Our study sample included a 1.5% (3435) geographically representative sample of AMA master file physicians from four major groups: Primary Care, Pediatrics, Emergency Medicine, and Specialists. One hundred forty-six (4.3%) individuals from our sample were either deceased or retired and 442 (12.8%) were unreachable because of incorrect contact information. This left 2847 eligible physicians for our study. Two hundred eighty-eight physicians (10.1%) refused to participate, while 69.2% (n = 1971) were non-responders. Thus, a total of 588 (20.6% of eligible) physicians responded to the survey. Two hundred sixty-six physicians (9.3%) responded after the 1st mailing. Telephone numbers were available for 1636 physicians, (57%) of whom were faxed the questionnaire; 199 (7.0%) responded. An additional 4.3% responded after the 2nd (n = 64) and 3rd (n = 59) mailing.
Demographic characteristics of the responders compared to the non-responders can be found in Table . Responders were more likely to be white, female, younger, and less likely to be a primary care physician. No significant differences were found between responders and non-responders regarding region of residence in the United States or the nature of laws pertaining to syringe acquisition in their state.
Demographic data comparing respondents to non-respondents (N = 3435)
Responses to questions related to IDUs in physicians' practices, as well as physicians' knowledge, attitudes, and practices related to drug users are found in Table . Forty percent reported having IDUs in their practice, but only 28% reported that they always raise the issue of drug use when reviewing a patient's medical history. Forty-seven percent reported that they would consider prescribing syringes to drug-injecting patients, though only 6 physicians reported they had actually done so without any other indication of medical need other than injection drug use, such as diabetes. Few (7.4%) reported knowing of other physicians who prescribed syringes. Most (79%) respondents did not report knowing whether prescribing syringes was legal in their state. Over 90% attributed the sharing of syringes among IDUs to barriers to access, but most also responded that the behavior was attributable to the influence of "the drug culture" and lack of information. When asked about syringe availability in their home state, either through needle exchange programs (NEP), or over the counter (OTC) pharmacy sales, the majority (>50% across all categories of respondents) reported "don't know" (data not shown).
Univariate responses to survey questions
Physicians willing to prescribe syringes were similar to others in gender, age, year of graduation, and specialty. Geographical differences were more pronounced, however, with those practicing in the South and Midwest being less likely to consider prescribing than those working in the West and Northeast (Table ). Physicians who reported having IDUs in their practice were 1.5 times (95% CI: 1.02–2.29) more likely to consider prescribing syringes and 2.9 times (95% CI: 1.13–7.39) more likely if they asked their patients about syringe sharing. The strongest predictor of being willing to prescribe was the physician's belief that IDUs share because they do not have access to sterile syringes, both in the univariate and multivariate model (OR 4.6; 95% CI 2.0–10.9 and AOR 4.75; 95% CI 2.0–11.4 respectively). Physicians who reported knowing how/where syringes were legally available to IDUs in their area were significantly more likely to be willing to prescribe syringes than those who did not know or who believed there was no legal access. This held true both in univariate and multivariate analyses. Willingness to prescribe syringes was significantly associated with having prescribed syringes to a person with diabetes who might be an IDU (OR, 2.8; 95% CI, 1.3–6.2) and knowing someone who prescribes syringes to IDUs (OR, 3.8; 95% CI, 1.5–10.1); although in the final model, prescribing syringes to an IDU person with diabetes was marginally significant (i.e., included 1.0 in the confidence interval (AOR, 2.3; 95% CI, 1.0–5.2)).
Bivariate and multivariate analyses comparing physicians willing to prescribe syringes by other variables
Table presents the results from TPB analyses. Willingness to prescribe syringes was significantly associated with each of the theory of planned behavior domains, with perceived control being the strongest predictor, both alone (OR = 2.9; CI: 2.4–3.7) and in the final model (OR = 2.0; CI: 2.0–2.5). Attitudes towards care for IDUs and perceptions of peer attitudes were also robust predictors of physician willingness to participate in syringe prescription programs. Physicians who regarded prescribing as an important tool for preventing HIV, and those who felt strongly enough to approve of "civil disobedience" if the practice were illegal, were more likely to be willing to consider prescribing than those who did not. Conversely, those who would be embarrassed about prescribing, and who felt their peers would disapprove of prescribing or breaking the law, were less likely to be willing to participate in such programs.
Theory of Planned Behavior Predictors (n = 427)