bs_bs_banner
Pain Medicine 2014; 14: 625–636 Wiley Periodicals, Inc.
Physicians’ Beliefs and Likelihood of Prescribing Opioid Tamper-Resistant Formulations for Chronic Noncancer Pain Patients
Dennis C. Turk, PhD,* Elizabeth J. Dansie, PhD,† Hilary D. Wilson, PhD,† Bruce Moskovitz, MD,‡ and Myoung Kim, PhD‡
Subjects. A nationally representative sample of 1,535 practicing physicians throughout the United States.
*Department of Anesthesiology & Pain Medicine and Center for Pain Research on Impact, Measurement, & Effectiveness, University of Washington;
Methods. A stepwise hierarchical multiple linear regression analysis was conducted to estimate if physician characteristics, opinions, or geographic region categorized according to state rates of mortality by drug overdose and milligrams of opioids prescribed by state were predictive of the likelihood of prescribing TRFs.
†
Outcomes Research, United Biosource Corporation, Seattle, Washington; ‡
LLC, Health Economics & Outcomes Research, Janssen Scientific Affairs, Titusville, New Jersey, USA Reprint requests to: Dennis C. Turk, PhD, Department of Anesthesiology & Pain Medicine, University of Washington, Box 356540, Seattle, WA 98195, USA. Tel: (206) 616-2626; Fax: (206) 543 2958; E-mail: [email protected]. Dr. Turk has received grant support and/or consulting fees from Eli Lilly, OrthoMcNeil-Janssen, Pfizer, Phillips Respironics, the National Institutes of Health, and the United States Food and Drug Administration.
Abstract Background. Tamper-resistant opioid formulations (TRFs) have recently been the target of active development in an effort to deter opioid misuse and abuse.
Results. Board certification in Pain Medicine and prescribing opioids to a higher volume of CNCP patients were significantly predictive of a reported likelihood of prescribing TRFs, in addition to concerns about possible misuse and abuse of opioids, beliefs in the effectiveness of opioids for CNCP, and greater satisfaction with education and training in pain management this set of factors accounted for 21% of the model variance. Rates of mortality by drug overdose and opioid prescription volume by location were not predictive of TRF usage. Conclusions. Reducing physician concerns about potential misuse and abuse of opioids through additional education in pain management and dissemination of information about the potential benefits and availability of TRFs should influence physicians’ attitudes about and the adoption of TRFs. Key Words. Opioids; Tamper-Resistant Formulations; Prescribing Patterns; Attitudes; Chronic Noncancer Pain
Objective. To understand factors that are predictive of physicians’ likelihoods of prescribing TRFs to patients with chronic noncancer pain (CNCP).
Introduction
Design. A cross-sectional survey was conducted, utilizing a questionnaire of clinicians’ attitudes and opinions about opioids for CNCP (Clinicians’ Attitudes about Opioids Scale) to explore beliefs about and likelihood of prescribing TRFs.
The use of long-term opioid therapy (LOT) for the management of chronic noncancer pain (CNCP) has risen dramatically since the mid-1990s [1,2]. Although opioids are considered appropriate for acute and cancer-related pain, their benefits and risks are less well understood and 625
Turk et al. more controversial in the context of extended use for CNCP patients [3,4]. Opioids remain the most commonly prescribed category of drug in the United States [5,6], where 3% of adults are prescribed LOT [4]. Americans currently consume approximately 80% of the global opioid supply [7]; however, there is considerable variation in prescribing rates of opioids both across and within states and regions [8]. Available evidence reveals that the largest increase in opioid prescribing can be attributed to the use of Schedule II opioids for CNCP [9]. Corresponding to the rise in opioid prescribing is the expansion in rates of opioid misuse (i.e., inappropriate use of a medication for medical purposes, rather than mindaltering effects), abuse (i.e., use of medication for mindaltering effect or in any manner not consistent with the way in which the opioid was prescribed by the treating physician), as well as both fatal and nonfatal overdoses [7,10–12]. For example, in 2010, 4.8% of the US population aged ≥12 (∼14.4 million) used opioids nonmedically [13]. Nonadherence rates for prescribed opioids range from 2% to a high of 53% depending on how misuse and abuse are defined and measured [14], while recent data on illicit use indicate that prescription opioids account for more overdose deaths than heroin and cocaine combined [12,15]. This relationship is more than circumstantial, as multiple investigations have been conducted to support the purported causal influence of increased opioid prescribing to higher rates of misuse, abuse, emergency room visits, and overdose [12,16–18]. Unfortunately, education directed toward prescribers to assist them to develop the most appropriate pain management strategy designed for individual patients is limited [19]. In response to the current concerns about LOT, the United States Food and Drug Administration (FDA) has identified the need for the development of tamper-resistant formulations (TRFs, also known as abuse-deterrent formulations) of extended-release opioids. Significant efforts and financial investment by the pharmaceutical industry have been devoted to develop abuse-deterrent formulations, and these efforts are ongoing [20]. The intention for the development of TRFs is to treat pain in patients who might benefit from opioid medication while concomitantly minimizing the rate of misuse, abuse, and other negative outcomes, hence increasing the benefit–risk ratio for opioids [20–22]. Several TRFs are in development that utilize a range of techniques to combat misuse and abuse, such as formulations of opioids that: 1) have properties that provide physical barriers to tampering (e.g., barriers to crushing, chewing, or grinding); 2) contain agonist– antagonist formulations that interfere with the euphoria experienced when the drug is abused; 3) contain an additional substance that creates an aversive side effect for the individual if abused; 4) contain a prodrug (which hinders the opioid effectiveness until it is transformed in the gastrointestinal tract); or 5) are more difficult to misuse simply because of the route of administration [21,23]. At the present time, five TRFs are on the market: OpanaR ER (Oxymorphone extended release), OxectaR (Oxycodone), OxyContinR (Oxycodone 626
controlled release), ExalgoR (osmotic-controlled release oral delivery system [OROS] hydromorphone) [24], and NucyntaR ER (tapentadol extended release) [25] and multiple TRFs are in development. There are substantial financial incentives for the development and marketing of these products. A budget-impact model for prescription TRFs estimated the market in the United States alone could range from $600 million to $1.6 billion per year [26], demonstrating the potential economic impact if these formulations gained widespread acceptance. It is inevitable that TRFs approved by the FDA would be more costly than existing formulations. However, the economic burden of opioid misuse and abuse is substantial, with some cost estimates ranging from $50–$75 billion annually in lost productivity, criminal justice expenditures, drug abuse treatments, and medical complications [27,28]. Thus, prescribers and payers will have to balance the additional costs of TRFs against the benefits from lowered costs resulting from potential savings accruing for reduced misuse and abuse in making decisions about when and for whom TRFs are warranted. As an illustration of this dilemma, in April 2013, the FDA rejected a generic version of OxyContin, leaving only the reformulated, tamper-resistant OxyContin that is less prone to crushing, snorting, and other forms of abuse [29]. Beyond costs, several additional factors that are not completely independent might impact the adoption of TRFs if traditional formulations remain available. For example, clinicians’ own concerns about the potential for misuse or diversion of opioids, beliefs in their ability to identify “atrisk” patients, variability in beliefs about the overall effectiveness of opioids for managing chronic pain and improving function, and the education in pain management the physician has received could all influence opioid prescribing patterns. Moreover, there is large variability in prescribing patterns [8] and overdose deaths [30] from opioids across geographic states, making it possible that geographic location could influence who would be likely to use opioid formulations with a lower potential for aberrant behaviors. Although several opioid TRFs are currently marketed [20,24], no investigations have reported on the factors that determine their acceptability, which factors influence providers’ attitudes and beliefs about the necessity and value of TRFs, or the likelihood that clinicians will actually adopt these formulations [21]. Recently a new measure, the Clinicians’ Attitudes about Opioids Scale (CAOS), was developed to assess the attitudes and practices of providers who prescribe opioids for chronic pain patients [31]. One of the subscales of the CAOS specifically assesses clinicians’ attitudes and beliefs about TRFs. The purpose of the present study was to explore the factors that play a role in the acceptance and likelihood of prescribing TRFs in a nationally representative United States sample of physicians. The following hypotheses were addressed: 1) higher evaluation of the perceived effectiveness of opioids and concerns for the risk of misuse and abuse inherent in opioid prescribing would
Prescribing Tamper-Resistant Formulations be associated with a greater likelihood of prescribing TRFs; and 2) physicians practicing in states with greater numbers of deaths attributed to opioids would be more likely to prescribe TRFs, as they would be more aware of the potential negative outcomes associated with prescription opioid abuse due to medical, regulatory, and media attention, in turn increasing their likelihood of adopting potentially less abusable formulations. Methods Participants Participants were recruited in 2010 using a large, proprietary, Web-enabled panel of United States physicians that is maintained by the private company, Toluna, Inc. (Wilton, CT, USA). The panel comprised approximately 100,000 American Medical Association (AMA)-verified physicians who had been recruited by means of the Internet to join the panel through a “double-opt-in” process, where their targeting information, such as primary specialty, subspecialty, purchasing authority, and involvement in Pharmacy and Therapeutics committees, is stored. The panel is designed to be representative of the United States population of physicians by age, race/ethnicity, geographic region, and medical specialty according to the AMA. The company conducts an extensive validation of every panelist to guarantee all respondents are licensed and AMAverified physicians prior to extending invitations to participate in a survey. Panel physicians may withdraw membership at any time and are compensated for their participation at a rate Toluna has determined to commensurate with the length of the survey and medical specialty. All physicians on the Toluna panel in the following specialties were eligible to participate in our survey: Anesthesiologists (Anesth), General/Family Practitioners (G/FP), Neurologists (Neurol), Orthopedists (Ortho), Pain Medicine specialists (PM), Internists (IM), Physical Medicine and Rehabilitation specialists (PM&R), Psychiatrists (Psych), Rheumatologists (Rheum), and Surgeons (Surg). Panel members received notification of the survey through email and were compensated between $50–75 depending on their specialty. The study was approved by the Institutional Review Board of the University of Washington, and all participants gave informed consent. Measure CAOS was used to assess current beliefs regarding opioids, opioid use, and patients with CNCP. The threephase questionnaire development process and psychometric evaluation of this scale is described in detail elsewhere [30]. The CAOS consists of 38 content questions and 13 demographic and descriptive questions concerning providers’ practice type, volume of CNCP patients in their practice, and their opioid-related prescription practices. The response scale for all content questions is 0–10, with 0 being “strongly disagree” and 10 being “strongly agree.” A principal component analysis revealed that the questions comprised five subscales that measured clinicians’ attitudes regarding opioids, as
well as considerations that may influence the likelihood of prescribing opioids or TRFs of such opioids. Total subscale scores were created by summing the scores for each item within each subscale. The measure was shown to be internally consistent (Cronbach’s alpha = 0.87) and demonstrated high stability over a 2-week period (test–retest reliability coefficients for each subscale = 0.62–0.79). As described in detail by Wilson and colleagues [30], questions comprising each subscale did not load on multiple factors, while correlations between factors were small to moderate, indicating that each subscale measured unique aspects of clinicians’ attitudes and beliefs about prescribing. These scale characteristics and scale development substantiated the validity of our use of this scale in statistical models of association as detailed below. The five factors composing the CAOS are as follows: 1. Impediments and concerns (Impediments): A 16-item subscale with higher scores reflecting increased concerns (e.g., “Physical dependence is an impediment to taking opioids for long periods of time”; “Long-term use of opioids is over-prescribed for patients with chronic non-cancer pain”). 2. Perceived effectiveness (Effectiveness): A nine-item subscale with higher scores associated with increased beliefs in the efficacy of opioids for treatment of CNCP (e.g., “In general, opioids are effective for mixed pain”; “Opioids are the most effective treatments available for persistent pain”). 3. Medical education (Education): A three-item subscale with higher scores indicative of increased confidence in education/training received for the treatment of CNCP (i.e., “My education regarding pain evaluation and treatment during medical school is appropriate”; “My education regarding pain evaluation and treatment during residency was appropriate”; “My current practice involving the use of opioids for the treatment of chronic non-cancer pain is consistent with my training.”). 4. Schedule II vs III Opioids (Schedule II vs III): A five-item subscale with higher scores associated with decreased preference and increased concern surrounding Schedule II opioids (e.g., “Increased administrative burden prevents me from prescribing Schedule II opioids”; “Fear of DEA scrutiny inhibits me from prescribing Schedule II opioids”). 5. Tamper-resistant formulations and dosing (TRFD): A 3-item subscale with higher scores reflecting a preference for more controlled dosing and TRFs (i.e., “I would be more likely to prescribe opioids if TRF formulas were available”; “I would prescribe a TRF form over a generic standard form for my patients even if it was more expensive”; “I would try to prescribe TRF over the generic/standard form only for those patients with higher abuse potential”). In addition to the CAOS items, we asked respondents to complete six additional questions that, although did not measure the same constructs represented by the five derived subscales, were considered important when 627
Turk et al. describing clinicians prescribing habits such as “Patient demographics (e.g., age, sex, race) influence my prescribing opioids on a long-term basis,” “My opioid prescribing is influenced by my pharmacy benefits package,” and “Increasing opioid-related mortalities has influenced my opioid prescribing habits.” Finally, we created two variables to explore the influence of geographic region on TRF usage. First, we created a three-category “Mortality overdose by region” (MOR) variable by categorizing states according to the drug overdose rates by state (e.g., 14.9% and higher, 9.5–14.8%, and less than 9.5%) [29]. Second, we used McDonald, Carlson, and Izrael’s [8] ranking of the geographic states (in the United States) according to estimates of the mean milligrams (mg) of opioids (in morphine equivalents) dispensed per resident by state to create a three-category “Quantity of opioids by region” (QOR) variable to indicate if participants resided in a state with high or low quantities of dispensed opioids (e.g., top 10 vs middle 30 vs bottom 10 prescribing states). Statistical Analysis Demographic and descriptive characteristics of participating physicians, as well as the distribution of survey items and CAOS total subscale scores, were assessed using frequencies and percentages for categorical variables and means and standard deviations for continuous variables. Quartiles were calculated for clinicians’ total scores on the TRFD subscale, in order to categorize clinicians according to their reported likelihood of prescribing TRFs. In order to understand those clinicians most and least likely to use TRFs better, demographic characteristics of the sample were also calculated for those clinicians that scored within the upper (Q4) and lower quartiles (Q1) on the TRFD factor. Chi-square (χ2) analyses were performed to determine if the clinicians in Q4 or Q1 of TRFD differed with respect to demographic and practice characteristics (e.g., gender, age, region of practice, employment setting). Pearson’s r correlations were calculated to examine the intercorrelations among the five CAOS total subscales scores. A stepwise hierarchical multiple linear regression analysis was conducted to address our primary research question, namely, what physician characteristics, opinions, and beliefs are predictive of the reported likelihood of prescribing TRFs. Unstandardized and standardized (β) coefficients, along with standard errors and significance values are reported to assess the strength and significance of the reported associations. The fit of each model was assessed by examining the adjusted R2 value and F value for change in R2 when progressing from Model 1 to Model 5. The six supplemental questionnaire items not part of the five CAOS subscales (but believed to help characterize clinicians’ beliefs about opioids and opioid prescribing) were examined for the entire sample using means and standard deviations. In addition, we sought to understand those more likely (Q4) and less likely (Q1) to report that 628
they would prescribe TRFs as operationalized by the TRFD subscale better. To do so, independent samples t-tests were estimated to assess significant differences on each of the six items between the extreme TRFD quartile groups. Results Sample Characteristics Overall, 47% of physicians invited to participate in the survey responded (N = 1,535). Response rates for the various specialties were as follows: Anesth (60%), PM (60%), G/FP (51%), IM (51%), Ortho (45%), PM&R (40%), Psych (40%), Surg (39%), Neurol (33%), and Rheum (15%). Demographic and descriptive characteristics of survey respondents are presented in (Table 1). This information is presented for the entire sample and after stratifying the sample according to the upper and lower quartiles of responses on TRFD. The majority of total sample of respondents were men (82.9%) in the age range of 45–60 years old (53.0%) and had been in practice for >19 years (31.7%) (for additional information about the total sample, see Wilson et al., 2013 [31]). A subset of 15.4% of participants practiced in one of the top 10 states and 19.5% in the bottom 10 states in mean milligram of opioids (in morphine equivalents) dispensed per resident. Sixteen percent were board certified in pain and almost two thirds (64.6%) practiced in a clinic or hospital setting. When comparing clinicians in the upper and lower quartiles of TRFD, a statistically significantly higher frequency of Q4 (highest) participants were board certified in Pain Medicine, and these respondents also reported a higher percentage of their weekly patients that have CNCP, of CNCP patients in which they prescribe opioids, and CNCP patients to whom they prescribe Schedule II opioids. Prediction of TRF Usage The distributions of all CAOS total subscale scores were examined, with mean values and bivariate correlations presented in (Table 2) for the entire sample, along with means stratified by the upper and lower TRFD quartile. All five subscales were normally distributed (skewness values lt; 0.04), thus no transformations of any of the variables were required. When considering the sample in its entirety, mean values indicated that participants have moderate beliefs about the effectiveness of opioids in addition to using Schedule II opioids, but their concerns about opioids and impediments to use surpass their beliefs about effectiveness. Interestingly, these individuals tended toward being only moderately satisfied with their education in pain management. Bivariate correlations indicated small to moderate associations among the CAOS subscales across the total sample, with the strongest associations evidenced between Schedule II vs III and Impediments, and TRFD and Effectiveness, indicating that clinicians that have elevated concerns about opioids concomitantly are less likely to prescribe Schedule II opioids,
Prescribing Tamper-Resistant Formulations
Table 1
Demographic and practice characteristics of survey respondents (N = 1,535)
Region of practice Pacific Northwest2 Southwest3 Midwest4 Southeast5 Northeast6 MOR Less than 9.5% 9.5–14.8% 14.9% and higher Gender (% male) Age 60 Years in practice 19 Board-certified pain (% yes)* Type of practice Solo Partnership Single specialty Multi-specialty Employment Setting Clinic/hospital/HMO Academic/Other Field of current practice Anesthesiology General/Family Medicine Internal Medicine Neurology Orthopedics Pain Management Physical Medicine/Rehab Surgery Rheumatology Psychiatry % of weekly patients involving CNCP* 0–10% 11–30% 31–50% >50% % of CNCP patients prescribe opioids* 0 1–10% 11–20% 21–30% 31–40% 41–50% >50%
Total Sample
TRF Quartile1 1
TRF Quartile 4
N or M
% or SD
N or M
% or SD
N or M
% or SD
236 69 283 466 481
15.4 4.5 18.4 30.4 31.3
52 21 80 139 131
12.3 5.0 18.9 32.9 31.0
59 12 59 118 118
16.1 3.3 16.1 32.2 32.2
404 803 328 1,272
26.3 52.3 21.4 82.9
113 217 93 353
26.7 51.3 22.0 83.5
103 180 83 305
24.3 54.6 21.1 83.3
592 813 130
38.5 53.0 8.5
151 234 38
35.7 55.3 9.0
133 202 31
36.3 55.2 8.5
379 324 346 486 245
24.7 21.1 22.5 31.7 16.0
105 76 105 137 46
24.9 18.0 24.8 32.4 10.9
75 88 87 116 78
20.5 24.0 23.8 31.7 21.3
367 319 450 399
23.9 20.8 29.3 26.0
88 88 150 97
20.8 20.8 35.5 22.9
88 82 96 100
24.0 22.4 26.2 27.3
1,022 513
66.6 33.4
270 153
63.8 36.2
246 120
67.2 32.8
197 69 573 77 122 49 45 158 47 198
12.8 4.5 37.3 5.0 7.9 3.2 2.9 10.3 3.0 12.9
46 13 142 22 56 4 12 41 19 68
10.9 3.1 33.6 5.2 13.2 0.9 2.8 9.7 4.5 16.1
55 16 148 19 21 11 11 32 7 46
15.0 4.4 40.4 5.2 5.7 3.0 3.0 8.7 1.9 12.6
497 598 210 230
32.4 39.0 13.6 15.0
168 159 47 49
39.8 37.6 11.1 11.5
91 164 46 65
24.9 44.8 12.6 17.7
175 359 201 180 116 151 353
11.4 23.4 13.1 11.7 7.6 9.8 23.0
90 130 43 40 23 36 61
21.3 30.7 10.2 9.5 5.4 8.5 14.1
22 70 53 42 32 34 113
6.0 19.1 14.5 11.5 8.7 9.3 30.9
629
Turk et al.
Table 1
Continued
% of CNCP patients prescribe Schedule* II opioids 0 1–10% 11–20% 21–30% 31–40% 41–50% >50%
Total Sample
TRF Quartile1 1
TRF Quartile 4
N or M
% or SD
N or M
% or SD
N or M
% or SD
235 402 178 144 110 96 370
15.3 26.2 11.5 9.4 7.2 6.3 24.1
120 113 42 34 18 27 69
28.4 26.7 9.9 8.0 4.3 6.4 16.3
29 86 48 27 32 28 116
7.9 23.5 13.1 7.4 8.7 7.7 31.7
* P < 0.05. 1 TRF Quartiles based on scores on the Tamper-resistant formulation and dosing subscale of the CAOS with Q1 = lowest interest and likelihood and Q4 = highest interest and likelihood of prescribing TRFs. 2 Pacific = Alaska, California, Hawaii, Oregon, Washington. 3 Southwest = Arizona, Colorado, Nevada, New Mexico, Utah. 4 Midwest = Iowa, Idaho, Illinois, Indiana, Kansas, Michigan, Minnesota, Montana, Nebraska, North Dakota, Ohio, South Dakota, Wisconsin, Wyoming. 5 Southeast = Alabama, Arkansas, Delaware, Florida, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, West Virginia. 6 Northeast = Connecticut, District of Columbia, Maine, Maryland, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, Vermont. CAOS = Clinicians’ Attitudes about Opioids Scale; CNCP = chronic noncancer pain; HMO = health maintenance organization; MOR = Mortality overdose by region; SD = standard deviation; TRF = tamper-resistant opioid formulations.
while participants who report an openness to using TRFs also report a belief in the effectiveness of opioids in general. The findings from the stepwise hierarchical multiple linear regression analysis are presented in (Table 3). To predict TRFD scores, respondent characteristics were entered into Step 1 of the regression model, followed sequentially by each of the four remaining CAOS total subscale scores in Steps 2–5 (i.e., Impediments, Effectiveness, Education, Schedule II vs III). The two clinician characteristics that were initially statistically significant
predictors in Step 1 of the model remained significant in the final model, including the percentage of CNCP patients to whom the clinician prescribes opioids and if the clinician is board certified in Pain Medicine (with the exception of patient demographics), where both characteristics are positively related to a reported likelihood of prescribing TRFs. MOR was not a significant predictor of TRFD scores. There was a statistically significant R2 change with the addition of each CAOS subscale when predicting TRFD, except for Schedule II vs III, accounting for 21.0% of the total variance in TRFD scores. Specifically, in the final model, an increased reporting of the
Table 2 Summary statistics for the five CAOS subscales for the entire sample and stratified by TRF quartile, and bivariate correlations among CAOS factors for the entire sample
TRFD2 Impediments Effectiveness Education Schedule II vs III
TRF Quartile1 1
TRF Quartile 4
Total Sample
M
SD
M
SD
M
SD
F1
F2
F3
F4
4.54 6.56 4.82 4.62 5.40
1.04 1.47 1.52 2.04 1.98
8.49 6.95 6.29 4.75 5.63
0.67 1.43 1.56 2.50 2.09
6.48 6.66 5.61 4.76 5.48
1.60 1.36 1.54 2.15 1.90
0.09* 0.39** 0.02 0.06
−0.24** 0.13** 0.39**
0.09** −0.01
0.10**
* P < 0.05; ** P < 0.01. 1 TRF quartiles based on scores on the Tamper-resistant formulation and dosing subscale of the CAOS with Q1 = lowest interest and likelihood Q4 the highest interest and likelihood of prescribing TRFs. 2 Tamper-resistant formulation and dosing subscale of CAOS. CAOS = Clinicians’ attitudes about opioids scale; SD = standard deviation; TRF = tamper-resistant opioid formulations; TRFD = tamper-resistant formulations and dosing.
630
0.11 0.11 0.10 0.12 0.08 0.01
0.32
−0.09 −0.02
−0.06 0.03
−0.02 0.06*
0.03 0.00
0.07**
0.02
0.06 21.43***
−0.12 −0.02 0.14
−0.10 −0.05
0.18
0.14
0.08 0.01 0.03
0.09 0.11
0.10
0.10
−0.04 −0.04 0.12***
−0.03 −0.01
0.04
0.03
0.21 270.94***
−0.12 −0.02 0.25 0.44
−0.10 −0.05
0.18
0.14
SE B B 0.03 −0.04 −0.05 0.02 0.10*** 0.06
Model 3
0.08 0.01 0.03 0.03
0.10 0.12
0.10
0.10
−0.10 −0.05
0.22
0.13
0.21 4.54*
−0.04 −0.11 −0.04 −0.02 0.21*** 0.26 0.42*** 0.44 0.04
−0.03 −0.01
0.04
0.03
SE B β B 0.03 −0.04 −0.06 0.02 0.10*** 0.06
Model 4
0.08 0.01 0.03 0.03 0.02
0.09 0.12
0.10
0.10
0.08 0.01 0.03 0.03 0.02 0.02
−0.10 0.09 −0.05 0.12
0.22 0.10
0.13 0.10
−0.03 −0.11 −0.04 −0.02 0.22*** 0.26 0.43*** 0.44 0.05* 0.04 0.00 0.21 0.01
−0.03 −0.01
0.05*
0.03
−0.03 −0.04 0.22*** 0.43*** 0.05* 0.00
−0.03 −0.01
0.05*
0.03
SE B β B SE B Β 0.03 −0.05* −0.06 0.03 −0.05* 0.02 0.10*** 0.06 0.02 0.10***
Model 5
* P < 0.05; ** P < 0.01; *** P < 0.001. 1 Subscales from CAOS, Clinicians’ Attitudes about Opioids Scale. 2 Patient demographics = “Patient demographics (for example age, sex, race) influence my prescribing opioids on a long-term basis.” 3 Reference group is designated as the states where the mean mg of opioids dispensed per resident fell within the Top 10 Highest and Top 10 Lowest prescribing states per resident. CNCP = Chronic noncancer pain; CAOS = Clinicians’ Attitudes about Opioids Scale; HMO = health maintenance organization; MOR = Mortality overdose region (the lowest category, less than 9.5%, is the reference category); SE = standard error.
0.05 10.31***
Model 2 SE B β B 0.03 −0.00 −0.05 0.02 0.18*** 0.06
0.09
B 0.00 0.11
Model 1
Summary of hierarchical linear regression analysis for variables predicting clinician prescribing of TRFs (N = 1,535)
Variable Years in practice % of CNCP prescribed opioid Gender Female Board-certified Pain Medicine Yes MOR 9.6% to 14.8% > 14.8% Employment setting Academic vs. Clinic/HMO Patient demographics Impediments1 Effectiveness1 Education1 Schedule II vs III1 R2 F for R2 change
Table 3
Prescribing Tamper-Resistant Formulations
631
Turk et al. effectiveness of opioids was the strongest predictor of the likelihood to prescribe TRFs, followed by reported concerns or impediments to opioid usage. Interestingly, a belief that education and training in pain management was appropriate was significantly related to an increased reported likelihood of prescribing TRFs. Sensitivity Analyses of Region Because MOR was not a significant predictor of the reported likelihood to prescribe TRFs, two sensitivity analyses were performed to understand the relationship between region and TRFD scores better. First, we recategorized the MOR variable by removing all states from the MOR categorization that had a participant sample size of less than 10, in order to remove cases that may be an over-representation of each state. This resulted in a removal of 80 cases surveyed in 18 states, with the MOR variable recategorized into three categories (top 10 states, middle 12 states, and bottom 10 states). The multiple regression analysis was then re-estimated with a sample of 1,455 participants, using this modified MOR variable in addition to all other predictors in the original regression analysis. Again, the model remained unchanged and MOR was not found to be a significant predictor of TRFD scores (results not reported). Next, the multiple regression analysis was re-estimated as initially specified, but using the QOR (quantity) categorization in place of MOR (mortality). Thus, we sought to test if practicing in a state with a higher mean of milligrams of opioids prescribed was predictive of potential TRF usage. Similar to mortality rates, quantity of opioids prescribed by region failed to be a significant predictor of TRFD scores and the strength of the associations between TRFD scores and
Table 4
any other predictors in the model did not significantly change (results not reported). Mean Differences Between Q4 and Q1 Responders The six supplemental questions were examined to characterize the total sample and the differences between individuals very likely and not very likely to prescribe TRFs. As displayed in Table 4, as might be expected, respondents falling in the upper quartile (Q4) of TRFD scores reported significantly higher mean values on all questions in comparison with those within the lower quartile (Q1). For example, those very likely to prescribe TRFs held stronger beliefs that patients take opioids for reasons other than their pain, their concerns about diversion influences their prescribing of opioids, and reported that increasing opioid-related mortalities has influenced their prescribing habits. Discussion The current study was undertaken to gain a better understanding of what factors influence physicians’ willingness and likelihood to use TRFs. To accomplish this, we administered the CAOS to a representative sample of physicians across the United States. In addition, we investigated whether beliefs about opioids and TRFs differed by the region of the country in which the physician practiced, to determine if physicians practicing in states with greater numbers of overdose deaths or where higher percentages of opioids are prescribed would be more disposed to prescribe TRFs for the management of CNCP. To our knowledge, this is the first investigation into the opinions physicians hold about TRFs. This study comprised a secondary analysis of results from a larger study [31].
Comparisons between clinicians by the upper and lower TRF subscale quartile
Question Supplemental items “The majority of my patients prescribed opioids for long periods of time take them as directed.” “Patients sometimes take opioids for reasons other than pain (eg, anxiety, stress, sleep)” “Concerns about diversion affect my willingness to prescribe opioids.” “Increasing opioid-related mortalities has influenced my opioid prescribing habits.” “My opioid prescribing is influenced by my pharmacy benefits package.” “I believe standard practice policies would be beneficial in helping to make decisions regarding the prescription of opioids for long-periods of time for certain chronic non-cancer pain patients.”
Total Sample Q11 (n = 423) Q4 (n = 346) Mean (SD) Mean (SD) Mean (SD) t-value P value
5.66 (2.06)
5.07 (2.17)
6.07 (2.14)
6.47
Pain Medicine 2014; 14: 625–636 Wiley Periodicals, Inc.
Physicians’ Beliefs and Likelihood of Prescribing Opioid Tamper-Resistant Formulations for Chronic Noncancer Pain Patients
Dennis C. Turk, PhD,* Elizabeth J. Dansie, PhD,† Hilary D. Wilson, PhD,† Bruce Moskovitz, MD,‡ and Myoung Kim, PhD‡
Subjects. A nationally representative sample of 1,535 practicing physicians throughout the United States.
*Department of Anesthesiology & Pain Medicine and Center for Pain Research on Impact, Measurement, & Effectiveness, University of Washington;
Methods. A stepwise hierarchical multiple linear regression analysis was conducted to estimate if physician characteristics, opinions, or geographic region categorized according to state rates of mortality by drug overdose and milligrams of opioids prescribed by state were predictive of the likelihood of prescribing TRFs.
†
Outcomes Research, United Biosource Corporation, Seattle, Washington; ‡
LLC, Health Economics & Outcomes Research, Janssen Scientific Affairs, Titusville, New Jersey, USA Reprint requests to: Dennis C. Turk, PhD, Department of Anesthesiology & Pain Medicine, University of Washington, Box 356540, Seattle, WA 98195, USA. Tel: (206) 616-2626; Fax: (206) 543 2958; E-mail: [email protected]. Dr. Turk has received grant support and/or consulting fees from Eli Lilly, OrthoMcNeil-Janssen, Pfizer, Phillips Respironics, the National Institutes of Health, and the United States Food and Drug Administration.
Abstract Background. Tamper-resistant opioid formulations (TRFs) have recently been the target of active development in an effort to deter opioid misuse and abuse.
Results. Board certification in Pain Medicine and prescribing opioids to a higher volume of CNCP patients were significantly predictive of a reported likelihood of prescribing TRFs, in addition to concerns about possible misuse and abuse of opioids, beliefs in the effectiveness of opioids for CNCP, and greater satisfaction with education and training in pain management this set of factors accounted for 21% of the model variance. Rates of mortality by drug overdose and opioid prescription volume by location were not predictive of TRF usage. Conclusions. Reducing physician concerns about potential misuse and abuse of opioids through additional education in pain management and dissemination of information about the potential benefits and availability of TRFs should influence physicians’ attitudes about and the adoption of TRFs. Key Words. Opioids; Tamper-Resistant Formulations; Prescribing Patterns; Attitudes; Chronic Noncancer Pain
Objective. To understand factors that are predictive of physicians’ likelihoods of prescribing TRFs to patients with chronic noncancer pain (CNCP).
Introduction
Design. A cross-sectional survey was conducted, utilizing a questionnaire of clinicians’ attitudes and opinions about opioids for CNCP (Clinicians’ Attitudes about Opioids Scale) to explore beliefs about and likelihood of prescribing TRFs.
The use of long-term opioid therapy (LOT) for the management of chronic noncancer pain (CNCP) has risen dramatically since the mid-1990s [1,2]. Although opioids are considered appropriate for acute and cancer-related pain, their benefits and risks are less well understood and 625
Turk et al. more controversial in the context of extended use for CNCP patients [3,4]. Opioids remain the most commonly prescribed category of drug in the United States [5,6], where 3% of adults are prescribed LOT [4]. Americans currently consume approximately 80% of the global opioid supply [7]; however, there is considerable variation in prescribing rates of opioids both across and within states and regions [8]. Available evidence reveals that the largest increase in opioid prescribing can be attributed to the use of Schedule II opioids for CNCP [9]. Corresponding to the rise in opioid prescribing is the expansion in rates of opioid misuse (i.e., inappropriate use of a medication for medical purposes, rather than mindaltering effects), abuse (i.e., use of medication for mindaltering effect or in any manner not consistent with the way in which the opioid was prescribed by the treating physician), as well as both fatal and nonfatal overdoses [7,10–12]. For example, in 2010, 4.8% of the US population aged ≥12 (∼14.4 million) used opioids nonmedically [13]. Nonadherence rates for prescribed opioids range from 2% to a high of 53% depending on how misuse and abuse are defined and measured [14], while recent data on illicit use indicate that prescription opioids account for more overdose deaths than heroin and cocaine combined [12,15]. This relationship is more than circumstantial, as multiple investigations have been conducted to support the purported causal influence of increased opioid prescribing to higher rates of misuse, abuse, emergency room visits, and overdose [12,16–18]. Unfortunately, education directed toward prescribers to assist them to develop the most appropriate pain management strategy designed for individual patients is limited [19]. In response to the current concerns about LOT, the United States Food and Drug Administration (FDA) has identified the need for the development of tamper-resistant formulations (TRFs, also known as abuse-deterrent formulations) of extended-release opioids. Significant efforts and financial investment by the pharmaceutical industry have been devoted to develop abuse-deterrent formulations, and these efforts are ongoing [20]. The intention for the development of TRFs is to treat pain in patients who might benefit from opioid medication while concomitantly minimizing the rate of misuse, abuse, and other negative outcomes, hence increasing the benefit–risk ratio for opioids [20–22]. Several TRFs are in development that utilize a range of techniques to combat misuse and abuse, such as formulations of opioids that: 1) have properties that provide physical barriers to tampering (e.g., barriers to crushing, chewing, or grinding); 2) contain agonist– antagonist formulations that interfere with the euphoria experienced when the drug is abused; 3) contain an additional substance that creates an aversive side effect for the individual if abused; 4) contain a prodrug (which hinders the opioid effectiveness until it is transformed in the gastrointestinal tract); or 5) are more difficult to misuse simply because of the route of administration [21,23]. At the present time, five TRFs are on the market: OpanaR ER (Oxymorphone extended release), OxectaR (Oxycodone), OxyContinR (Oxycodone 626
controlled release), ExalgoR (osmotic-controlled release oral delivery system [OROS] hydromorphone) [24], and NucyntaR ER (tapentadol extended release) [25] and multiple TRFs are in development. There are substantial financial incentives for the development and marketing of these products. A budget-impact model for prescription TRFs estimated the market in the United States alone could range from $600 million to $1.6 billion per year [26], demonstrating the potential economic impact if these formulations gained widespread acceptance. It is inevitable that TRFs approved by the FDA would be more costly than existing formulations. However, the economic burden of opioid misuse and abuse is substantial, with some cost estimates ranging from $50–$75 billion annually in lost productivity, criminal justice expenditures, drug abuse treatments, and medical complications [27,28]. Thus, prescribers and payers will have to balance the additional costs of TRFs against the benefits from lowered costs resulting from potential savings accruing for reduced misuse and abuse in making decisions about when and for whom TRFs are warranted. As an illustration of this dilemma, in April 2013, the FDA rejected a generic version of OxyContin, leaving only the reformulated, tamper-resistant OxyContin that is less prone to crushing, snorting, and other forms of abuse [29]. Beyond costs, several additional factors that are not completely independent might impact the adoption of TRFs if traditional formulations remain available. For example, clinicians’ own concerns about the potential for misuse or diversion of opioids, beliefs in their ability to identify “atrisk” patients, variability in beliefs about the overall effectiveness of opioids for managing chronic pain and improving function, and the education in pain management the physician has received could all influence opioid prescribing patterns. Moreover, there is large variability in prescribing patterns [8] and overdose deaths [30] from opioids across geographic states, making it possible that geographic location could influence who would be likely to use opioid formulations with a lower potential for aberrant behaviors. Although several opioid TRFs are currently marketed [20,24], no investigations have reported on the factors that determine their acceptability, which factors influence providers’ attitudes and beliefs about the necessity and value of TRFs, or the likelihood that clinicians will actually adopt these formulations [21]. Recently a new measure, the Clinicians’ Attitudes about Opioids Scale (CAOS), was developed to assess the attitudes and practices of providers who prescribe opioids for chronic pain patients [31]. One of the subscales of the CAOS specifically assesses clinicians’ attitudes and beliefs about TRFs. The purpose of the present study was to explore the factors that play a role in the acceptance and likelihood of prescribing TRFs in a nationally representative United States sample of physicians. The following hypotheses were addressed: 1) higher evaluation of the perceived effectiveness of opioids and concerns for the risk of misuse and abuse inherent in opioid prescribing would
Prescribing Tamper-Resistant Formulations be associated with a greater likelihood of prescribing TRFs; and 2) physicians practicing in states with greater numbers of deaths attributed to opioids would be more likely to prescribe TRFs, as they would be more aware of the potential negative outcomes associated with prescription opioid abuse due to medical, regulatory, and media attention, in turn increasing their likelihood of adopting potentially less abusable formulations. Methods Participants Participants were recruited in 2010 using a large, proprietary, Web-enabled panel of United States physicians that is maintained by the private company, Toluna, Inc. (Wilton, CT, USA). The panel comprised approximately 100,000 American Medical Association (AMA)-verified physicians who had been recruited by means of the Internet to join the panel through a “double-opt-in” process, where their targeting information, such as primary specialty, subspecialty, purchasing authority, and involvement in Pharmacy and Therapeutics committees, is stored. The panel is designed to be representative of the United States population of physicians by age, race/ethnicity, geographic region, and medical specialty according to the AMA. The company conducts an extensive validation of every panelist to guarantee all respondents are licensed and AMAverified physicians prior to extending invitations to participate in a survey. Panel physicians may withdraw membership at any time and are compensated for their participation at a rate Toluna has determined to commensurate with the length of the survey and medical specialty. All physicians on the Toluna panel in the following specialties were eligible to participate in our survey: Anesthesiologists (Anesth), General/Family Practitioners (G/FP), Neurologists (Neurol), Orthopedists (Ortho), Pain Medicine specialists (PM), Internists (IM), Physical Medicine and Rehabilitation specialists (PM&R), Psychiatrists (Psych), Rheumatologists (Rheum), and Surgeons (Surg). Panel members received notification of the survey through email and were compensated between $50–75 depending on their specialty. The study was approved by the Institutional Review Board of the University of Washington, and all participants gave informed consent. Measure CAOS was used to assess current beliefs regarding opioids, opioid use, and patients with CNCP. The threephase questionnaire development process and psychometric evaluation of this scale is described in detail elsewhere [30]. The CAOS consists of 38 content questions and 13 demographic and descriptive questions concerning providers’ practice type, volume of CNCP patients in their practice, and their opioid-related prescription practices. The response scale for all content questions is 0–10, with 0 being “strongly disagree” and 10 being “strongly agree.” A principal component analysis revealed that the questions comprised five subscales that measured clinicians’ attitudes regarding opioids, as
well as considerations that may influence the likelihood of prescribing opioids or TRFs of such opioids. Total subscale scores were created by summing the scores for each item within each subscale. The measure was shown to be internally consistent (Cronbach’s alpha = 0.87) and demonstrated high stability over a 2-week period (test–retest reliability coefficients for each subscale = 0.62–0.79). As described in detail by Wilson and colleagues [30], questions comprising each subscale did not load on multiple factors, while correlations between factors were small to moderate, indicating that each subscale measured unique aspects of clinicians’ attitudes and beliefs about prescribing. These scale characteristics and scale development substantiated the validity of our use of this scale in statistical models of association as detailed below. The five factors composing the CAOS are as follows: 1. Impediments and concerns (Impediments): A 16-item subscale with higher scores reflecting increased concerns (e.g., “Physical dependence is an impediment to taking opioids for long periods of time”; “Long-term use of opioids is over-prescribed for patients with chronic non-cancer pain”). 2. Perceived effectiveness (Effectiveness): A nine-item subscale with higher scores associated with increased beliefs in the efficacy of opioids for treatment of CNCP (e.g., “In general, opioids are effective for mixed pain”; “Opioids are the most effective treatments available for persistent pain”). 3. Medical education (Education): A three-item subscale with higher scores indicative of increased confidence in education/training received for the treatment of CNCP (i.e., “My education regarding pain evaluation and treatment during medical school is appropriate”; “My education regarding pain evaluation and treatment during residency was appropriate”; “My current practice involving the use of opioids for the treatment of chronic non-cancer pain is consistent with my training.”). 4. Schedule II vs III Opioids (Schedule II vs III): A five-item subscale with higher scores associated with decreased preference and increased concern surrounding Schedule II opioids (e.g., “Increased administrative burden prevents me from prescribing Schedule II opioids”; “Fear of DEA scrutiny inhibits me from prescribing Schedule II opioids”). 5. Tamper-resistant formulations and dosing (TRFD): A 3-item subscale with higher scores reflecting a preference for more controlled dosing and TRFs (i.e., “I would be more likely to prescribe opioids if TRF formulas were available”; “I would prescribe a TRF form over a generic standard form for my patients even if it was more expensive”; “I would try to prescribe TRF over the generic/standard form only for those patients with higher abuse potential”). In addition to the CAOS items, we asked respondents to complete six additional questions that, although did not measure the same constructs represented by the five derived subscales, were considered important when 627
Turk et al. describing clinicians prescribing habits such as “Patient demographics (e.g., age, sex, race) influence my prescribing opioids on a long-term basis,” “My opioid prescribing is influenced by my pharmacy benefits package,” and “Increasing opioid-related mortalities has influenced my opioid prescribing habits.” Finally, we created two variables to explore the influence of geographic region on TRF usage. First, we created a three-category “Mortality overdose by region” (MOR) variable by categorizing states according to the drug overdose rates by state (e.g., 14.9% and higher, 9.5–14.8%, and less than 9.5%) [29]. Second, we used McDonald, Carlson, and Izrael’s [8] ranking of the geographic states (in the United States) according to estimates of the mean milligrams (mg) of opioids (in morphine equivalents) dispensed per resident by state to create a three-category “Quantity of opioids by region” (QOR) variable to indicate if participants resided in a state with high or low quantities of dispensed opioids (e.g., top 10 vs middle 30 vs bottom 10 prescribing states). Statistical Analysis Demographic and descriptive characteristics of participating physicians, as well as the distribution of survey items and CAOS total subscale scores, were assessed using frequencies and percentages for categorical variables and means and standard deviations for continuous variables. Quartiles were calculated for clinicians’ total scores on the TRFD subscale, in order to categorize clinicians according to their reported likelihood of prescribing TRFs. In order to understand those clinicians most and least likely to use TRFs better, demographic characteristics of the sample were also calculated for those clinicians that scored within the upper (Q4) and lower quartiles (Q1) on the TRFD factor. Chi-square (χ2) analyses were performed to determine if the clinicians in Q4 or Q1 of TRFD differed with respect to demographic and practice characteristics (e.g., gender, age, region of practice, employment setting). Pearson’s r correlations were calculated to examine the intercorrelations among the five CAOS total subscales scores. A stepwise hierarchical multiple linear regression analysis was conducted to address our primary research question, namely, what physician characteristics, opinions, and beliefs are predictive of the reported likelihood of prescribing TRFs. Unstandardized and standardized (β) coefficients, along with standard errors and significance values are reported to assess the strength and significance of the reported associations. The fit of each model was assessed by examining the adjusted R2 value and F value for change in R2 when progressing from Model 1 to Model 5. The six supplemental questionnaire items not part of the five CAOS subscales (but believed to help characterize clinicians’ beliefs about opioids and opioid prescribing) were examined for the entire sample using means and standard deviations. In addition, we sought to understand those more likely (Q4) and less likely (Q1) to report that 628
they would prescribe TRFs as operationalized by the TRFD subscale better. To do so, independent samples t-tests were estimated to assess significant differences on each of the six items between the extreme TRFD quartile groups. Results Sample Characteristics Overall, 47% of physicians invited to participate in the survey responded (N = 1,535). Response rates for the various specialties were as follows: Anesth (60%), PM (60%), G/FP (51%), IM (51%), Ortho (45%), PM&R (40%), Psych (40%), Surg (39%), Neurol (33%), and Rheum (15%). Demographic and descriptive characteristics of survey respondents are presented in (Table 1). This information is presented for the entire sample and after stratifying the sample according to the upper and lower quartiles of responses on TRFD. The majority of total sample of respondents were men (82.9%) in the age range of 45–60 years old (53.0%) and had been in practice for >19 years (31.7%) (for additional information about the total sample, see Wilson et al., 2013 [31]). A subset of 15.4% of participants practiced in one of the top 10 states and 19.5% in the bottom 10 states in mean milligram of opioids (in morphine equivalents) dispensed per resident. Sixteen percent were board certified in pain and almost two thirds (64.6%) practiced in a clinic or hospital setting. When comparing clinicians in the upper and lower quartiles of TRFD, a statistically significantly higher frequency of Q4 (highest) participants were board certified in Pain Medicine, and these respondents also reported a higher percentage of their weekly patients that have CNCP, of CNCP patients in which they prescribe opioids, and CNCP patients to whom they prescribe Schedule II opioids. Prediction of TRF Usage The distributions of all CAOS total subscale scores were examined, with mean values and bivariate correlations presented in (Table 2) for the entire sample, along with means stratified by the upper and lower TRFD quartile. All five subscales were normally distributed (skewness values lt; 0.04), thus no transformations of any of the variables were required. When considering the sample in its entirety, mean values indicated that participants have moderate beliefs about the effectiveness of opioids in addition to using Schedule II opioids, but their concerns about opioids and impediments to use surpass their beliefs about effectiveness. Interestingly, these individuals tended toward being only moderately satisfied with their education in pain management. Bivariate correlations indicated small to moderate associations among the CAOS subscales across the total sample, with the strongest associations evidenced between Schedule II vs III and Impediments, and TRFD and Effectiveness, indicating that clinicians that have elevated concerns about opioids concomitantly are less likely to prescribe Schedule II opioids,
Prescribing Tamper-Resistant Formulations
Table 1
Demographic and practice characteristics of survey respondents (N = 1,535)
Region of practice Pacific Northwest2 Southwest3 Midwest4 Southeast5 Northeast6 MOR Less than 9.5% 9.5–14.8% 14.9% and higher Gender (% male) Age 60 Years in practice 19 Board-certified pain (% yes)* Type of practice Solo Partnership Single specialty Multi-specialty Employment Setting Clinic/hospital/HMO Academic/Other Field of current practice Anesthesiology General/Family Medicine Internal Medicine Neurology Orthopedics Pain Management Physical Medicine/Rehab Surgery Rheumatology Psychiatry % of weekly patients involving CNCP* 0–10% 11–30% 31–50% >50% % of CNCP patients prescribe opioids* 0 1–10% 11–20% 21–30% 31–40% 41–50% >50%
Total Sample
TRF Quartile1 1
TRF Quartile 4
N or M
% or SD
N or M
% or SD
N or M
% or SD
236 69 283 466 481
15.4 4.5 18.4 30.4 31.3
52 21 80 139 131
12.3 5.0 18.9 32.9 31.0
59 12 59 118 118
16.1 3.3 16.1 32.2 32.2
404 803 328 1,272
26.3 52.3 21.4 82.9
113 217 93 353
26.7 51.3 22.0 83.5
103 180 83 305
24.3 54.6 21.1 83.3
592 813 130
38.5 53.0 8.5
151 234 38
35.7 55.3 9.0
133 202 31
36.3 55.2 8.5
379 324 346 486 245
24.7 21.1 22.5 31.7 16.0
105 76 105 137 46
24.9 18.0 24.8 32.4 10.9
75 88 87 116 78
20.5 24.0 23.8 31.7 21.3
367 319 450 399
23.9 20.8 29.3 26.0
88 88 150 97
20.8 20.8 35.5 22.9
88 82 96 100
24.0 22.4 26.2 27.3
1,022 513
66.6 33.4
270 153
63.8 36.2
246 120
67.2 32.8
197 69 573 77 122 49 45 158 47 198
12.8 4.5 37.3 5.0 7.9 3.2 2.9 10.3 3.0 12.9
46 13 142 22 56 4 12 41 19 68
10.9 3.1 33.6 5.2 13.2 0.9 2.8 9.7 4.5 16.1
55 16 148 19 21 11 11 32 7 46
15.0 4.4 40.4 5.2 5.7 3.0 3.0 8.7 1.9 12.6
497 598 210 230
32.4 39.0 13.6 15.0
168 159 47 49
39.8 37.6 11.1 11.5
91 164 46 65
24.9 44.8 12.6 17.7
175 359 201 180 116 151 353
11.4 23.4 13.1 11.7 7.6 9.8 23.0
90 130 43 40 23 36 61
21.3 30.7 10.2 9.5 5.4 8.5 14.1
22 70 53 42 32 34 113
6.0 19.1 14.5 11.5 8.7 9.3 30.9
629
Turk et al.
Table 1
Continued
% of CNCP patients prescribe Schedule* II opioids 0 1–10% 11–20% 21–30% 31–40% 41–50% >50%
Total Sample
TRF Quartile1 1
TRF Quartile 4
N or M
% or SD
N or M
% or SD
N or M
% or SD
235 402 178 144 110 96 370
15.3 26.2 11.5 9.4 7.2 6.3 24.1
120 113 42 34 18 27 69
28.4 26.7 9.9 8.0 4.3 6.4 16.3
29 86 48 27 32 28 116
7.9 23.5 13.1 7.4 8.7 7.7 31.7
* P < 0.05. 1 TRF Quartiles based on scores on the Tamper-resistant formulation and dosing subscale of the CAOS with Q1 = lowest interest and likelihood and Q4 = highest interest and likelihood of prescribing TRFs. 2 Pacific = Alaska, California, Hawaii, Oregon, Washington. 3 Southwest = Arizona, Colorado, Nevada, New Mexico, Utah. 4 Midwest = Iowa, Idaho, Illinois, Indiana, Kansas, Michigan, Minnesota, Montana, Nebraska, North Dakota, Ohio, South Dakota, Wisconsin, Wyoming. 5 Southeast = Alabama, Arkansas, Delaware, Florida, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, West Virginia. 6 Northeast = Connecticut, District of Columbia, Maine, Maryland, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, Vermont. CAOS = Clinicians’ Attitudes about Opioids Scale; CNCP = chronic noncancer pain; HMO = health maintenance organization; MOR = Mortality overdose by region; SD = standard deviation; TRF = tamper-resistant opioid formulations.
while participants who report an openness to using TRFs also report a belief in the effectiveness of opioids in general. The findings from the stepwise hierarchical multiple linear regression analysis are presented in (Table 3). To predict TRFD scores, respondent characteristics were entered into Step 1 of the regression model, followed sequentially by each of the four remaining CAOS total subscale scores in Steps 2–5 (i.e., Impediments, Effectiveness, Education, Schedule II vs III). The two clinician characteristics that were initially statistically significant
predictors in Step 1 of the model remained significant in the final model, including the percentage of CNCP patients to whom the clinician prescribes opioids and if the clinician is board certified in Pain Medicine (with the exception of patient demographics), where both characteristics are positively related to a reported likelihood of prescribing TRFs. MOR was not a significant predictor of TRFD scores. There was a statistically significant R2 change with the addition of each CAOS subscale when predicting TRFD, except for Schedule II vs III, accounting for 21.0% of the total variance in TRFD scores. Specifically, in the final model, an increased reporting of the
Table 2 Summary statistics for the five CAOS subscales for the entire sample and stratified by TRF quartile, and bivariate correlations among CAOS factors for the entire sample
TRFD2 Impediments Effectiveness Education Schedule II vs III
TRF Quartile1 1
TRF Quartile 4
Total Sample
M
SD
M
SD
M
SD
F1
F2
F3
F4
4.54 6.56 4.82 4.62 5.40
1.04 1.47 1.52 2.04 1.98
8.49 6.95 6.29 4.75 5.63
0.67 1.43 1.56 2.50 2.09
6.48 6.66 5.61 4.76 5.48
1.60 1.36 1.54 2.15 1.90
0.09* 0.39** 0.02 0.06
−0.24** 0.13** 0.39**
0.09** −0.01
0.10**
* P < 0.05; ** P < 0.01. 1 TRF quartiles based on scores on the Tamper-resistant formulation and dosing subscale of the CAOS with Q1 = lowest interest and likelihood Q4 the highest interest and likelihood of prescribing TRFs. 2 Tamper-resistant formulation and dosing subscale of CAOS. CAOS = Clinicians’ attitudes about opioids scale; SD = standard deviation; TRF = tamper-resistant opioid formulations; TRFD = tamper-resistant formulations and dosing.
630
0.11 0.11 0.10 0.12 0.08 0.01
0.32
−0.09 −0.02
−0.06 0.03
−0.02 0.06*
0.03 0.00
0.07**
0.02
0.06 21.43***
−0.12 −0.02 0.14
−0.10 −0.05
0.18
0.14
0.08 0.01 0.03
0.09 0.11
0.10
0.10
−0.04 −0.04 0.12***
−0.03 −0.01
0.04
0.03
0.21 270.94***
−0.12 −0.02 0.25 0.44
−0.10 −0.05
0.18
0.14
SE B B 0.03 −0.04 −0.05 0.02 0.10*** 0.06
Model 3
0.08 0.01 0.03 0.03
0.10 0.12
0.10
0.10
−0.10 −0.05
0.22
0.13
0.21 4.54*
−0.04 −0.11 −0.04 −0.02 0.21*** 0.26 0.42*** 0.44 0.04
−0.03 −0.01
0.04
0.03
SE B β B 0.03 −0.04 −0.06 0.02 0.10*** 0.06
Model 4
0.08 0.01 0.03 0.03 0.02
0.09 0.12
0.10
0.10
0.08 0.01 0.03 0.03 0.02 0.02
−0.10 0.09 −0.05 0.12
0.22 0.10
0.13 0.10
−0.03 −0.11 −0.04 −0.02 0.22*** 0.26 0.43*** 0.44 0.05* 0.04 0.00 0.21 0.01
−0.03 −0.01
0.05*
0.03
−0.03 −0.04 0.22*** 0.43*** 0.05* 0.00
−0.03 −0.01
0.05*
0.03
SE B β B SE B Β 0.03 −0.05* −0.06 0.03 −0.05* 0.02 0.10*** 0.06 0.02 0.10***
Model 5
* P < 0.05; ** P < 0.01; *** P < 0.001. 1 Subscales from CAOS, Clinicians’ Attitudes about Opioids Scale. 2 Patient demographics = “Patient demographics (for example age, sex, race) influence my prescribing opioids on a long-term basis.” 3 Reference group is designated as the states where the mean mg of opioids dispensed per resident fell within the Top 10 Highest and Top 10 Lowest prescribing states per resident. CNCP = Chronic noncancer pain; CAOS = Clinicians’ Attitudes about Opioids Scale; HMO = health maintenance organization; MOR = Mortality overdose region (the lowest category, less than 9.5%, is the reference category); SE = standard error.
0.05 10.31***
Model 2 SE B β B 0.03 −0.00 −0.05 0.02 0.18*** 0.06
0.09
B 0.00 0.11
Model 1
Summary of hierarchical linear regression analysis for variables predicting clinician prescribing of TRFs (N = 1,535)
Variable Years in practice % of CNCP prescribed opioid Gender Female Board-certified Pain Medicine Yes MOR 9.6% to 14.8% > 14.8% Employment setting Academic vs. Clinic/HMO Patient demographics Impediments1 Effectiveness1 Education1 Schedule II vs III1 R2 F for R2 change
Table 3
Prescribing Tamper-Resistant Formulations
631
Turk et al. effectiveness of opioids was the strongest predictor of the likelihood to prescribe TRFs, followed by reported concerns or impediments to opioid usage. Interestingly, a belief that education and training in pain management was appropriate was significantly related to an increased reported likelihood of prescribing TRFs. Sensitivity Analyses of Region Because MOR was not a significant predictor of the reported likelihood to prescribe TRFs, two sensitivity analyses were performed to understand the relationship between region and TRFD scores better. First, we recategorized the MOR variable by removing all states from the MOR categorization that had a participant sample size of less than 10, in order to remove cases that may be an over-representation of each state. This resulted in a removal of 80 cases surveyed in 18 states, with the MOR variable recategorized into three categories (top 10 states, middle 12 states, and bottom 10 states). The multiple regression analysis was then re-estimated with a sample of 1,455 participants, using this modified MOR variable in addition to all other predictors in the original regression analysis. Again, the model remained unchanged and MOR was not found to be a significant predictor of TRFD scores (results not reported). Next, the multiple regression analysis was re-estimated as initially specified, but using the QOR (quantity) categorization in place of MOR (mortality). Thus, we sought to test if practicing in a state with a higher mean of milligrams of opioids prescribed was predictive of potential TRF usage. Similar to mortality rates, quantity of opioids prescribed by region failed to be a significant predictor of TRFD scores and the strength of the associations between TRFD scores and
Table 4
any other predictors in the model did not significantly change (results not reported). Mean Differences Between Q4 and Q1 Responders The six supplemental questions were examined to characterize the total sample and the differences between individuals very likely and not very likely to prescribe TRFs. As displayed in Table 4, as might be expected, respondents falling in the upper quartile (Q4) of TRFD scores reported significantly higher mean values on all questions in comparison with those within the lower quartile (Q1). For example, those very likely to prescribe TRFs held stronger beliefs that patients take opioids for reasons other than their pain, their concerns about diversion influences their prescribing of opioids, and reported that increasing opioid-related mortalities has influenced their prescribing habits. Discussion The current study was undertaken to gain a better understanding of what factors influence physicians’ willingness and likelihood to use TRFs. To accomplish this, we administered the CAOS to a representative sample of physicians across the United States. In addition, we investigated whether beliefs about opioids and TRFs differed by the region of the country in which the physician practiced, to determine if physicians practicing in states with greater numbers of overdose deaths or where higher percentages of opioids are prescribed would be more disposed to prescribe TRFs for the management of CNCP. To our knowledge, this is the first investigation into the opinions physicians hold about TRFs. This study comprised a secondary analysis of results from a larger study [31].
Comparisons between clinicians by the upper and lower TRF subscale quartile
Question Supplemental items “The majority of my patients prescribed opioids for long periods of time take them as directed.” “Patients sometimes take opioids for reasons other than pain (eg, anxiety, stress, sleep)” “Concerns about diversion affect my willingness to prescribe opioids.” “Increasing opioid-related mortalities has influenced my opioid prescribing habits.” “My opioid prescribing is influenced by my pharmacy benefits package.” “I believe standard practice policies would be beneficial in helping to make decisions regarding the prescription of opioids for long-periods of time for certain chronic non-cancer pain patients.”
Total Sample Q11 (n = 423) Q4 (n = 346) Mean (SD) Mean (SD) Mean (SD) t-value P value
5.66 (2.06)
5.07 (2.17)
6.07 (2.14)
6.47
