Abstract
Objective
To compare adherence to opioid prescribing guidelines and potential opioid misuse in patients of resident versus attending physicians.
Design
Retrospective cross-sectional study.
Setting
Large primary care practice at a safety-net hospital in New England.
Subjects
Patients 18–89 years old, with at least one visit to the primary care clinic within the past year and were prescribed long-term opioid treatment for chronic non-cancer pain.
Methods
Data were abstracted from the EMR by a trained data analyst through a clinical data warehouse. The primary outcomes were adherence to any one of two American Pain Society Guidelines; 1) documentation of at least one opioid agreement (contract) ever, and 2) any urine drug testing in the past year; and 3) evidence of potential prescription misuse defined as ≥2 early refills. We employed logistic regression analysis to assess whether patients’ physician status predicts guideline adherence and/or potential opioid misuse.
Results
Similar proportions of resident and attending patients had a controlled substance agreement (45.1% of resident patients vs. 42.4% of attending patient, p=0.47) or urine drug testing (58.6% of resident patients vs. 63.6% of attending patients, p=0.16). Resident patients were more likely to have two or more early refills in the past year relative to attending patients (42.8% vs. 32.5%; p=0.004). In the adjusted regression analysis, resident patients were more likely to receive early refills (OR 1.82, 95% CI 1.26–2.62) than attending patients.
Conclusions
With some variability, residents and attending physicians were only partly compliant with national guidelines. Residents were more likely to manage patients with a higher likelihood of opioid misuse.
Introduction
Prescription opioid misuse is a significant public health problem. In 2009, nearly 15,500 intentional and unintentional deaths were attributed to prescription opioids (1). Primary care providers are the principal prescribers of opioid medications (2,3) for chronic non-cancer pain and thus serve as a major source of potentially harmful opioids. In response to the epidemic of prescription opioid misuse, the American Pain Society generated best practice guidelines for prescribing and mitigating risk of prescription opioids. The recommendations include risk stratification, controlled substance agreements and periodic urine drug testing (4).
Residents provide a substantial proportion of the care for vulnerable patient populations (5,6) in safety-net hospitals; such hospitals serve predominantly low-income and/or un-insured patients, many of whom may be at risk for prescription opioid misuse. Although resident physicians have been shown to provide higher quality of care in the outpatient setting for some chronic diseases as compared to attending physicians (7), it is not clear whether that practice extends to opioid prescribing for chronic pain. One study demonstrated higher use of contracts by residents, however no risk adjustments were made for patient characteristics (8). Because practice patterns established in residency are likely to be the basis for lifelong practice, a better understanding of these patterns is needed. Furthermore, if attending physician practices are not adherent to guidelines, it may indicate a need for education and practice changes for attendings as well as residents. This is especially important for attendings who precept residents and thus help shape resident practice.
We conducted a retrospective cross sectional study at an urban safety-net hospital comparing adherence to guidelines on opioid monitoring and prevalence of opioid misuse among patients of resident versus attending physicians. Our hypothesis was that resident physicians provide similar care to attending physicians, given that they are being trained and monitored by these same physicians.
Methods
We conducted a retrospective cross-sectional study at the general internal medicine (GIM) primary care practice of Boston Medical Center (BMC), which cares for approximately 30,000 unique patients. Data was abstracted from the electronic health record (EHR) through the institution’s clinical data warehouse. The Institutional Review Board at Boston University approved this study.
Study Sample
We identified patients age 18 to 89 years who met the following criteria: 1) one or more completed visits to the GIM practice from August 31, 2011 to September 1, 2012; 2) received long-term opioid treatment (defined as three or more opioid prescriptions written at least 21 days apart within a six-month period) for chronic non-cancer pain; (2) 3) A GIM primary care provider (physician or nurse practitioner) signed the opioid prescriptions. We excluded patients who were receiving care for cancer (except non-melanoma skin cancer) as defined by ICD-9-CM codes on the EMR problem list and three or more visits in the past year to the hematology-oncology clinic. We defined attending patients as all patients receiving two or more opioid prescriptions from either attending physicians or nurse practitioners. Patients were classified as resident patients if a resident signed at least two of their opioid prescriptions. At BMC, resident physicians do not sign opioid refills for attending physicians unless the patients are seen as part of a resident office visit. Resident physicians in Massachusetts sign their own controlled substance prescriptions and do not require a co-signature from an attending physician.
Data Collection
An experienced data analyst extracted data in the clinical data warehouse, based on data from the electronic health record (Logician) and from parameters set by the study team. The analyst removed any identifying information while creating the dataset that study team analyzed.
Opioid Treatment
We included all oral analgesic opioids, either as a single agent or in combination, topical analgesics, such as transdermal patches, and intranasal preparations. The following opioids were identified: codeine, fentanyl (oral and transdermal), hydromorphone, meperidine, methadone, morphine sulfate, oxycodone and propoxyphene. We excluded patients receiving intravenous, intramuscular and sublingual preparations of opioids, buprenorphine as well as those receiving tramadol. A prior study similarly excluded patients on tramadol (9). At the time of our data collection and analysis, the Federal Drug Enforcement Administration (DEA) did not consider tramadol a controlled substance. However, as of August 2014, tramadol will be considered a Class IV controlled substance (10). Methadone taken for opioid dependency would only be dispensed at a federally licensed program thus all methadone prescriptions would be for chronic pain.
Outcome variables
The primary outcomes were adherence to pain management guidelines and evidence of opioid misuse analyzed according to physician status (resident vs. attending physician). We examined adherence to any one of two American Pain Society Guidelines (4): 1) EHR documentation of at least one controlled substance agreement (contract) ever, and 2) any urine drug testing in the past year. These outcomes were chosen because they could be identified with specificity through documents and laboratory results in the EHR.
We defined potential prescription misuse as receipt of two or more early refills of an opioid medication in the past year, a cut-off that has been used in previous studies (2,11). Two or more refills are more likely to indicate a pattern indicative of opioid misuse as compared to one early refill, which may be given for legitimate reasons such as dose escalation. An early refill was defined as a prescription written 7–25 days after the previous prescription for the same medication. We used a lower limit of seven days to exclude any refills that were simply reprints of a prescription. Since some patients may require a prescription a few days early within a typical 28-day refill cycle for personal reasons, such as a vacation, we used an upper limit of 25 days. In another study a similar 3-day window period was used to identify an early refill (12). We conducted a manual review of a random sample of 100 charts to determine the proportion of early refills that were not suggestive of misuse e.g. for planned dose escalation, progression of disease, or for travel. Using the upper and lower limits of early refills as stated above, we determined that for 80% of the early refills, the prescriber did not document a reason for the early refill, and thus we interpreted this as evidence of potential patient opioid misuse.
Covariates
We identified demographic variables including gender, age, race, language and primary insurance as covariates in multiple regression models. Medical comorbidities were examined using the Deyo modification of the Charlson Comorbidity Index (13). If the Charlson-Deyo score was one or greater we classified patients as having significant co-morbidity. We measured healthcare utilization as the number of PCP and emergency department (ED) visits in the past year. We also classified patients according to whether opioids were initiated prior to 2010, a potential indication of a longer patient-provider relationship. We did not have data on outside facility visits.
We also examined covariates known to affect receipt of guideline-concordant care for opioid prescribing (4). These include risk factors for opioid misuse identified in previous research: 1) age < 45 years at the time of study initiation (September 1, 2011), 2) current or past drug use disorder, 3) current or past alcohol use disorder, 4) current or past tobacco use and 5) current or past mental health disorder (i.e. anxiety, depression, bipolar disorder, post-traumatic stress disorder or schizophrenia) (14). Drug, alcohol, and tobacco use and mental health disorders were identified through ICD-9-CM codes listed on the EMR problem list or through billing codes. Each of the individual risk factors was given a score of one if positive. We then summed the number of risk factors for each patient.
Statistical analysis
We conducted chi-square and t-tests to examine whether patient characteristics and outcome variables (i.e., controlled substance agreement ever, urine drug testing in the past year and two or more early refills in the past year) differed by patients’ physician status. Multiple logistic regression analyses were performed to examine the association between physician status and each of the three binary outcome variables. We included covariates in the adjusted models that we determined to have a priori clinical relevance or that were statistically significant in the bivariate analyses with a p-value ≤ 0.1. We also examined the correlation among the predictor variables to determine the existence of collinearity. Because one physician can attend to multiple patients, we expected clustering among patients within physicians. Therefore, we used Generalized Estimating Equations (GEE) method to further adjust for clustering among patients. For our final models, we used a two-sided type I error rate of <0.05. All analyses were conducted using the SAS, Version 9.1.
Results
We identified 1,285 patients who were on chronic opioid treatment and met our inclusion criteria; 215 were resident patients, and 1,070 were attending patients (Table 1). Relative to attending patients, resident patients were more likely to be male, younger and on Medicaid. A higher proportion of attending patients had opioids initiated prior to 2010 and was dual-eligible for Medicare/Medicaid, indicating chronic medical or mental disability. Resident patients had fewer primary care visits and more ED visits relative to attending patients. The mean number of opioid risk factors was higher in resident patients (2.09 vs. 1.85 p=0.006).
Table 1.
Patient Characteristics According to Physician Status
| Characteristics | Resident Patients* (n=215) |
Attending Patients† (n=1070) |
P- value‡ |
|---|---|---|---|
| Male, n (%) | 128 (59.5) | 528 (49.4) | 0.006 |
| Mean Age-years (SD) | 49.60 (13.9) | 54.8 (12.1) | <0.001 |
| Race, n (%) | 0.28 | ||
| Non-Hispanic black | 101 (47.0) | 558 (52.2) | |
| Non-Hispanic white | 81 (37.7) | 341 (31.9) | |
| Hispanic | 25 (11.6) | 114 (10.7) | |
| Other | 8 (3.7) | 57 (5.3) | |
| Language, n (%) | 0.15 | ||
| English | 201 (93.5) | 967 (90.4) | |
| Non-English | 14 (6.5) | 103 (9.6) | |
| Insurance, n (%) | <0.001 | ||
| Medicaid | 91 (42.3) | 339 (31.7) | |
| Medicare | 24 (11.2) | 151 (14.1) | |
| Dual-eligible§ | 64 (29.8) | 370 (34.6) | |
| Private | 17 (7.9) | 149 (13.9) | |
| Other | 19 (8.8) | 61 (5.7) | |
| Significant medical comorbidity‖, n (%) | 150 (69.8) | 759 (70.9) | 0.73 |
| Opioids initiated prior to 2010¶, n (%) | 135 (62.8) | 860 (80.4) | <0.001 |
| PCP visits, past year, n (%) | 0.001 | ||
| 1–3 | 65 (30.2) | 208 (19.4) | |
| 4–6 | 67 (31.2) | 430 (40.2) | |
| >7 | 83 (38.6) | 432 (40.4) | |
| ED visits, past year, n (%) | <0.001 | ||
| 0 | 59 (27.4) | 531 (49.6) | |
| 1–3 | 86 (40.0) | 427 (39.9) | |
| 4–6 | 31 (14.4) | 77 (7.2) | |
| >7 | 39 (18.1) | 35 (3.3) | |
| Opioid misuse risk factors#, n (%) | |||
| Age <45 | 76 (35.4) | 211 (19.7) | <0.001 |
| Drug use disorder | 103 (47.9) | 481 (45.0) | 0.43 |
| Alcohol use disorder | 42 (19.5) | 147 (13.7) | 0.03 |
| Tobacco use | 96 (44.7) | 452 (42.2) | 0.51 |
| Mental health disorder | 133 (61.9) | 692 (64.7) | 0.43 |
| Mean no. of opioid risk factors# (SD) | 2.09 (1.3) | 1.85 (1.2) | 0.006 |
Any patient who received >2 opioid prescriptions from a resident physician in the past year
Includes both attending physicians and nurse practitioners
Chi-square test for categorical variables and t-test for continuous variables
Patients who receive both Medicare and Medicaid
Charlson-Deyo score >1
Patients who were started on opioid prescriptions prior to 2010 as documented in our electronic medical record (EMR)
Opioid risk factors based on ICD-9 codes as noted in EHR
ICD: international classification of diseases
PCP: Primary care providers (includes both physicians and nurse practitioners)
Table 2 compares receipt of guideline-concordant care and potential opioid misuse between the two groups. Fewer than half of patients in either group had a controlled substance agreement (45.1% of resident patients vs. 42.4% of attending patient, p=0.47), and the proportion that received urine drug testing was similar in both groups (58.6% of resident patients vs. 63.6% of attending patients, p=0.16). Resident patients were more likely to have two or more early refills in the past year relative to attending patients (42.8% vs. 32.5%; p=0.004). We did not detect high correlation among the predictor variables.
Table 2.
Guideline Adherence* to Opioid Risk-Reduction Strategies, and Possible Opioid Misuse
| Variable | Resident Patients† (n=215) | Attending Patients‡ (n=1070) | P-value§ |
|---|---|---|---|
| Controlled substance Agreement, ever‖, n (%) | 97 (45.1) | 454 (42.4) | 0.47 |
| Urine Drug Tests¶, n (%) | 126 (58.6) | 681 (63.6) | 0.16 |
| Early Refills#, n (%) | 0.004 | ||
| 0/1 | 123 (57.2) | 722 (67.5) | |
| >2 | 92 (42.8) | 348 (32.5) |
Adherence to American Pain Society Guidelines
Any patient who received >2 opioid prescriptions from a resident physician in the past year
Includes both attending physicians and nurse practitioners
Chi-square test
Documentation of a controlled substance agreement in EMR, ever
Any urine drug testing in the past year
Prescription written 7–25 days after the previous prescription of the same drug
EMR: electronic medical record
In the adjusted regression analysis using the GEE method (Table 3), patients of resident physicians were more likely to receive early refills (OR 1.82, 95% CI 1.26–2.62) than patients of attending physicians, but did not differ from patients of attending physicians in their receipt of controlled substance agreements or urine drug testing. Male patients were more likely than female patients to have a controlled substance agreement or to receive urine drug testing. White patients and those who spoke English were also more likely to receive urine drug testing. Compared to those with fewer than four primary care visits in the prior year, patients with four or more visits were more likely to have controlled substance agreements (OR 1.98, 95% CI 1.42–2.75) and to receive urine drug testing (OR=3.15, 95% CI 2.35–4.23), yet had an increased risk of early refills (OR 2.65, 95% CI 1.73–4.07). Patients with some opioid risk factors (drug and tobacco use) were more likely to have controlled substance agreements and to receive urine drug testing, while patients with other opioid risk factors (alcohol use and mental health problems) did not have higher receipt of agreements and urine testing. Patients with drug use, tobacco use and mental health disorders had an increased risk of receiving early refills (OR 1.37, 95% CI 1.05–1.79); OR 1.31, 95% CI 1.04–1.66 and OR 1.62, 95% CI 1.31–2.01, respectively).
Table 3.
Predictors of having a controlled substance agreement*, urine drug testing† or opioid misuse‡ based on GEE method.
| Controlled Substance Agreement* |
Urine Drug Testing† |
Early Refills‡ | ||||
|---|---|---|---|---|---|---|
| Variable | Adjusted OR§ |
95% CI | Adjusted OR§ |
95% CI | Adjusted OR§ |
95% CI |
| Physician Status | ||||||
| Attending physician‖ | 1.0 | (reference) | 1.0 | (reference) | 1.0 | (reference) |
| Resident Physician¶ | 1.01 | (0.70, 1.45) | 0.78 | (0.54, 1.13) | 1.82 | (1.26, 2.62)# |
| Gender | ||||||
| Female | 1.0 | (reference) | 1.0 | (reference) | 1.0 | (reference) |
| Male | 1.28 | (1.06, 1.56)# | 1.52 | (1.17, 1.97)# | 1.00 | (0.78, 1.29) |
| Race | ||||||
| Non-white race | 1.0 | (reference) | 1.0 | (reference) | 1.0 | (reference) |
| White race | 1.28 | (0.99, 1.66) | 1.51 | (1.16, 1.96)# | 1.52 | (1.18, 1.97)# |
| Age | ||||||
| Age > 45 years | 1.0 | (reference) | 1.0 | (reference) | 1.0 | (reference) |
| Age < 45 years | 1.25 | (0.92, 1.71) | 1.11 | (0.85, 1.46) | 1.18 | (0.89, 1.56) |
| Language | ||||||
| Non-English Language | 1.0 | (reference) | 1.0 | (reference) | 1.0 | (reference) |
| English Language | 1.63 | (1.08, 2.48)# | 1.86 | (1.25, 2.76)# | 1.46 | (0.92, 2.30) |
| Insurance | ||||||
| Private insurance | 1.0 | (reference) | 1.0 | (reference) | 1.0 | (reference) |
| Non-private insurance | 1.35 | (1.00, 1.83) | 0.98 | (0.70, 1.37) | 0.76 | (0.58, 1.01) |
| Opioids initiated | ||||||
| Opioids initiated prior to 2010 | 1.0 | (reference) | 1.0 | (reference) | 1.0 | (reference) |
| Opioids initiated after 2010 | 0.79 | (0.57, 1.09) | 0.87 | (0.65, 1.18) | 1.16 | (0.82, 1.63) |
| Primary Care visits | ||||||
| 1–3 | 1.0 | (reference) | 1.0 | (reference) | 1.0 | (reference) |
| >4 | 1.98 | (1.42, 2.75)# | 3.15 | (2.35, 4.23)# | 2.65 | (1.73, 4.07)# |
| ED visits | ||||||
| >1 ED visits | 1.0 | (reference) | 1.0 | (reference) | 1.0 | (reference) |
| No ED visits | 1.25 | (1.06, 1.47)# | 1.48 | (1.15, 1.90)# | 1.27 | (1.00, 1.61) |
| Alcohol Use** | ||||||
| No alcohol use disorder | 1.0 | (reference) | 1.0 | (reference) | 1.0 | (reference) |
| Alcohol use disorder | 0.91 | (0.69, 1.20) | 0.74 | (0.53, 1.04) | 0.94 | (0.69, 1.28) |
| Drug Use** | ||||||
| No drug use disorder | 1.0 | (reference) | 1.0 | (reference) | 1.0 | (reference) |
| Drug use disorder | 3.05 | (2.30, 4.03)# | 6.13 | (4.34, 8.66)# | 1.37 | (1.05, 1.79)# |
| Tobacco Use** | ||||||
| No tobacco use disorder | 1.0 | (reference) | 1.0 | (reference) | 1.0 | (reference) |
| Tobacco use disorder | 1.27 | (1.05, 1.54)# | 1.43 | (1.14, 1.79)# | 1.31 | (1.04, 1.66)# |
| Mental health** | ||||||
| No mental health disorder | 1.0 | (reference) | 1.0 | (reference) | 1.0 | (reference) |
| Mental health Disorder | 1.04 | (0.84, 1.28) | 1.19 | (0.94, 1.51) | 1.62 | (1.31, 2.01)# |
Documentation of a controlled substance agreement in the electronic medical record (EMR), ever
Any urine drug testing in the past year
Opioid misuse defined as two or more early refills (prescription written 7–25 days after the previous prescription of the same drug) in the past year
Adjusted for resident status, gender, race, language, insurance, opioid initiation after 2010, PCP visits, ED visits, opioid misuse risk factors (age, drug use disorder, alcohol use disorder, tobacco use, mental health disorder as documented in the EMR)
OR=odds ratio
CI=confidence interval
Includes both attending physicians and nurse practitioners
Any patient who received >2 opioid prescriptions from a resident physician in the past year
p-value: < 0.05
Based on EMR ICD-9 (international classification of diseases) codes on EMR problem list and billing codes
GEE: Generalized Estimating Equations
Discussion
Attending and resident physicians provided similar levels of guideline concordant monitoring to patients on chronic opioid therapy. Fewer than half of patients in either group received a controlled substance agreement, and less than two-thirds received any urine drug testing. Over one-third of patients received multiple early refills, a marker of potential opioid misuse, with resident physicians providing more early refills than attending physicians.
Our hypothesis that residents and attendings would deliver similar care was partially supported by evidence that both groups had similar patterns in monitoring patients. However, the resident patient population was at higher risk for possible opioid misuse, as demonstrated by both the number of risk factors as well as higher odds of early refills, suggesting that this level of monitoring was probably inadequate.
A combination of provider and patient factors may have led to increased potential opioid misuse in the resident patient population. Resident patients had more opioid misuse risk factors relative to attending patients, a finding consistent with a prior study (15). There may be other, unmeasured characteristics of resident patients. Some patients with intention to deceive their provider to obtain opioid prescriptions may seek new clinicians more frequently, and resident physicians may be more likely to have available new patient appointments given resident turnover every three years. These same patients may also seek out residents with the hope that residents may be more likely to prescribe opioid medications due to clinical inexperience. Because residents have less continuity with their primary care patients and have fewer sessions per week (and hence a reduced presence in clinic), it is possible that in the residents’ absence, other covering physicians provide early opioid refills, especially if the patients’ pain is not controlled. Resident patients may also be less engaged in primary care as indicated through a lower number of primary care visits as compared to attending patients. Attending physicians, through long-term relationships with their patients, may inevitably ‘train’ their patients in compliance and specific expectations in order to receive continued opioid prescriptions. They may have also stopped opioid therapy in patients who did not adhere with monitoring. Finally, resident physicians in Massachusetts (unlike those in other states) are not required to obtain an attending co-signature on opioid prescriptions. It is possible that the requirement of an attending co-signature promotes additional case discussion between attending and resident that leads to improved monitoring. It may also be that those long-standing patients of attending patients have demonstrated reliability on opioids, whereas long-standing patients who were unable to manage opioids were stopped prior to the study, which would account for the lower risk profile of attending patients.
Our findings differ in some respects from prior literature. The level of monitoring in our study was significantly higher than that noted in previous studies where fewer than 10% of similar patients received urine drug testing (2,9). Low prevalence of monitoring has been attributed to lack of training in treating and monitoring patients on chronic opioids (16). We suspect that the level of urine drug testing in our study is higher than in prior studies due to local expertise in opioid prescribing for chronic pain, which may influence practice patterns. In addition, there is increasing national awareness of monitoring and prescribing strategies due to educational campaigns, such as Risk Evaluation and Mitigation Strategy (REMS) (17). It is likely that monitoring practices have increased nationally compared to earlier studies. However the levels of monitoring observed in our study are still significantly lower than that endorsed by clinical guidelines (4) i.e. all patients on chronic prescription opioids should have at least one urine drug test per year. Prior studies have shown that physicians monitor black patients with urine drug testing more closely than white patients, despite the fact that the latter are more likely to abuse prescription opioids (9,14). This was not the case in our study; whites received more intensive monitoring than black patients.
Overall, a higher proportion of patients in our study received early refills relative to patients in a previous study (2). This finding may be due to a high prevalence of opioid misuse risk factors among the patients served by this primary care practice, reflecting a vulnerable safety-net patient population (9).
Our study has several limitations. Data were abstracted from the EMR and therefore mental health, tobacco use, alcohol use and substance use disorders were derived from billing information or ICD codes, which may be incomplete or unreliable (18). We did not have information about early refills provided by prescribers outside of the primary care practice. Thus, the prevalence of early refills in our study is likely an underestimate. However, because 20% of early refills on our chart review were prescribed for legitimate reasons (e.g. for vacation supply or dose escalation) using early refills as a proxy for opioid misuse may over-estimate the magnitude of such misuse. While each early refill may not indicate misuse, it does create an opportunity for misuse or diversion to occur. Our attempts to control for any systematic differences due to length of time on chronic opioid therapy may not fully capture the differences. The findings from our study in a practice setting where there is a high level of local addiction expertise and in which resident opioid prescriptions are not co-signed by attendings may not be generalizable to other settings.
Our findings highlight the need to improve monitoring of patients with chronic pain on opioids. Such efforts should include education of PCPs at all stages of training about how to monitor patients on chronic opioid therapy according to the best available evidence. PCPs can also be trained in Screening, Brief Intervention and Referral to Treatment (SBIRT) developed by Substance Abuse and Mental Health Services Administration (SAMHSA) to help detect and intervene if required, among individuals with substance use disorder or a high risk for developing substance use disorders (19). Use of prescription monitoring programs (PMPs) can help prescribers identify patients at risk for misuse. Future studies could explore whether attending co-signatures on resident opioid prescriptions reduce opioid misuse among resident patients. As practices evolve into patient-centered medical homes, use of population management (e.g. registries of patients on chronic opioid therapy) may improve monitoring and reduce opioid misuse. Use of EHRs in medical homes may promote adoption of clinical guidelines through clinical decision support tools, reminders and alerts to PCPs (19). Such decision support tools to operationalize clinical guidelines may increase adherence to recommendations (20).
Acknowledgments
This research was supported by a grant from the National Institute on Drug Abuse R01DA034252
Footnotes
Disclosure/Conflicts of Interest: No conflicts of Interest
References
- 1. [Accessed Jul 18 2013];CDC - Focus Area: Prescription Drug Overdose - Injury Center. Available from: http://www.cdc.gov/injury/about/focus-rx.html.
- 2.Starrels JL, Becker WC, Weiner MG, Li X, Heo M, Turner BJ. Low use of opioid risk reduction strategies in primary care even for high risk patients with chronic pain. J Gen Intern Med. 2011;26(9):958–964. doi: 10.1007/s11606-011-1648-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Juurlink DN, Dhalla IANL. Improving Opioid Prescribing: the New York City recommendations. JAMA. 2013;309(9):879–880. doi: 10.1001/jama.2013.1139. [DOI] [PubMed] [Google Scholar]
- 4.Chou R, Fanciullo GJ, Fine PG, Adler Ja, Ballantyne JC, Davies P, et al. Clinical guidelines for the use of chronic opioid therapy in chronic noncancer pain. J Pain. 2009;10(2):113–130. doi: 10.1016/j.jpain.2008.10.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Walker KO, Calmes D, Hanna N. The impact of public hospital closure on medical and residency education: implications and recommendations. J Natl Med Assoc. 2008;100(12):1377–1383. doi: 10.1016/s0027-9684(15)31537-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Chatterjee P, Joynt KE, Orav EJ, Jha AK. Patient experience in safety-net hospitals: implications for improving care and value-based purchasing. Arch Intern Med. 2012;172(16):1204–1210. doi: 10.1001/archinternmed.2012.3158. [DOI] [PubMed] [Google Scholar]
- 7.Zallman L, Ma J, Xiao L, Lasser KE. Quality of US primary care delivered by resident and staff physicians. J Gen Intern Med. 2010;25(11):1193–1197. doi: 10.1007/s11606-010-1456-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Touchet BK, Yates WR, Coon KA. Opioid contract use is associated with physician training level and practice specialty. J Opioid Manag. 2005;1(4):195–200. doi: 10.5055/jom.2005.0042. [DOI] [PubMed] [Google Scholar]
- 9.Becker WC, Starrels JL, Heo M, Li X, Weiner MGTB. Racial differences in Primary Care Opioid Risk Reduction Strategies. Annnals Fam Med. 2011;9(3):219–225. doi: 10.1370/afm.1242. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Ode DEAOD. TRAMADOL (Trade Names: Ultram®, Ultracet®) [Accessed on July 14, 2014];:7183. Available from: http://www.deadiversion.usdoj.gov/drug_chem_info/tramadol.pdf.
- 11.Reid MC, Engles-Horton LL, Weber MB, Kerns RD, Rogers EL, O’Connor PG. Use of opioid medications for chronic noncancer pain syndromes in primary care. J Gen Intern Med. 2002;17(3):173–179. doi: 10.1046/j.1525-1497.2002.10435.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Katz N, Panas L, Kim M, Audet AD, Bilansky A, Eadie J, et al. Usefulness of prescription monitoring programs for surveillance — analysis of Schedule II opioid prescription data in Massachusetts, 1996 – 2006. Pharmacoepidemiol Drug Saf. 2010;19(2):115–123. doi: 10.1002/pds.1878. [DOI] [PubMed] [Google Scholar]
- 13.Lasser KE, Kim TW, Alford DP, Cabral H, Saitz R, Samet JH. Is unhealthy substance use associated with failure to receive cancer screening and flu vaccination? A retrospective cross-sectional study. BMJ Open. 2011;1(1):e000046. doi: 10.1136/bmjopen-2010-000046. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Liebschutz JM, Saitz R, Weiss RD, Averbuch T, Meltzer EC, Claggett-borne E, et al. Clinical factors associated with prescription drug use disorder in urban primary care patients with chronic pain. J Pain. 2010;11(11):1047–1055. doi: 10.1016/j.jpain.2009.10.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Colburn JL, Jasinski DRRDA. Long-term opioid therapy, aberrant behaviors, and substance misuse: comparison of patients treated by resident and attending physicians in a general medical clinic. J opioid Manag. 2012;8(3):153–160. doi: 10.5055/jom.2012.0111. [DOI] [PubMed] [Google Scholar]
- 16.Liebschutz JM, Alford DP. Safe opioid prescribing: a long way to go. J Gen Intern Med. 2011;26(9):951–952. doi: 10.1007/s11606-011-1797-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Nelson LS. Curbing the Opioid Epidemic. JAMA. 2013;308(5):2012–2013. doi: 10.1001/jama.2012.8165. [DOI] [PubMed] [Google Scholar]
- 18.SE G. The sensitivity and specificity of computererized databases for the diagnosis of rheumatoid arthritis. Arthritis Rheum. 1994;37(6):821–823. doi: 10.1002/art.1780370607. [DOI] [PubMed] [Google Scholar]
- 19.Cheatle MD, Klocek JW, McLellan AT. Managing pain in high-risk patients within a patient-centered medical home. Transl Behav Med. 2012;2(1):47–56. doi: 10.1007/s13142-012-0113-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Trafton JA, Martins SB, Michel MC, Wang D, Tu SW, Clark DJ, et al. Designing an automated clinical decision support system to match clinical practice guidelines for opioid therapy for chronic pain. Implement Sci. 2010;5:26. doi: 10.1186/1748-5908-5-26. [DOI] [PMC free article] [PubMed] [Google Scholar]