution of controlled dangerous substances (CDS).
Statistical analysis
Bivariate analysis was used to compare demographic factors and outcomes among subjects with and without criminal charges in the two years after initiation of treatment. We also used bivariate analysis to compare the mean number of criminal cases in the two years before and after initiation of treatment. We also compared the number of criminal cases before and after initiation of treatment among subjects who remained in treatment at 12 months and those who were opioid negative for 6 or more months. Chi-square tests were used to analyze categorical variables and paired t tests for continuous variables. Wilcoxon signed ranks and Mann Whitney U tests were used to compare variables that were not normally distributed. P values less than 0.05 were considered statistically significant. Multivariate analysis was used to analyze factors associated with criminal charges after initiation of treatment. Variables with a p value < 0.1 were entered into the logistic regression model; for factors that were highly co-linear (Pearson's correlation coefficient > 0.4), only one was entered. Analysis was performed using PASW software (version 18). This study was approved by the Johns Hopkins Institutional Review Board.