Overview of Study
We identified spine and pain clinics serving NC residents; this included 3 practices located in Virginia adjacent to sparsely populated areas in the north of NC. We then contacted the practice managers (or similarly qualified respondents) at these clinics to determine eligibility for participation. To be eligible, the respondent had to confirm that the clinic treated patients with chronic (defined as 3 months or more) back and neck pain. Respondents at eligible clinics were then surveyed to gather information about the organization of the clinic and the types of patients served. This study was approved by the Institutional Review Board of the University of North Carolina at Chapel Hill.
Development of Survey
The study team, which consisted of an internal medicine physician, physical therapist, registered nurse, social worker, epidemiologist, and economist, developed and reviewed the survey, which gathered information on organizational characteristics of the clinics, including: location, ownership, numbers of physician and nonphysician providers, average number of patient visits per week, average charge for an initial evaluation, types of problems treated at the clinic, and patient insurance. The survey also gathered specific information on types of physician and nonphysician providers at the clinic, and the various services offered. To reduce respondent burden, we created check lists of the various providers and services that could be offered, including an “other” category.
Identification of Spine and Pain Clinics
Two approaches were used to identify spine and pain clinics (): (1) surveying NC residents with chronic back and/or neck pain and (2) searching the NC yellow pages.
Flow diagram showing survey recruitment, eligibility, and response.
We first identified clinics by surveying a representative sample of NC adults 21 years and older (n = 873) who reported having chronic back and/or neck pain. These individuals were participants in a population-based phone survey we conducted on characteristics and care-seeking behaviors of NC residents with chronic back and neck pain. We defined chronic pain as: (1) pain and activity limitations nearly every day for the past 3 months; or (2) more than 24 episodes of pain that limited activity for 1 day or more in the past year. Subjects were asked whether, within the previous year, they had been seen at a spine or pain clinic— defined as a practice with multiple practitioners from different specialties who specialize in treating back/neck pain. Subjects who were seen at a spine/pain clinic (n = 120) provided a clinic name, location, and/or name of the treating provider. On the basis of this information, we identified 62 unique potential spine/pain clinics. We attempted to contact these clinics primarily by telephone, and by email, fax, and mail. Eleven could not be contacted after at least 6 attempts, 10 were contacted but found ineligible (i.e., did not treat chronic back and neck pain), and 1 refused participation. The response rate for this mode of recruitment was 77% (40 of 52 eligible or unknown-eligibility sites).
Our second approach was a computerized search of the NC yellow pages for multidisciplinary spine or pain practices. Single-specialty practices, such as chiropractic or orthopedic practices treating back and neck pain, were not sampled. This generated 166 yellow pages listings that were independently examined by 2 reviewers (T.C. and A.J.); 151 (91%) of these listings were found to be duplicates and/or had already been identified by our first identification approach described above. There were many duplicates with this approach because clinics were often listed repeatedly under multiple counties within driving distance. Of the 15 remaining sites, 4 were nonresponders, 5 were ineligible, 0 refused, and 6 responded. The response rate was 60% (6 of 10 eligible). Surveys were completed by a total of 46 clinics with an overall response rate of 75%.
The survey was primarily administered by phone, but respondents were also given the options of email, fax, and regular mail as secondary modes of survey completion. By phone, the survey took 15 to 20 minutes. Sites contacted by phone were offered $25 to participate. Sites that did not respond to phone queries were provided $20 cash in a hard copy mailing of the survey that solicited their participation.
All analyses were conducted using SAS software version 9.1.17
We used frequencies, proportions, and distributions to describe the data.
To facilitate analyses, we grouped the 19 provider types based on the results of the descriptive analyses and the clinical experience of the team. We created 2 groupings according to function (9 categories) or discipline (4 categories). The 9-category grouping comprised (1) surgeons (orthopedic and neurosurgeons), (2) anesthesiologists, (3) physical medicine and rehabilitation physicians (PM&R), (4) primary care physicians (family medicine, internal medicine), (5) medical specialty physicians (neurologists, rheumatologists, occupational medicine), (6) nonphysician allopathic clinicians (nurse practitioners, physician assistants, clinical pharmacists/PharmDs, nutritionists), (7) physical therapists, (8) mental health professionals (psychiatrists, psychologists), and (9) alternative medicine practitioners (massage therapists, acupuncturists, chiropractors). The 4-category grouping comprised (1) allopathic medicine (surgeons, anesthesiologists, PM&R, primary care, medical specialists, and nonphysician allopathic clinicians), (2) physical therapists, (3) mental health practitioners, and (4) alternative medicine practitioners.
Similarly, we grouped the 35 possible services offered into 8 categories: imaging, physical methods, exercise/nutrition/ education, alternative therapies, injections, outpatient procedures, medications, and off-site services.
We assessed associations between provider and service types by examining cell count proportions for 2*2 tables within each potential provider-service combination. To examine bivariate associations, we used Fisher exact tests for differences in proportions.
We created several variables to measure multidisciplinary function. On the basis of the Shealy and Cady criteria,11
we created a variable to indicate whether clinics were multidisciplinary by this standard (i.e.
, had an MD, RN, PT, and mental health specialist). We created a second variable to reflect the IASP criteria for multidisciplinary function: 3 physician specialties including a psychiatrist, or 2 physician specialties and a clinical psychologist.10
We created 4 additional variables based on provider and service offerings as follows: 2 provider indexes based on different types of providers at each site using the 9-category (index from 1 to 9) and 4-category (index from 1 to 4) groupings, and 2 service indexes based on the 35 services (index from 1 to 35) and the 8-category grouping (index from 1 to 8).