We used 12 years of data (2003–2014) from the National Ambulatory Medical Care Survey, a national probability sample survey of office-based physicians engaged in patient care. Data collected on sampled visits include patient characteristics; visit characteristics; physician diagnosis; services/tests ordered (including referral to PT and/or another physician); and treatments/medications prescribed. Data on physician specialty are also collected.
We limited our analysis to visits to PCPs (i.e., general/family practice, internal medicine, pediatrics, obstetrics/gynecology) for musculoskeletal conditions based on ICD-9-CM codes and created the following diagnostic subgroups: arthropathies and related disorders (710–719); spine disorders (720–724, 846, 847); soft tissue disorders (725–729); osteopathies, chondropathies, musculoskeletal deformities (730–739); fractures/dislocations (820–829, 830–839); extremity sprains/strains (840–845, 848); contusions (922–924). In addition to PT and physician referral data, we extracted data on patient characteristics (age, sex, race, ethnicity, insurance status) and the major reason for visit (i.e., acute problem, chronic problem, pre- or post-surgery, preventive).
Analyses were conducted using Stata (14.2) and accounting for the multistage survey design.4 We pooled the data over 2-year intervals to increase estimate precision.5 To account for potential changes in the characteristics of ambulatory care visits over time, we estimated a multivariate logistic regression model with PT referral as the outcome; the 2-year intervals as categorical, explanatory variables; and age, sex, race, ethnicity, insurance, major reason for visit, and diagnostic subgroups as our control variables. We then calculated average adjusted prevalence rates for each 2-year period and assessed for a linear trend over time.6 We also estimated number of visits referred. We used the same approach to estimate referral rates and number of visits referred to specialty physicians. We conducted our analyses for the entire sample (i.e., all musculoskeletal-related ambulatory care visits) and for the three most prevalent diagnostic subgroups: arthropathies (32.3% of all musculoskeletal visits), spine (28.3%), and soft tissue disorders (17.9%).