Table 4. Association Between Physician Characteristics and Low-Value Services.
Physician Characteristic | Within-Region Analysisa | Within-Organization Analysisb | ||
---|---|---|---|---|
Additional Services per 100 Beneficiaries per Year (95% CI) | P Value | Additional Services per 100 Beneficiaries per Year (95% CI) | P Value | |
Academic degree (MD vs DO) | −1.10 (−1.66 to −0.55) | <.001 | −1.05 (−1.54 to 0.55) | <.001 |
Foreign medical graduate | 1.80 (1.37 to 2.23) | <.001 | 0.71 (0.38 to 1.04) | <.001 |
Graduate of medical school in US News and World Report 2013 top 20 rank | −0.39 (−0.88 to 0.09) | .11 | −0.41 (−0.83 to 0.01) | .05 |
Agec | 0.23 (0.10 to 0.36) | .001 | 0.16 (0.05 to 0.27) | .004 |
Female | 0.82 (0.54 to 1.11) | <.001 | 1.13 (0.88 to 1.37) | <.001 |
Academic title (vs no professorship) | ||||
Full professor | −4.80 (−6.01 to −3.59) | <.001 | −0.95 (−2.09 to 0.19) | .10 |
Associate professor | −4.14 (−5.38 to −2.90) | <.001 | −0.86 (−2.00 to 0.27) | .14 |
Assistant professor | −3.03 (−3.66 to −2.41) | <.001 | −0.38 (−0.99 to 0.24) | .23 |
Any publication authorship | −0.13 (−0.49 to 0.22) | .47 | −0.16 (−0.44 to 0.120) | .25 |
No. of publications authored | 0.01 (−0.01 to 0.03) | .45 | 0.01 (−0.01 to 0.02) | .24 |
Clinical trial investigator | −0.22 (−2.82 to 2.37) | .87 | −0.08 (−1.90 to 1.73) | .93 |
NIH grant recipient | −0.49 (−2.94 to 1.96) | .69 | −0.34 (−2.05 to 1.38) | .70 |
Any pharmaceutical and/or device company payments | 1.86 (1.60 to 2.12) | <.001 | 0.43 (0.18 to 0.699) | .001 |
Pharmaceutical and/or device company payment amountd | 0.11 (0.03 to 0.19) | .01 | 0.09 (0.03 to 0.15) | .002 |
Medicare FFS panel sizee | 10.59 (8.14 to 13.05) | <.001 | 2.53 (0.72 to 4.34) | .006 |
Abbreviations: DO, doctor of osteopathy; FFS, fee-for-service; MD, doctor of medicine; NIH, National Institutes of Health.
In addition to patient sociodemographic and clinical characteristics, this model contains indicators for each patient hospital referral region, effectively comparing physicians within the same geographic service area. Confidence intervals were estimated using robust variance estimators, clustered at the physician level.
In addition to patient sociodemographic characteristics, clinical characteristics, and hospital referral region, this model also contains indicators for provider organization tax identification number, effectively comparing physicians within the same provider organization. Confidence intervals were estimated using robust variance estimators, clustered at the physician level.
Coefficient is scaled to represent the additional services per 100 beneficiaries that are associated with a 10-year increase in age.
Coefficient is scaled to represent the additional services per 100 beneficiaries that are associated with a $5000 increase in pharmaceutical and/or device company payments.
Coefficient is scaled to represent the additional services per 100 beneficiaries that are associated with an increase of 100 patients per year.