Variation among Primary Care Physicians in the Use of Imaging for Older Patients with Acute Low Back Pain
Variation among Primary Care Physicians in the Use of Imaging for Older Patients with Acute Low Back Pain
Alai Tan 0 1 2
Jie Zhou 0
Yong-Fang Kuo 0 1 2 3
James S. Goodwin 0 1 2 3
0 Sealy Center on Aging, University of Texas Medical Branch , Galveston, TX , USA
1 Department of Preventive Medicine and Community Health, University of Texas Medical Branch , Galveston, TX , USA
2 Institute for Translational Sciences, University of Texas Medical Branch , Galveston, TX , USA
3 Department of Internal Medicine-Geriatrics, University of Texas Medical Branch , Galveston, TX , USA
BACKGROUND: Diagnostic imaging is not recommended in the evaluation and management of non-specific acute low back pain. OBJECTIVE: To estimate the variation among primary care providers (PCPs) in the use of diagnostic imaging for older patients with non-specific acute low back pain. DESIGN AND PARTICIPANTS: Retrospective cohort study using 100 % Texas Medicare claims data. We identified 145,320 patients aged 66 years and older with nonspecific acute low back pain during the period January 1, 2007, through November 30, 2011, cared for by 3297 PCPs. MAIN MEASURES: We tracked whether each patient received lumbar imaging (radiography, computed tomography [CT], or magnetic resonance imaging [MRI]) within 4 weeks of the initial visit. Multilevel logistic regression models were used to estimate physician-level variation in imaging use. KEY RESULTS: Among patients, 27.2 % received radiography and 11.1 % received CT or MRI within 4 weeks of the initial visit for low back pain. PCPs varied substantially in the use of imaging. The average rate of radiography within 4 weeks was 53.9 % for PCPs in the highest decile, compared to 6.1 % for PCPs in the lowest decile. The average rates of CT/MRI within 4 weeks were 18.5 % vs. 3.2 % for PCPs in the highest and lowest deciles, respectively. The specific physician seen by a patient accounted for 25 % of the variability in whether imaging was performed, while only 0.44 % of the variance was due to measured patient characteristics and 1.4 % to known physician characteristics. Use of imaging by individual physicians was stable over time. CONCLUSIONS: PCPs vary substantially in the use of imaging for non-specific acute low back pain. Providerlevel measures can be employed to provide feedback to physicians in an effort to modify imaging use.
back pain; care management; practice variation; primary care; quality assessment; J Gen Intern Med 31(2); 156-63 DOI; 10; 1007/s11606-015-3475-3 © Society of General Internal Medicine 2015
Acute low back pain is the second most common symptomatic
reason for primary care visits in the United States, with an
estimated annual prevalence of 15–20 % among adults.1
Only a small proportion of these patients have an
identifiable underlying cause; for example, less than 2 % have
disc herniation.2 Most patients improve within a month
with conservative management and do not require
immediate diagnostic testing.3,4 Randomized controlled trials
have found no significant differences in clinical outcomes
(pain, function, and quality of life) in patients with
nonspecific acute low back pain between those receiving
immediate lumbar imaging (radiography, computed
tomography [CT] or magnetic resonance imaging [MRI]) and
those who receive usual clinical care without imaging.5–7
The lack of benefits is accompanied by potential harms.
Imaging may detect abnormalities unrelated to the pain,
leading to additional tests, interventions, and referrals that do not
improve outcomes.8,9 Cumulative low-level radiation
exposure is associated with an increased risk of cancer, with about
1,200 additional future cases of cancer anticipated for the 2.2
million lumbar CT scans performed annually in the US.10–12
Medical societies recommend against imaging for acute low
back pain.13–20 Table 1 summarizes the current guidelines and
recommendations. Imaging for low back pain is a target in
the BChoosing Wisely^ campaign to eliminate unnecessary
tests and procedures.13 Physicians, however, continue to
use imaging in this manner, despite guidelines and
recommendations.21 One third of patients with acute low back
pain receive imaging without a clinical indication, with
85 % of the imaging occurring within 28 days of the
A number of studies have shown that individual physician
variation is a major determinant of whether a patient receives a
test, independent of patient characteristics.23,24 This is
especially the case when the test is not generally recommended in a
given clinical presentation.25 Studies report that imaging use
for low back pain varies by physician characteristics.22,26 To
our knowledge, no data exist on variation among individual
primary care physicians (PCPs) in obtaining imaging for
patients with acute low back pain.
In the present study, we examined the use of imaging for
acute low back pain and variability in use at the level of
individual PCPs using Medicare claims data.
The study was approved by the Institutional Review Board of
the University of Texas Medical Branch.
We used 100 % Medicare claims data from Texas for January
1, 2006, through December 31, 2011. These include the
Medicare Denominator File for demographic and enrollment
information, the Carrier File for physician claims, the Outpatient
Statistical Analysis File (OUTSAF) for outpatient claims, and
the Medicare Provider Analysis and Review (MEDPAR) File
for inpatient claims. Physician characteristics were extracted
from the American Medical Association (AMA) Physician
Masterfile, which were linked with Medicare claims data via
the National Provider Identification (NPI) number.
The overall approach was to identify older patients with
nonspecific acute low back pain during the period January 1,
2007, through November 30, 2011, and the PCPs who initially
saw those patients. We employed an algorithm developed
by the National Committee for Quality Assurance to
identify non-specific acute low back pain.14 Medicare claims
from January 1, 2006, through December 31, 2011, were
included to ensure a 12-month look-back period and a 4-week
Specifically, we identified patients aged 66 years and
older who had an outpatient claim for evaluation and
management with a diagnosis of low back pain. We
excluded cases with initial evaluation made in an
emergency room. To ensure that we had complete patient
claims from which to extract comorbidity and outcome
variables, we included only patients with full coverage of
Medicare Parts A and B and with no HMO coverage
from 1 year before to 4 weeks after the diagnosis of low back
pain (n = 1,118,453 episodes). Patients with claims with a
diagnosis code for low back pain within the previous year
were excluded. We also excluded those with Bred flags^ as
indicators for imaging. These included any claims at the initial
visit or in the previous year for trauma, cancer, intravenous
drug use, or neurologic impairment14 (n=388,511 episodes).
Appendix 1 shows the algorithm used to identify episodes
of non-specific acute low back pain. For patients with
multiple episodes, we included the first episode only,
leaving 293,333 patients.
Physicians who initially saw these patients were identified
from the NPI number, and their specialties were identified
from the Health Care Financing Administration (HCFA)
specialty field. We included only PCPs in the present study
(170,880 patients cared for by 7707 PCPs). To ensure reliable
physician-level measures, we restricted analyses to PCPs with
at least 20 patients with non-specific acute low back pain
during the period 2007–2011, resulting in 145,320 patients
(85 % of eligible patients), who saw 3297 PCPs (43 % of
The Medicare Denominator File was used to extract
information on patient age, sex, and race/ethnicity. Medicaid eligibility
was measured using the state buy-in information in the
Medicare Denominator File. The percentage of high school
graduates in the patient’s ZIP Code area was obtained from US
Census data. We categorized the patient’s county of residence
as metro, non-metro, or rural, using US Department of
Agriculture definitions.27 Comorbidity was assessed using the
Charlson comorbidity index based on outpatient, inpatient,
and carrier claims 1 year prior to the diagnosis of low back
pain.28,29 For each patient, we also extracted the number of
visits within the previous year to the physician who provided
care for low back pain.
Physician age, gender, board certification, years in practice,
and medical school training in or outside the US were obtained
from the AMA Physician Masterfile. PCPs’ specialties were
obtained from the HCFA specialty field in the carrier files.
Due to the small number of general practitioners (n=59)
and geriatricians (n = 16), we classified PCPs as either
family medicine (combining family medicine and general
practice) or internal medicine (combining general internal
medicine and geriatrics).
Use of Diagnostic Imaging
We tracked each patient’s physician and outpatient claims to
identify any lumbar or sacral spine imaging, including
radiography, CT, or MRI, within 4 weeks of an initial visit for low
back pain (Appendix 1). We then identified whether the
physician who ordered the imaging was the same PCP who saw
the patient at the initial visit.
We summarized, in percentage statistics, (i) the overall
percentage of patients who received imaging by any physician
and (ii) the percentage of patients who received imaging by the
same PCP as they saw at the initial visit. To focus on
examining PCP-level imaging use, we restricted the outcome
variable in the subsequent analyses to imaging ordered by the
same PCP as seen at the initial visit. Descriptive analysis was
used to summarize patient and physician characteristics and
rates of imaging use. Our data have a two-level structure, with
patients clustered within PCPs. Multilevel logistic regression
modeling30 was used to generate PCP-level estimates on rates
of imaging use, adjusting for patient characteristics and the
clustering of patients within physicians. The average adjusted
imaging rate derived from the multilevel model was used to
identify physicians with imaging rates significantly higher or
lower than average, at a significance level of 0.05.31 We then
plotted the PCP-level adjusted rates of imaging, ranking from
low to high.
We also evaluated the stability of the PCP-level profiling
by comparing the imaging rates for each PCP in two
periods (1/1/2007–6/30/2009 and 7/1/2009–12/31/2011),
using the Spearman rank correlation. This analysis included
only the 1142 PCPs with 20 or more patients with low
back pain in each period.
SAS software version 9.2 (SAS Institute Inc., Cary, NC. USA)
was used for the statistical analyses.
We repeated the above-described analyses for another two
cohorts. Given the variability in physician coding practice,
we analyzed a cohort excluding patients with a red flag code
within 30 days after the initial visit. The 127,226 patients in
this cohort saw 2990 PCPs for acute low back pain in 2007–
2011. Our study subjects were older adults. Older patients may
receive imaging for suspicion of cancer.15 To test the impact of
patient age, we also restricted the analyses to the 24,413
patients aged 66–69 years cared for by 1568 PCPs for acute
low back pain.
Of 145,320 patients with non-specific acute low back pain,
27.2 % received radiography and 11.1 % received CT/MRI
within 4 weeks of the initial visit. The proportion of patients
receiving imaging from the same PCP seen at the initial visit
was 23.6 % for radiography and 8.1 % for CT/MRI.
Table 2 presents the percentage of patients who received
imaging within 4 weeks from the same PCP as was seen at the
initial visit, stratified by patient characteristics. Also shown are
the odds ratios generated from multilevel logistic regression
modeling, adjusted for patient characteristics and the
clustering of patients within physicians. Older age and less
frequent visits to the PCP in the prior year were associated
with a higher likelihood of receiving radiography. Factors
associated with higher likelihood of receiving CT/MRI
include younger age, male gender, non-Hispanic white
race, fewer comorbidities, ineligibility for Medicaid, and
greater number of visits to the PCP in the prior year.
Table 3 shows the percentage of patients who received
imaging within 4 weeks, stratified by PCP characteristic.
Unadjusted rates are presented, along with adjusted odds ratios
generated from multilevel logistic regression modeling.
Patients were more likely to receive radiography or CT/MRI if
the PCP was trained in the US, was board-certified, had more
years in practice, and had a higher volume of Medicare
patients. PCPs in family medicine were more likely to order
radiography but less likely to order CT/MRI. The intraclass
correlation coefficient was 0.251 based on a random
interceptonly model for imaging within 4 weeks, implying that 25.1 %
of the variation in whether a patient received imaging was
attributable to variation among physicians. In contrast, only
0.44 % of the variation was explained by known patient
characteristics and 1.41 % by known physician characteristics.
Figure 1(A) presents a cumulative distribution of rates for
each PCP in obtaining radiography within 4 weeks of the
initial visit, adjusted for patient characteristics. The average
rate of radiography use at the physician level was 23.6 %.
Among the 3297 physicians, 659 (20.0 %) had significantly
lower rates and 567 (17.2 %) had significantly higher rates
than the average. There was a ninefold difference in the
average rate for physicians in the highest versus lowest deciles
(53.9 % vs. 6.1 %). With an intraclass correlation coefficient of
0.25 and≥20 patients per physician, the reliability of
physicianlevel estimates is≥0.87.
Figure 1(B) illustrates the variation among PCPs in ordering
a low-back CT/MRI within 4 weeks of the initial visit,
adjusted for patient characteristics. The average rate at the
physician*Estimated from a multilevel model with both patient and physician characteristics. All characteristics listed in Table 2 were adjusted for in this analysis
level was 7.7 %. Among the 3297 physicians, only 72 (2.2 %)
had significant lower and 243 (7.4 %) had significant higher
rates than the average. The average rate was 18.5 % for
physicians in the highest decile, compared to 3.2 % for
physicians in the lowest decile.
We also examined stability over time in the PCP rates
of obtaining imaging, comparing rates for two periods
(1/1/2007–6/30/2009 vs. 7/1/2009–12/31/2011). Of the
physicians ranked in the lowest quartile in the first period,
over 75 % (83.5 % for radiography and 75.3 % for CT/MRI)
were ranked in the lowest two quartiles in the second period.
Of physicians ranked in the highest quartile in the first period,
over 70 % (79.3 % for radiography and 70.2 % for CT/MRI)
were ranked in the highest two quartiles in the second period.
The Spearman rank correlation between the PCP rates for the
two periods were highly correlated (r = 0.72 and 0.51 for
radiography and CT/MRI, respectively).
Compared to the primary analysis, the sensitivity analysis
had slightly lower imaging rates. For the cohort excluding
patients with a red flag code within 30 days after the initial
visit, 25.6 % received radiography and 9.6 % received CT/
MRI within 4 weeks of the initial visit. For the cohort of
patients aged 66–69 years, 23.6 % received radiography and
10.3 % received CT/MRI within 4 weeks of the initial visit.
Sensitivity analysis revealed the same patient and physician
characteristics associated with imaging use as in the primary
analysis. The substantial variation among PCPs (intraclass
correlation coefficients=0.26 and 0.25) and the high
correlation between PCP-level rates for the two time periods (r=0.70
and 0.60, respectively) were also similar to those in the
Among Texas Medicare patients with non-specific acute low
back pain during the period 2007–2011, almost one-third
received radiography and 11 % received CT or MRI within
4 weeks of the initial visit. A national sample of Medicare
claims data for 2000–2005 yielded similar rates.22 Imaging for
patients with acute low back pain without a Bred flag^
indicates potential overuse, and should be avoided.14,22
We found substantial variation among PCPs in imaging for
older patients with acute low back pain, with
betweenphysician variation accounting for 25.1 % of total variation.
This degree of variability is consistent with the concept of
preference-based rather than evidence-based practice as
described by Wennberg and colleagues.25 A recent study found a
growing gap between guideline recommendations and actual
clinical management of back pain.32 Doctors with financial
ties to MRI scanner businesses tend to refer more patients for
imaging,33 but we were unable to examine this association in
our analysis. Nevertheless, design and use of provider
performance measures is an important strategy for curbing the
influence of such financial ties and to promote high-value,
Initial enthusiasm about physician-level quality measures
has been tempered by several concerns, including the reliability
of these measures and their stability over time.35,36 For many
quality measures, provider variation is low, with intraclass
correlation coefficients of <0.10.35,37 This in turn leads to
low reliability of provider performance estimates, unless
the number of patients per provider is larger than is
feasible for most measures. However, the high physician
variability in early imaging for acute low back pain allows for
generation of estimates with reliability of >0.80 for
physicians with at least 20 such patients. Also, the majority of
top and low performers in one period were also top and
low performers in the subsequent period. Our study
suggests that it is feasible to generate provider-level estimates
of imaging use with high reliability and stability. Such
measures can then provide feedback to physicians to modify
The Healthcare Effectiveness Data and Information Set
quality measures, most derived from administrative claims
data, are widely used to compare performance across
health plans. Physician-level measures can be similarly
employed. One barrier to quality improvement is that
physicians are generally unaware of their quality relative
to their peers. Feedback raises awareness and motivates
physician to improve. A meta-analysis found that
physician feedback is more effective when provided frequently,
in writing, with an explicit action plan, and tied to
incentives or accountability.38 Future studies are needed to
translate our proposed physician measures of imaging for
low back pain into plans for actual interventions and then
to assess the effectiveness of such interventions.
One should not assume that the percentage of patients with
low back pain receiving imaging should ideally be zero.
Several considerations might lead a physician to order such
imaging; not all can be captured in Medicare billing data.
Instead, our assumption is that the percentage of patients for
whom such considerations are relevant is relatively low, and
that these cases should average out among PCPs. It is
improbable that patients with acute complex low back pain would
cluster within individual PCPs. The focus is on performance
relative to other PCPs rather than on the absolute percentage of
patients who undergo imaging.
Compared to the large between-physician variation, the
variance explained by measured patient and physician
characteristics was minimal (<2 %). Of note, the risk of receiving
CT/MRI was higher for younger, healthier patients. Physicians
with more years in practice were more likely to order imaging.
This finding resonates with the call for continuing medical
education targeting older physicians.39
Our study has several limitations. First, physician coding
practice varies. Indications for imaging may appear after the
index visit. We conducted sensitivity analysis by excluding
patients with a red flag code within 30 days after the initial
visit; the high variation among PCPs remained. Second, the
study included older patients. Our sensitivity analysis in
patients aged 66–69 years can partially mitigate concerns with
the effect of age. Nevertheless, studies including younger
patients are needed to confirm our findings. Moreover,
administrative claims data cannot capture patient preference.
Physicians may order imaging at the patient’s request in order to
increase patient satisfaction.40 We also could not completely
exclude patients with certain red flags (e.g., weight loss or
remote history of cancer) using administrative claims data.
However, these cases should average out among PCPs, and
were unlikely to explain the marked variation among PCPs in
imaging use. The present study focused on PCPs. Higher rates
of imaging use, especially CT/MRI for acute low back pain,
have been reported for specialists.23 Nevertheless, restricting
the analyses to PCPs with >20 patients captured 85 % of
Medicare patients with non-specific acute low back pain in
the primary care setting. Another limitation is that our findings
may be not generalizable to regions outside Texas, given
differences in population composition and healthcare
utilization patterns.41 Nevertheless, our estimated imaging rates are
comparable to those from the national data.22 Future studies in
other geographic regions are needed.
In conclusion, despite guideline recommendations, a
substantial proportion of patients with acute low back pain receive
imaging. The use of imaging varied greatly among PCPs,
and this high level of variability permits generation of
provider-level measures with good reliability. Such
measures can be employed to provide feedback to physicians
to modify overuse.
Acknowledgments: This study was supported by the National
Institutes of Health (K05CA134923, UL 1TR000071 and
P30AG024832) and the Agency for Healthcare Research and Quality
(1R24HS022134-01). The funding agency had no role in the study
design, conduct, analysis, or interpretation of data or in the
preparation, review, or approval of the manuscript.
Sarah Toombs Smith, PhD, Science Editor and Assistant Professor at
the Sealy Center on Aging, University of Texas Medical Branch at
Galveston, provided editorial assistance in manuscript preparation.
Dr. Toombs Smith received no compensation for her assistance apart
from her employment at the institution where the study was
Conflict of Interest: The authors declare that they do not have a
conflict of interest.
Author Contributions: Dr. Zhou had full access to all of the data in the
study and takes responsibility for the integrity of the data and the
accuracy of the data analysis. Study concept and design: Tan and
Goodwin. Acquisition of data: Goodwin. Analysis and interpretation
of data: All authors. Drafting of the manuscript: Tan. Critical revision of
the manuscript for important intellectual content: Goodwin and Kuo.
Statistical analysis: Zhou. Obtained funding: Goodwin. Administrative,
technical and material support: Tan, Kuo and Goodwin. Study
Corresponding Author: Alai Tan, MD, PhD; Institute for Translational
Sciences, University of Texas Medical Branch, Galveston, TX,
USA (e-mail: ).
1. Hart LG , Deyo RA , Cherkin DC . Physician office visits for low back pain. Frequency, clinical evaluation, and treatment patterns from a U.S. National survey . Spine (Phila Pa 1976 ). 1995 ; 20 ( 1 ): 11 - 9 .
2. Atlas SJ , Deyo RA . Evaluating and managing acute low back pain in the primary care setting . J Gen Intern Med . 2001 ; 16 ( 2 ): 120 - 31 .
3. Henley E. Understanding and treating low back pain in family practice . J Fam Pract . 2000 ; 49 ( 9 ): 793 - 5 .
4. Pengel LH , Herbert RD , Maher CG , Refshauge KM . Acute low back pain: systematic review of its prognosis . BMJ . 2003 ; 327 ( 7410 ): 323 .
5. Chou R , Fu R , Carrino JA , Deyo RA . Imaging strategies for lowback pain: systematic review and meta-analysis . Lancet . 2009 ; 373 ( 9662 ): 463 - 72 .
6. Gilbert FJ , Grant AM , Gillan MG , et al. Low back pain: influence of early MR imaging or CT on treatment and outcome-multicenter randomized trial . Radiology . 2004 ; 231 ( 2 ): 343 - 51 .
7. Kendrick D , Fielding K , Bentley E , Kerslake R , Miller P , Pringle M. Radiography of the lumbar spine in primary care patients with low back pain: randomised controlled trial . BMJ . 2001 ; 322 ( 7283 ): 400 - 5 .
8. Webster BS , Cifuentes M. Relationship of early magnetic resonance imaging for work-related acute low back pain with disability and medical utilization outcomes . J Occup Environ Med . 2010 ; 52 ( 9 ): 900 - 7 .
9. Deyo RA . Cascade effects of medical technology . Annu Rev Public Health . 2002 ; 23 : 23 - 44 .
10. Fazel R , Krumholz HM , Wang Y , et al. Exposure to low-dose ionizing radiation from medical imaging procedures . N Engl J Med . 2009 ; 361 ( 9 ): 849 - 57 .
11. Smith-Bindman R , Miglioretti DL , Larson EB . Rising use of diagnostic medical imaging in a large integrated health system . Health Aff (Millwood) . 2008 ; 27 ( 6 ): 1491 - 1502 .
12. Berrington de González A , Mahesh M , Kim KP , et al. Projected cancer risks from computed tomographic scans performed in the United States in 2007 . Arch Intern Med . 2009 ; 169 ( 22 ): 2071 - 7 .
13. US Medical Specialty Societies. Choosing Wisely Campaign: Five Things Physicians and Patients Should Question . Available at: http://www. choosingwisely.org/doctor-patient-lists/. Accessed July 7 , 2015 .
14. National Committee for Quality Assurance . NQA-Endosed national voluntary concensus standards for physician-focused ambulatory care: Appendix A -Natinal Committee for Quality Assurance measure technical specifications , April, 2008 V. 7 . Available at: http://www.ncqa.org/Portals/0/HEDISQM/NQF_Posting_Appendix.pdf. Accessed July 7 , 2015 .
15. Davis PC WFI , Cornelius RS , Angtuaco EJ , Broderick DF , Brown DC , Garvin CF , Hartl R , Holly L , McConnell CT Jr, Mechtler LL , Rosenow JM , Seidenwurm DJ , Smirniotopoulos JG . Expert Panel on Neurologic Imaging. ACR Appropriateness Criteria® low back pain . Reston , VA: American College of Radiology (ACR); 2011 .
16. Chou R , Qaseem A , Owens DK , Shekelle P. Physicians CGCotACo . Diagnostic imaging for low back pain: advice for high-value health care from the American College of Physicians . Ann Intern Med . 2011 ; 154 ( 3 ): 181 - 9 .
17. Chou R , Qaseem A , Snow V , et al. Diagnosis and treatment of low back pain: a joint clinical practice guideline from the American College of Physicians and the American Pain Society . Ann Intern Med . 2007 ; 147 ( 7 ): 478 - 91 .
18. Casazza BA . Diagnosis and treatment of acute low back pain . Am Fam Physician . 2012 ; 85 ( 4 ): 343 - 50 .
19. Delitto A , George SZ , Van Dillen LR , et al. Low back pain: clinical practice guidlines linked to the international classification of functioning, disability and health from the Orthopaedic Section of the American Physical Therapy Association . J Orthop Sports Phys Ther . 2012 ; 42 ( 4 ): A1 - 57 .
20. American Chronic Pain Association. Consumers guide: Practice guidelines for low back pain . Available at: http://theacpa.org/uploads/documents/ Consumer%20Guidelines %20for%20Low%20Back%20PainFinal%202-6- 08.pdf. Accessed July 7 , 2015 .
21. Kale MS , Bishop TF , Federman AD , Keyhani S. Trends in the overuse of ambulatory health care services in the United States . JAMA Intern Med . 2013 ; 173 ( 2 ): 142 - 8 .
22. Pham HH , Landon BE , Reschovsky JD , Wu B , Schrag D . Rapidity and modality of imaging for acute low back pain in elderly patients . Arch Intern Med . 2009 ; 169 ( 10 ): 972 - 81 .
23. Cherkin DC , Deyo RA , Wheeler K , Ciol MA . Physician variation in diagnostic testing for low back pain. Who you see is what you get . Arthritis Rheum . 1994 ; 37 ( 1 ): 15 - 22 .
24. Jaramillo E , Tan A , Yang L , Kuo YF , Goodwin JS . Variation among primary care physicians in prostate-specific antigen screening of older men . JAMA . 2013 ; 310 ( 15 ): 1622 - 4 .
25. Wennberg J . Tracking medicine . New York: Oxford University Press; 2010 : 117 - 155 .
26. Carey TS , Garrett J . Patterns of ordering diagnostic tests for patients with acute low back pain . The North Carolina Back Pain Project. Ann Intern Med . 1996 ; 125 ( 10 ): 807 - 14 .
27. US Department of Agriculture. Rural-urban Continuum Codes . Available at: http://www.ers.usda.gov /data-products/rural-urban-continuumcodes/documentation .aspx. Accessed July 7 , 2015 .
28. Charlson ME , Pompei P , Ales KL , MacKenzie CR . A new method of classifying prognostic comorbidity in longitudinal studies: development and validation . J Chronic Dis . 1987 ; 40 ( 5 ): 373 - 83 .
29. Klabunde CN , Potosky AL , Legler JM , Warren JL . Development of a comorbidity index using physician claims data . J Clin Epidemiol . 2000 ; 53 ( 12 ): 1258 - 67 .
30. Leyland A , Goldstein H , eds. Multilevel modelling of health statistics. New York: Wiley; 2001 .
31. Normand S-LT , Glickman ME , Gatsonis CA. Statistical methods for profiling providers of medical care: issues and applications . J Am Stat Assoc . 1997 ; 92 ( 439 ): 803 - 14 .
32. Mafi JN , McCarthy EP , Davis RB , Landon BE . Worsening trends in the management and treatment of back pain . JAMA Intern Med . 2013 ; 173 ( 17 ): 1573 - 81 .
33. Lungren MP , Amrhein TJ , Paxton BE , et al. Physician self-referral: frequency of negative findings at MR imaging of the knee as a marker of appropriate utilization . Radiology . 2013 ; 269 ( 3 ): 810 - 15 .
34. Baker DW , Qaseem A , Reynolds PP , Gardner LA , Schneider EC . Design and use of performance measures to decrease low-value services and achieve cost-conscious care . Ann Intern Med . 2013 ; 158 ( 1 ): 55 - 9 .
35. Hofer TP , Hayward RA , Greenfield S , Wagner EH , Kaplan SH , Manning WG . The unreliability of individual physician Breport cards^ for assessing the costs and quality of care of a chronic disease . JAMA . 1999 ; 281 ( 22 ): 2098 - 2105 .
36. Parry GJ , Gould CR , McCabe CJ , Tarnow-Mordi WO . Annual league tables of mortality in neonatal intensive care units: longitudinal study. International Neonatal Network and the Scottish Neonatal Consultants and Nurses Collaborative Study Group . BMJ . 1998 ; 316 ( 7149 ): 1931 - 5 .
37. Nelson EC , Gentry MA , Mook KH , Spritzer KL , Higgins JH , Hays RD . How many patients are needed to provide reliable evaluations of individual clinicians? Med Care . 2004 ; 42 ( 3 ): 259 - 66 .
38. Ivers N , Jamtvedt G , Flottorp S , et al. Audit and feedback: effects on professional practice and healthcare outcomes . Cochrane Database Syst Rev . 2012 ; 6 , CD000259 .
39. Choudhry NK , Fletcher RH , Soumerai SB . Systematic review: the relationship between clinical experience and quality of health care . Ann Intern Med . 2005 ; 142 ( 4 ): 260 - 73 .
40. Rosenbaum L. The whole ball game-overcoming the blind spots in health care reform . N Engl J Med . 2013 ; 368 ( 10 ): 959 - 62 .
41. Kaiser Family Foundation. Texas: Demographics and the economy indicators . Available at: http://kff.org/state-category/demographics-andthe-economy/?state=TX. Accessed July 7 , 2015 .