Characteristics of the Sample
A total of 111 of 178 surveys were returned, or 62.3% (range for individual sites of 46.6% to 100%). Respondents included 67 attending physicians (60 family practice, 4 internal medicine, 3 osteopathy), 19 nurse practitioners (NPs), 3 physician assistants (PAs), and 22 family practice residents. The sample was 55% female and 82% white. A mean of 37.5% of adult patients seen in the targeted week across all providers was reported as having a current chronic pain complaint. About one-fourth of patients had back pain (23.6%), followed by joint pain (17.1%), headache (12.1%), generalized pain (7.8%), neck pain (7.5%), abdominal pain (6.8%), fibromyalgia (5.7%), and arm pain (5.2%). Pelvic pain, neuropathic pain, and complex regional pain syndrome were each reported at less than 5%.
Problems Providing Optimum Pain Relief
Respondents rated 11 possible problems in providing effective pain treatment for their patients (). The majority of providers rated patient self-management, patient psychological factors, and patient compliance as frequently or always preventing optimal pain treatment. In contrast, much smaller proportions of providers rated any of the 5 provider or system factors as frequently or always implicated in failure to help patients obtain optimal relief.
Use of Opioids
Providers were asked to rate their likelihood of prescribing opioids for chronic pain when other treatments were ineffective on a scale of 0=not at all likely to 4=very likely. Only 28.8% rated themselves as highly likely to prescribe opioids (rating of “3” or “4”). They also rated issues preventing prescribing opioids for chronic pain (). Fear of the patient becoming addicted, followed by fear of patients selling the opioids were noted as the strongest reasons prevention opioid prescribing. Few respondents indicated a large effect of law enforcement scrutiny on their willingness to prescribe opioids.
Analyses were conducted to compare demographic characteristics, and ratings of treatment problems, opioid issues, and co-occuring patient psychosocial characteristics, for providers reported less likely to prescribe opioids (n =72, rating of 0 to 2), compared with those highly likely to prescribe opioids (n =32, rating of 3 or 4). No significant demographic differences (gender, race, provider type, or practice site) were found. However, there was a trend (P =.10) for a higher proportion of attendings to be more likely to prescribe opioids (35.8%), than residents (22.7%) or NP/PAs (13.6%).
In the analysis of treatment problems and opioid issues, using t tests to compare mean rating scores on each item, providers less likely to prescribe opioids rated patient psychological factors as a significantly greater problem than those more likely to prescribe (2.91 vs 2.59, P =.019, ES=0.51), and reported a higher rating for “difficulty coordinating or adding on pain management/treatment” (2.38 vs 1.97, P =.03, ES=0.50). They also rated potential for patient addiction as more important in preventing opioid prescribing (M =2.72 vs 2.16, P =.004, ES=0.67). However, analysis of patient co-occuring psychosocial characteristics found a counter-intuitive association between likelihood of prescribing opioids and estimates of the percent of pain patients who had substance abuse histories (χ2 =4.64, P =.031). One-quarter of providers (25.8%) reporting higher likelihood of prescribing opioids indicated that greater than 50% of their patients with pain also had substance abuse histories, compared with 9.6% of providers less likely to prescribe.
To further understand correlates of high versus low likelihood of opioid prescribing, a logistic regression was conducted with provider type (attending vs others), and the 4 variables described above as significantly associated with opioid prescribing as independent variables. Four variables remained independently associated with likelihood of prescribing, 2 indicating increased likelihood and 2 indicating decreased likelihood: (1) provider type (attendings more likely to prescribe, odds ratio [OR]=5.3, P =.012; 95% confidence interval [CI]=1.4 to 19.3); (2) rating a higher percent of patients with substance abuse histories, OR=4.3, P =.013 (95% CI=1.4 to 13.8); (3) rating patient psychological factors as a greater problem, OR=0.36, P =.029 (95% CI=0.14 to 0.90); and (4) rating potential for patient addiction as more of an issue, OR=0.31, P =.002 (95% CI=0.15 to 0.64).
Satisfaction with Treatment and Training
Providers rated satisfaction with treating their patients with chronic pain as quite low (M =1.90, SD=0.81), on a scale where 0=not at all satisfied and 4=very satisfied. However, a higher proportion (37.5%) of providers who rated themselves as more likely to prescribe opioids also rated themselves as highly satisfied with their pain care (satisfaction rating of 3 or 4), while only 16.7% of providers who were least likely to prescribe opioids rated themselves as highly satisfied (P =.03). On a scale of 0=insufficient, 1=adequate, 2=good, and 3=very good, the mean rating of chronic pain education for NP/PA programs was 0.5 (SD=0.80), for medical school 0.35 (SD=0.67), and for postgraduate medical education 0.7 (SD=0.84). The majority of attending physicians rated their medical school education (81.5%), and residency training (54.7%), about chronic pain treatment as insufficient. Residents rated their undergraduate and graduate medical education on chronic pain as somewhat better, with less than half rating each as insufficient (47.6% and 42.1%, respectively).