Table 4

Adjusteda Proportions of Utilization by Identified Primary Care Physician (PCP) versus Non-PCP over Time

Year
(sample n, PCP)
[Sample n, non-PCP]		1999–2000
(n=1,216)
[n=1,855]		2001–2002
(n=1,176)
[n=2,446]		2003–2004
(n=1,294)
[n=2,405]		2005–2006
(n=1,375)
[n=2,561]		2007–2008
(n=1,408)
[n=2,848]		2009–2010
(n=1,597)
[n=2,209]		P-value
Medications
  Narcotics
    PCP212525313631<.001
    Non-PCP161420233128<.001
  Tramadol
    PCP3.31.91.32.84.04.1.03
    Non-PCP2.12.31.81.74.04.9.01
  Neuropathic
    PCP2.93.84.14.77.56.2<.001
    Non-PCP3.13.84.64.9128.0<.001
  Benzo/Relaxant
    PCP222425253028.01
    Non-PCP121217161919.004
  NSAIDs/APAPb
    PCP423638312824<.001
    Non-PCP282227242323.47
Referrals
  Physical Therapy
    PCP126.29.39.98.19.7.92
    Non-PCP282323191825.54
  Other Physician
    PCP6.91112121416<.001
    Non-PCP6.11411111511.22
Imaging
  X-Ray
    PCP131313121212.59
    Non-PCP211413171918.49
  CT/MRI
    PCP4.95.45.77.98.16.6.02
    Non-PCP9.59.98.7141414.002

aIn this table, our logistic regression models adjusted for age, sex, race, region, insurance type, visit year, duration of symptoms or context of visit (e.g., acute/new onset, chronic routine), and whether the visit was located in a metropolitan area.

bInteraction with time was significant (p=.01).

Abbreviations:
PCP, Primary Care Physician.