Post on 29-Dec-2015
Intensity of Imaging for Low Back Pain in Elderly Patients
AcademyHealth Annual MeetingJune 2007
HH Pham, MD, MPH, D Schrag, MD, MPH
C Corey, MS, J Reschovsky, PhD
HR Rubin, MD, PhD, BE Landon, MD, MBA
Background
Medicare spending on imaging services has increased dramatically since 2000 with unclear clinical benefits for beneficiaries
Guidelines allow discretion for imaging of elderly patients with acute low back pain
Little representative data on non-clinical factors associated with intensity of imaging
Research questions
What physician, practice, market, and non-clinical patient factors are associated with more intensive imaging for acute low back pain?
Does the economic environment in which physicians practice influence discretionary use of imaging?
Data sources (1)
2000-2001 Community Tracking Study Physician Survey• Nationally representative, clustered in 60 communities• Non-federal, completed training, 20+ hrs of clinical care/week• 12,406 respondents, ~50% PCPs• 59% response rate
Questions• Specialty, board certification, FMG status• Practice type, revenue sources (Medicaid, Medicare), capitation• Ability to obtain specialist and imaging referrals• Overall effect of financial incentives (increase/decrease services)• Compensation based on quality, profiling, patient satisfaction• Practice ownership
Data sources (2)
Complete 2000-2002 Medicare claims for 1.09 million beneficiaries seen by CTS physicians in year 2000
Geographic data from Area Resources File on number of patient care radiologists per capita, household income, and education levels
Design and Analysis
Back pain diagnosis identified for year 2001 Followed for 6 months after back pain diagnosis Modeled “intensity” of imaging
• never imaged imaged 29 -180 days imaged within 28 days• “Intensity” measured for:
- (a) any imaging modality; and (b) only CT/MRI
Excluded patients diagnosed by a radiologist Adjusted for comorbidities during year 2000, physician,
practice, and area factors (site fixed effects) Repeated analyses, excluding patients with visits to
other physicians between diagnosis and imaging dates
63,075 (15%) patients of 318,148 linked to a CTS PCP and had a diagnosis of acute low back pain in 2001
24,515 (39%) meeting clinical inclusion criteria (no potential indications for imaging 6 months prior to LBP diagnosis or
between diagnosis and imaging dates
5,964 (28%) imaged within 28 days
5,330 (90%) by XR725 (12%) by CT/MRI
15,011 (67%) never imaged
1,017 (4%) imaged between 29-180 days734 (73%) by XR
314 (31%) by CT/MRI
21,992 (89%) meeting inclusion criteria and not diagnosed by a radiologist
Study population
Clinical exclusions
Modified NCQA’s measure of inappropriate imaging for acute LBP
Cancers* Neurologic deficits* Trauma,* falls, injury Infections – endocarditis, osteomyelitis, TB, etc. IV drug use* Anemia – not hereditary, Fe deficiency, or blood loss Constitutional symptoms – weight loss, fever, night
sweats, fatigue/malaise, loss of appetite
Care relationships between acute LBP patients and their plurality PCP
Median (IQR) % of E&M visits with PCP 63 (47-80)
Had a visit with their CTS PCP within 6 months of LBP diagnosis 81%
Diagnosed by their CTS PCP 52%
Diagnosed in their CTS PCP’s practice 60%
Diagnosed by any PCP
Specialties of other diagnosing clinicians (outside of their CTS PCP’s practice)
62%
Orthopedic surgeon 9%
Chiropractor 15%
Site of imaging studies performed within 28 days of diagnosis
Modality
Total imaged
N
Patients imaged in
PCP’s practice
N (%)
Any 6,981 2,439 (37.5)
X-Ray 6,064 2,192 (38.9)
CT or MRI 1,039 280 (27.1)
Timing of imaging after LBP diagnosis
Imaging procedure
Patients, N
Number of days between diagnosis and imaging,
Median (IQR)
Any modality 6,981 0 (0-7)
CT 165 9 (3-35)
MRI 879 13 (4-36)
Predictors of intensity of imagingPatient factors and radiologist supply
Characteristic
Any modality
Adjusted OR (95% CI)
CT or MRI
Adjusted OR (95% CI)
Female 1.01 (0.92-1.11) 0.81 (0.67-0.96)*
Medicaid eligible 0.81 (0.70-0.94)* 0.94 (0.71-1.25)
Race (vs. white)
Black 0.83 (0.77-0.96)* 0.67 (0.46-0.99)*
Other 0.95 (0.71-1.26) 0.91 (0.58-1.42)
Radiologists/1000 (vs. lowest quartile)
Highest quartile 1.10 (0.97-1.25) 1.31 (1.02-1.69)*
No effect for median household income in the patient zip code; % adults with 12+ yrs of education in the county; or Klabunde or Charlson scores
Predictors of intensity of imagingPhysician factors
Characteristic
Any modality
Adjusted OR (95% CI)
CT or MRI
Adjusted OR (95% CI)
FP/GP specialty (vs. IM) 0.95 (0.87-1.03) 0.83 (0.67-1.02)
Effect of incentives (vs. increase services)
To reduce services 0.83 (0.68-1.01) 0.73 (0.51-1.00)*
No effect on services 1.03 (0.94-1.12) 1.00 (0.80-1.25)
No effect for years in practice; board certification; IMG status; compensation based on productivity, quality, profiling or
patient satisfaction measures, or practice ownership
Predictors of intensity of imagingPractice factors
Characteristic
Any modality
Adjusted OR (95% CI)
CT or MRI
Adjusted OR (95% CI)
% Revenue from capitation (vs. none)
1-10% 1.05 (0.94-1.17) 0.84 (0.68-1.03)
11-25% 0.98 (0.85-1.13) 0.74 (0.54-1.00)*
>25% 0.94 (0.79-1.12) 0.67 (0.50-0.90)**
Practice type (vs. solo/2)
Small group (3-10) 1.19 (1.03-1.37)* 1.10 (0.79-1.52)
Medium group (11-50) 1.49 (1.21-1.84)*** 0.94 (0.58-1.50)
Large group (>50) 1.22 (1.00-1.49)* 1.29 (0.85-1.96)
Medical school 0.84 (0.62-1.14) 0.64 (0.29-1.43)
No consistent effect for revenue from managed care, Medicare, or Medicaid
Limitations
No certainty regarding appropriateness of imaging• Not benchmarking – only comparing relative performance• Unlikely systematic under-coding of exclusions by physician or
practice characteristics, or by white patient race and higher SES• Uncertainty is comparable to claims-based measures of underuse
Lack data on presence of imaging equipment in practices
Cannot identify physician(s) responsible for referrals• For imaging or to specialists• But consistent relationships between characteristics of the CTS
PCP and intensity of imaging
Conclusions
Substantial minority of elderly patients with uncomplicated LBP are imaged early, often in their physician’s practice• Most cases of rapid imaging use XR’s, not CT/MRI
Overall financial incentives matter, but no association with specific types of performance-based compensation
Subgroups of patients who tend to receive fewer services may sometimes benefit
Incentives to increase or decrease services may have mixed effects on quality that may go undetected if the majority of performance metrics reflect underuse
Geographic variation in percent of patients imaged within 28 days
Before exclusions
After clinical exclusions
Any Modality
After clinical exclusions
CT/MRI
CTS Market Unadjusted % Unadjusted % Adjusted % Unadjusted % Adjusted %
Seattle 22.6 20.7 29.2 2.6 4.7
Phoenix 24.6 23.9 28.9 2.9 4.6
Miami 34.0 26.3 21.5 5.9 3.5
Newark 32.3 27.4 27.5 7.6 4.9
Cleveland 29.7 27.5 29.3 4.5 5.0
Indianapolis 29.5 28.5 28.3 5.8 4.9
Lansing 27.3 28.5 28.9 3.3 5.1
Greenville 30.0 29.8 28.4 4.6 4.6
Little Rock 29.0 30.4 29.3 6.6 5.2
Orange Cty 28.1 30.9 26.6 5.0 4.7
Boston 29.5 31.8 29.4 11.1 5.8
Syracuse 31.6 33.3 30.5 4.9 5.3