A Strategy to Reduce Opioid Drug Prescribing by Clinicians

Overview

Under the auspices of the Oral Health Section of the NC Department of Health and Human Services clinician-investigators from the UNC Oral and Maxillofacial Surgery Department are conducting a survey to assess current analgesic prescribing practices in NC with a goal of eventually reaching consensus among clinicians for a wide range of procedures and conditions where pain control is important for successful patient outcomes. We hope that about 850 dentists and 600 physicians will agree to take part in this research study. Participation in this Qualtrics survey e-mailed in mid-February 2019 will take about 15 minutes or less.

METHODS Participants: Dentists, emergency room physicians and primary care physicians who are in active practice will be surveyed using the lists of active practitioners with Drug Enforcement Administration(DEA) licenses provided by the NC Department of Public Health. Inclusion criteria for dentists: Active dentist with DEA license including: endodontists, general practice dentists, oral and maxillofacial surgeons, and periodontists. Given the small numbers per specialty, all endodontists (n=128), periodontists(n=114), and oral and maxillofacial surgeons (n=178) will be surveyed. A stratified random sample (n = 400) of general dentists (n=3905) will be selected from the list using region of the state as the stratification factor. Inclusion criteria for emergency room clinical staff: MD/DO/PA with a primary area of practice as emergency medicine or urgent care and area(s) of practice as emergency medicine, urgent care, family medicine, or adolescent \& young adult medicine All DO (n = 191) will be surveyed. A stratified random sample (n= 400) of MD's (N = 1407) and PA's will be selected from the list using region of the state as the stratification factor. Qualtrics Survey: The survey will be divided into two general areas: 1) personal and practice demographics (sex, age, time in practice (years), practice size) and 2) clinical scenarios typical of patients seen by the respective practitioners who might be expected to experience acute pain. For each of the scenarios, practitioners will be asked the type(s) of medication that would be prescribed, the strength in mg, and the number of doses for each medication. The survey will be pre-tested by a sample of potential survey subjects who will be asked to provide open-ended comments regarding ease of completion, confusing items, and word changes. The survey methods of Salant and Dillman will be used as a guide.(8) A Qualtics linkage file will be created accessible only to the project investigators. The linkage file will only be used to identify non-respondents for follow-up. Surveys will be coded to maintain respondents' confidentiality. Linkage file will be in a pass word protected file separate from survey responses. Distribution: A mixed mode of distribution of the surveys will be used. Qualtrics software (Qualtrics, Provo, UT) will be used to create electronic versions of the survey to be sent to providers via email with an accompanying letter describing the purpose of the project and how confidentiality will be maintained. Teleform (Cardiff Software, Vista, CA) will be used as a paper alternative for providers who do not wish to respond to the electronic version. For these individuals a cover letter will be included describing the study along with a postage paid return envelope. Up to three attempts will be made at two-week intervals, depending on response versus non-response, to optimize subject participation.(8) The mode for the third round will be determined by the response rate from each of the first two rounds. Statistical Analysis: All survey responses will be de-identified to maintain provider confidentiality. The electronic and Teleform responses will be merged and analyzed using SAS (version 9.4, Cary, NC). Respondents will be excluded from data analysis if they do not complete the data for the prescribing practices.The primary outcome variable will be the clinicians prescribing opioid drugs for each survey scenario or not. Secondary analyses will explore possible explanatory variables for the primary outcome. Bivariate analysis will be done by status (degree and practice mode) to explore the effect of the explanatory variables (age; gender; time in practice; practice location; and practice size) on the preferred medication option for each clinical scenario. Comparisons among degree/specialty holders is not possible given that the scenarios presented are unique to the different groups. The use of Chi-square or Fisher's Exact test for the bivariate analysis will be determined based on the cell frequencies. Level of significance was set at 0.05.