A p p ro p r i a t e U s e o f L u m b a r

Imaging for Evaluation of Low

Back Pain
Roger Chou, MDa,b,*, Richard A. Deyo, MD, MPHa,c,d,e,f,g,
Jeffrey G. Jarvik, MD, MPHh,i,j

KEYWORDS
 Low back pain  Radiography  MRI  CT

KEY POINTS
 Strong evidence shows that routine back imaging does not improve patient outcomes, exposes
patients to unnecessary harms, and increases costs.
 Diagnostic imaging studies should only be performed in patients who have severe or progressive
neurologic deficits or with features suggesting a serious or specific underlying condition.
 Advanced imaging with MRI or CT should be reserved for patients with a suspected serious under-
lying condition or neurologic deficits or who are candidates for invasive interventions.
 To be effective, efforts to reduce imaging overuse should be multifactorial and address clinician
behaviors, patient expectations and education, and financial incentives.
 Radiologists can help reduce imaging overuse by accurately reporting and providing consultative
expertise regarding the prevalence and potential clinical significance (or insignificance) of imaging
findings.

Funding statement: No funding was received for this manuscript.

Disclosures: Roger Chou was the lead author on guidelines developed by the American Pain Society and
American College of Physicians on diagnosis and management of low back pain, including recommendations
on imaging, and has consulted with Wellpoint Inc, Blue Cross Blue Shield Association, and Palladian Health on
implementing low back pain guidelines.
Jeffrey G. Jarvik is a consultant to General Electric Healthcare serving on their Comparative Effectiveness
Advisory Board. He also consults with HealthHelp, a radiology benefits management company. He is a
cofounder of PhysioSonics, a company that uses high-intensity focused ultrasound for diagnostic purposes,
is a stockholder, and receives royalties for intellectual property.
a
Department of Medicine, Oregon Health & Science University, 3181 Southwest Sam Jackson Park Road,
Portland, OR 97239, USA; b Department of Medical Informatics and Clinical Epidemiology, Oregon Health &
Science University, 3181 Southwest Sam Jackson Park Road, Portland, OR 97239, USA; c Department of Family
Medicine, Oregon Health & Science University, 3181 Southwest Sam Jackson Park Road, Portland, OR 97234,
USA; d Department of Public Health, Oregon Health & Science University, 3181 Southwest Sam Jackson Park
Road, Portland, OR 97234, USA; e Department of Preventive Medicine, Oregon Health & Science University,
3181 Southwest Sam Jackson Park Road, Portland, OR 97234, USA; f Center for Research on Occupational and
Environmental Toxicology, Oregon Health & Science University, 3181 Southwest Sam Jackson Park Road, Portland,
OR 97234, USA; g Kaiser Permanente Center for Health Research, 3800 North Interstate Avenue, Portland, OR
radiologic.theclinics.com

972271078, USA; h Department of Radiology, Harborview Medical Center, University of Washington, 325
Ninth Avenue, Box 359728, Seattle, WA 98104, USA; i Department of Neurological Surgery, Harborview Medical
Center, University of Washington, 325 Ninth Avenue, Box 359728, Seattle, WA 98104, USA; j Department of
Health Services, Harborview Medical Center, University of Washington, 325 Ninth Avenue, Box 359728, Seattle,
WA 98104, USA
* Corresponding author. Department of Medicine, Oregon Health & Science University, 3181 Southwest Sam
Jackson Park Road, Portland, OR 97239.
E-mail address: [email protected]

Radiol Clin N Am 50 (2012) 569585

doi:10.1016/j.rcl.2012.04.005
0033-8389/12/$ see front matter 2012 Elsevier Inc. All rights reserved.
570 Chou et al

Low back pain is extremely common, ranking as imaging has long been noted as a problem,14
the second most common symptomatic reason yet the use of imaging (particularly advanced
for office visits in the United States.1,2 About imaging) continues to increase rapidly.15 This
one-third of adults in the United States report article reviews costs associated with spinal imag-
back pain during the past 3 months,1 and nearly ing, current imaging practice patterns and trends,
three-quarters of adults report at least one epi- evidence on benefits and harms associated with
sode of low back pain during their lifetime.3 spinal imaging, factors that promote or are permis-
Low back pain is also very costly. In 1998, total sive of imaging overuse, and potential strategies
health care expenditures for individuals with back for improving imaging practices.
pain in the United States were estimated at $90
billion,4 and costs have since risen. The inflation- COSTS
adjusted increase (in 2005 U.S. dollars) in average Direct Costs
total health expenditures for people with back and
neck problems was 65% ($4795 per year in 1997 Direct costs of imaging include costs of equipment
to $6096 per year in 2005).5 Low back pain also and facilities, radiologic department staff, pro-
results in high indirect costs from disability, lost fessional fees for interpreting the test, and other
time from work, and decreased productivity while overhead. Because direct costs are often difficult
at work,6 and is the most common cause for to measure, reimbursement rates or charges are
activity limitations in younger adults. In the United often used as surrogate measures. Although esti-
States, 14% of workers lose at least 1 day of work mates vary substantially depending on geographic
each year because of low back pain.7 location, insurance status, and other factors, reim-
Lumbar spine imaging (plain radiography, CT, bursement rates and charges for lumbar spine CT
and MRI) is often performed in patients with low generally run 5 to 10 times higher than lumbosacral
back pain. Although clinical practice guidelines spine plain radiography, and MRI 10 to 15 times
recommend imaging only in the presence of higher (Table 1). Despite its relatively lower cost,
progressive neurologic deficits or signs or symp- lumbosacral spine radiography is a major contrib-
toms suggesting a serious or specific underlying utor to costs because of its frequent use. In 2004,
condition (the so-called red flags of low back an estimated 66 million lumbar radiographs were
pain),8 imaging is often performed in the absence performed in the United States.16
of a clear clinical indication for it.9 This fact is con-
Downstream Costs
cerning, because routine imaging does not seem
to improve clinical outcomes, exposes patients In addition to direct costs, imaging can also lead to
to unnecessary harms, and contributes to the downstream cascade effects, referring to the sub-
rising costs associated with low back pain.1012 sequent tests, referrals, and interventions per-
Eliminating unnecessary tests would help rein in formed as a result of imaging.17 In some cases,
costs associated with low back pain while main- the end result can be an invasive and expen-
taining high-quality care.13 Overuse of low back sive operation or other procedure of limited or

Table 1
Costs of spine imaging and fusion surgery

Intervention Cost
a
Lumbar spine radiography (two or three views) $54
Lumbar spine CT scana $344 (without contrast)
$426 (with contrast)
Lumbar spine MRIa $645 (without contrast)
$794 (with contrast)
Fusion surgeryb Without bone morphogenetic proteins: median,
$57,393 (interquartile range, $39,660$83,608)
With bone morphogenetic proteins: median,
$74,254 (interquartile range, $54,737$102,663)

a
Medicare reimbursement for San Francisco area, in 2011 dollars, calculated at http://www.trailblazerhealth.com/Tools/
Fee%20Schedule/MedicareFeeSchedule.aspx.
b
Total charge for hospitalization, excluding professional fees, based on 2006 Nationwide Inpatient Sample data.
Data from Cahill KS, Chi JH, Day A, et al. Prevalence, complications, and hospital charges associated with use of bone-
morphogenetic proteins in spinal fusion procedures. JAMA 2009;302(1):5866.
 Appropriate Use of Lumbar Imaging 571

questionable benefit. In 2006, according to the they would obtain imaging for uncomplicated low
Nationwide Inpatient Sample, the median total back pain.30
cost for fusion surgery without bone morphoge- Why are practice variations a cause for con-
netic proteins was nearly $60,000, excluding pro- cern? If they occur in otherwise similar populations
fessional fees.18 Although the increased number and settings, variations may indicate inequalities
of unnecessary operations that occur from un- in resource use or areas in which care is hap-
needed imaging tests is difficult to estimate, data hazard or arbitrary.33 In addition, research on
show that rates of spine MRIs increased sharply regional variations in the United States suggests
at the same time as back surgeries.11,19 Similarly, that high-use areas are generally not associated
over about a 10-year period starting in the mid- with better clinical outcomes but contribute signif-
90s, rates of interventional procedures such as icantly to overall health care costs.34,35 This finding
epidural steroid and facet joint injections more often signifies inefficiencies in medical care, which
than tripled,20 a pattern that roughly parallels can be caused by clinical uncertainty or a failure to
trends in increased use of MRI tests. implement evidence-based practice.
Increased use of surgery and interventional
procedures would not necessarily be a problem if Imaging Rates
the procedures resulted in important clinical bene- A study based on a national database of private
fits. However, even though rates of surgery are two insurance claims (covering 8 million beneficiaries)
to five times higher in the United States than in found that more than 40% of patients with acute
other developed countries,21 no evidence shows low back pain underwent imaging.36 The median
that patients with low back pain fare better in the time to imaging was the same day as the index
United States than in other countries, and random- diagnosis. Data indicate that imaging rates con-
ized trials suggest that surgery and interventional tinue to increase, despite efforts to curb overuse.
procedures are associated with limited or unclear An Australian study showed that imaging rates
benefit in patients with nonradicular low back for new low back pain problems in patients seen
pain.2224 In the case of spinal fusion, the wide- in general practice increased slightly despite the
spread use of expensive add-ons, such as in- publication of guidelines recommending against
strumentation and bone-morphogenetic proteins, routine imaging.37
have further increased costs, despite little evi-
dence of improved patient outcomes, and in Routine Imaging
some cases emerging evidence of harms.18,2527
In fact, despite spending more on low back pain In one survey, approximately 40% of family prac-
and performing more invasive procedures, clinical tice and 13% of internal medicine physicians re-
progress is difficult to discern. In adults with back ported ordering routine diagnostic imaging for
or neck problems in the United States, self- acute low back pain.31 Another survey of physi-
reported measures of mental health, physical fun- cians found that in the absence of any worrisome
ctioning, work or school limitations, and social features, approximately one-quarter would order
limitations were all similar or poorer in 2005 com- a lumbosacral spine radiograph for acute low
pared with 1997.5 Some data suggest the situation back pain without sciatica, and about two-thirds
may be getting even worse. In North Carolina, the for low back pain with sciatica.38 Data on actual
proportion of adults reporting chronic low back imaging practices are consistent with the survey
pain that impaired activity more than doubled results. One study found that among 35,000 Medi-
between 1992 and 2006, from 3.9% to 10.2%.28 care beneficiaries with acute low back pain and no
diagnostic code indicating a serious underlying
condition, nearly 30% underwent imaging (lumbar
IMAGING PRACTICES radiography or advanced imaging) within 28 days.9
Practice Variations
Advanced Imaging
Clinicians vary substantially in how frequently they
obtain low back pain imaging. One study found Use of advanced spinal imaging is increasing
that Medicare beneficiaries living in high-use rapidly. Among Medicare part B beneficiaries,
geographic areas in the United States were more the number of lumbar MRI scans performed in-
than five times more likely to undergo lumbar spine creased approximately fourfold between 1994
MRI and CT than if they lived in low-use areas.11 and 2005 (Fig. 1).15 Similarly, in a large health
In addition, wide variations in diagnostic testing care organization, the rate of MRIs tripled between
rates have been observed between, and within, 1997 and 2006.39 In North Carolina, more than
medical specialties.2932 One survey found inter- one-third of patients with chronic low back pain
nists almost evenly divided regarding whether underwent lumbar spine MRI or CT within the
572 Chou et al

identify a disease or condition, or how well the

test provides prognostic information. However,
even accurate tests do not necessarily result in
improved patient outcomes. The ultimate effects
of diagnostic testing depend on how clinicians
and patients use the test results, the effectiveness
of subsequent treatments, and harms related to
the diagnostic test and subsequent tests and
treatments. Well-conducted randomized trials are
at the top of the diagnostic evidence hierarchy
because they provide the most direct information
about the clinical benefits and harms of alternative
testing strategies.4446 In the case of low back
Fig. 1. Number of lumbar spine MRIs among Medicare pain, these studies are particularly important
beneficiaries, from Part B claims. (From Deyo RA, Mirza because no adequate reference standard exists
SK, Turner JA, et al. Overtreating Chronic Back Pain: to distinguish symptomatic from asymptomatic
Time to Back Off? J Am Board Fam Med 2009;22:628. common degenerative or age-related findings,
Reproduced by permission of the American Board of
which would be required to estimate the diag-
Family Medicine).
nostic accuracy of the tests for symptomatic low
back pain.
past year,40 and other studies show even higher Spine imaging is one of the few areas of diag-
rates.41 In the emergency department setting, nostic imaging in which multiple randomized trials
one recent study found that use of CT or MRI for reporting clinical outcomes are available. A meta-
low back pain tripled from 2002 to 2006 (3.2% vs analysis of six randomized trials (n 5 1804) of
9.6%; P<.01 for trend).42 MRIs are often ordered patients with primarily acute or subacute low
in patients who have not undergone any treat- back pain and no red flags found no differences
ments, despite recommendations for a trial of between routine lumbar imaging (plain radiog-
therapy before imaging in patients without red raphy, MRI, or CT) and usual care without routine
flags. According to Medicares Hospital Compare imaging on measures of pain, function, quality
Web site, approximately one-third of Medicare of life, or overall patient-rated improvement
patients with low back pain who underwent an (Table 2).10 In fact, for short-term pain, function,
outpatient lumbar spine MRI had not received and quality of life, trends slightly favored usual
any prior conservative treatment.43 care without routine imaging. Despite the percep-
tion that routine imaging can help alleviate patient
EFFECTIVENESS anxiety about back pain,47 routine imaging also
was not associated with better psychological
The ultimate goal of any diagnostic test is to outcomes.10 Patient satisfaction was reported in
improve clinical outcomes. Most studies of diag- only a few trials and effects were mixed, with
nostic tests estimate how accurately they can some trials showing no effect and others showing

Table 2
Results from meta-analysis of randomized controlled trials of routine imaging versus usual care
without routine imaging

Outcome Short-Term (<3 mo) SMD Long-Term (>6 mo to 1 y) SMD

Pain 0.19 ( 0.010.39), three trials 0.04 ( 0.150.07), four trials
Function 0.11 ( 0.290.50), three trials 0.01 ( 0.170.19), four trials
Quality of life 0.10 ( 0.530.34), two trials 0.15 ( 0.330.04), three trials
Mental health 0.12 ( 0.370.62), two trials 0.01 ( 0.320.34), three trials
Overall improvement RR, 0.83 (0.651.06), four trials RR, 0.82 (0.641.05), one trial

Abbreviations: RR, relative risk; SMD, standardized mean difference.

Data are mean SMD or point estimate for RR (95% CI). A negative SMD favors routine imaging for pain and function,
whereas a positive SMD favors routine imaging for quality of life and mental health. For overall improvement, an
RR <1 favors routine imaging.
Data from Chou R, Fu R, Carrino JA, et al. Imaging strategies for low-back pain: systematic review and meta-analysis.
Lancet 2009;373(9662):46372.
 Appropriate Use of Lumbar Imaging 573

that routine imaging was associated with higher Studies also show that of the small proportion of
satisfaction.10 Three of the trials restricted en- patients with a serious or specific underlying con-
rollment to patients older than 50 or 55 years, dition, almost everyone will have an identifiable
and most of the trials enrolled at least some risk factor. In a retrospective study of 963 patients
patients with radiculopathy. The conclusions of with acute low back pain, all 8 with tumors or frac-
the meta-analysis did not seem affected by tures had clinical risk factors.54 A prospective
whether radiography or advanced imaging (MRI study found no cases of cancer in 1170 patients
or CT) was evaluated. younger than 50 years with acute low back pain
and no history of cancer, weight loss, other sign
of systemic illness, or failure to improve.55 Simi-
COST-EFFECTIVENESS
larly, four trials (n 5 399) that enrolled patients
A prerequisite to evaluating the cost-effectiveness without risk factors and obtained imaging in all
of a clinical service is to understand its clinical participants or recorded diagnoses through at
effectiveness.13 In this case, for patients with no least 6 months of clinical follow-up found that no
red flags, routine imaging is no more effective serious conditions were missed.10
than usual care without routine imaging. Perform-
ing imaging is also more expensive. Services that
are more costly than the alternative, yet offer no Weak Correlation Between Imaging Findings
clear clinical advantages (or do more harm than and Symptoms
good), cannot be cost-effective, because they Another reason that routine imaging is not benefi-
will always be associated with higher (or negative) cial is that most lumbar imaging findings are
cost-effectiveness ratios (in this case, the incre- common in people without low back pain. In fact,
mental cost of routine imaging compared with no these imaging findings are only weakly associated
routine imaging divided by the incremental clinical with back symptoms. A systematic review reported
benefit of routine imaging compared with no odds ratios that ranged from 1.2 to 3.3 for the asso-
routine imaging).13,48 ciation between low back pain and disc degenera-
tion on plain radiography, and no association with
WHY DOESNT ROUTINE IMAGING LEAD spondylosis or spondylolisthesis (Fig. 2).56
TO BETTER CLINICAL OUTCOMES?
Favorable Natural History
In most patients with acute back pain, with or
without radiculopathy, substantial improvement in
pain and function occurs in the first 4 weeks, regard-
less of whether and how patients are treated.49,50
Routine imaging is unlikely to improve on this
already favorable prognosis. Thus, the natural
history of low back pain helps explain why routine
imaging does not result in better clinical outcomes.

Low Prevalence of Serious Underlying

Conditions
Another reason routine imaging is not beneficial is
that the frequency of conditions that require urgent
identification (eg, because of the potential for
permanent neurologic sequelae with delayed
diagnosis) is low. In patients with low back pain
in primary care settings, approximately 0.7%
have metastatic cancer, 0.01% spinal infection,
and 0.04% cauda equina syndrome.51,52 Although
vertebral compression fractures (4%) and inflam-
Fig. 2. Lateral radiograph of the lumbar spine shows
matory back disease (<1%) are more common, several common findings: (1) spondylolisthesis: grade 1
the diagnostic urgency for these conditions is not (<25%) anterolisthesis of L4 on L5 (arrows); (2) marked
as great, because they are not generally associ- disc space narrowing at L4/5; (3) sclerosis of inferior
ated with progressive or irreversible neurologic end plate of L4 and superior end plate of L5; and (4)
impairment.52,53 multilevel spondylosis deformans (arrowheads).
574 Chou et al

Although advanced imaging provides both in-

creased contrast and spatial resolution compared
with plain radiography, resulting in greater anatomic
detail, detection of many of the findings seen on
advanced imaging usually does not provide addi-
tional clinically important information. A systematic
review57 reported odds ratios that ranged from
1.8 to 2.8 for the association between chronic
low back pain and disc degeneration on MRI (Figs.
35), similar to the risk estimates observed in studies
of plain radiography. Consistent with these findings,
a randomized trial failed to show any incremental
value of rapid MRI over radiography for evaluating
low back pain in patients referred for imaging by their
primary care physician.58
In fact, most of the findings on advanced
imaging are so common in asymptomatic adults
that they could be viewed as normal signs of
aging. In one cross-sectional study of asymptom-
atic persons aged 60 years or older, 36% had
a herniated disc, 21% had spinal stenosis, and
more than 90% had a degenerated or bulging
disc.59 Other studies have reported similar Fig. 4. Sagittal T2-weighted MR image of the lumbar
results.6062 Recently published studies indicate spine. White arrow points to an annular fissure at
that imaging findings frequently precede symp- L4/5. Black arrow points to a sequestered disc frag-
toms, and changes on imaging do not correlate ment behind the L5 vertebral body.
well with the clinical course. One of the few
prospective studies found that among patients
the onset of low back pain, 84% had unchanged
with documented lumbar imaging findings before
or even improved findings after symptoms devel-
oped.63 Another prospective study found that
presence of disc protrusion on baseline MRI was
a negative predictor of subsequent back pain
(hazard ratio [HR], 0.5; 95% CI, 0.30.9) and pres-
ence of disc extrusion was not predictive (HR, 1.2;
95% CI, 0.43.4).64 Although nerve root contact
and central stenosis were associated with trends

Fig. 3. Sagittal T2-weighted MR image of the lumbar

spine. Black arrow at L4/5 shows disc height loss and Fig. 5. Axial T2-weighted MR image of the lumbar
a mild bulge. White arrow points to the L5/S1 disc, spine. Arrows point to an annular fissure associated
which has low signal, indicating desiccation. with a focal disc protrusion.
 Appropriate Use of Lumbar Imaging 575

toward increased risk of future back pain (HR, 2.2; Table 3

95% CI, 0.68.0 and HR, 1.9; 95% CI, 0.84.8, Average effective radiation doses
respectively), the associations were not statisti-
cally significant. In fact, the only statistically signif- Average Radiation
icant predictor was not an imaging finding, but Imaging Procedure Dose (millisieverts)
rather presence of depression (HR, 2.3; 95% CI,
Chest radiograph 0.02
1.24.4). (posterior-anterior
view)
Minimal Impact on Clinical Decision Making
Lumbar spine plain 1.5
Back imaging also may not affect patient outcomes radiography
because results typically have no important impact Lumbar spine CT 6
on clinical decision making. Imaging studies rarely Diagnostic cardiac 7
reveal unexpected findings. One review of 68,000 catheterization
lumbar radiographic examinations in persons 20
Data from Fazel R, Krumholz HM, Wang Y, et al. Exposure
to 50 years of age with low back pain estimated
to low-dose ionizing radiation from medical imaging
clinically unsuspected findings in 1 out of approxi- procedures. N Engl J Med 2009;361:84957.
mately every 2500 patients.65 In two studies of
patients who underwent lumbar radiography CT.70 In a 20-year-old woman, the estimated
(sample size approximately 100 in each), imaging risk was about twice as high. Lumbar CT may also in-
affected the management plan in only 1 or 2 volve use of iodinated contrast, which is associated
patients each.66,67 Similarly, a randomized trial with nephropathy and hypersensitivity reactions.
found no differences between patients who under- Because lumbar plain radiography is performed
went routine advanced imaging and those who did much more frequently than lumbar CT, it accounts
not undergo imaging in diagnosis and treatment for a greater proportion of the total radiation dose
plans.68 The limited therapeutic impact of routine from medical imaging procedures in the United
imaging could be because the clinical significance States (3.3% vs 0.7% for lumbar CT), despite a lower
of most imaging findings, and therefore what to average effective radiation dose (1.5 mSv).70 The
do about them, is largely unknown. In fact, no average radiation exposure from lumbar radiography
evidence shows that selecting therapies based on is 75 times higher than from chest radiography.70
the presence of the most common imaging findings This fact is of particular concern for women of
(eg, presence of degenerative discs, facet joint child-bearing age, because of the proximity of
arthritis, or bulging discs without nerve root com- lumbar radiography to the gonads, which are difficult
pression) improves outcomes compared with to effectively shield. According to some estimates,
a more generalized approach.8 Imaging findings the amount of female gonadal irradiation from lumbar
are also a poor predictor of prognosis or response radiography is equivalent to the exposure from a daily
to treatment. One study found no association chest radiograph for several years.52
between the presence of common degenerative
changes on MRI and outcomes after therapy.69 Labeling
Spine imaging could result in unintended harms
HARMS
from labeling effects, which occur when patients
Direct Harms
are told that they have a condition of which they
Radiation exposure were not previously aware.72 A classic example
Lumbar plain radiography and CT contributes to of labeling was a study of blood pressure
an individuals cumulative low-level radiation ex- screening in steel workers in Canada, which found
posure, which could promote carcinogenesis increased rates of absenteeism 1 year later in
(Table 3). Lumbar spine CT is associated with an persons diagnosed with hypertension, particularly
average effective radiation dose of 6 millisieverts in persons previously unaware of their diagnosis.73
(mSv).70 Based on the 2.2 million lumbar CT scans The authors concluded that labeling causes
performed in the United States in 2007, one study patients to adopt a sick role and treat them-
projected 1200 additional future cancers.71 An- selves as more fragile. Similar effects may occur
other study estimated that cancer would be in patients who learn that they have findings, often
expected to occur as a result of radiation exposure described as abnormalities, on lumbar imaging. In
in approximately 1 of every 270 women aged 40 one acute low back pain trial that performed
years who underwent coronary angiography,39 a lumbar spine MRI in all patients, those randomized
procedure associated with a similar average ef- to routinely receive their results reported smaller
fective radiation dose compared with lumbar spine improvements in self-rated general health than
576 Chou et al

those who were blinded to the results.74 In another symptoms. In addition, patients may equate a
trial, patients with subacute or chronic back pain decision to not obtain imaging or provide a precise
who underwent routine radiography reported diagnosis with low-quality or suboptimal care, or
more pain and worse overall health status after 3 interpret the decision to not perform imaging as
months and were more likely to seek follow-up implying that their pain is not legitimate or im-
care than those who did not undergo radiog- portant.78 In patients with chronic back pain, the
raphy.75 Knowledge of clinically irrelevant imaging desire for diagnostic tests is a frequent reason
findings might hinder recovery by causing patients for repeat office visits.80
to worry more, focus excessively on minor back Patient preferences about diagnostic testing
symptoms, or avoid exercise and other recom- seem to be communicated to physicians, who
mended activities because of fears that they could frequently accede to patient desires or requests
cause more structural damage, a pattern of for imaging.47 In one study, an increased likelihood
maladaptive coping referred to as fear avoid- of obtaining low back pain imaging was strongly
ance.76 These behaviors are associated with the associated with how intensely patients believed
development of chronic low back pain,77 can be imaging was necessary.81 A survey of physicians
difficult to change, and may be insidious, affecting in the United States showed that over one-third
patients even when they are not consciously would order a lumbar MRI for uncomplicated
aware of them. These potential harms emphasize acute low back pain if a patient insisted on it,
the need for imaging professionals to choose even after explaining that it was unnecessary.82
descriptive language with care, and to recognize Imaging decisions may themselves affect patient
their obligation to educate referring physicians expectations, because those who undergo imaging
and patients regarding the insignificance of age- for one episode of low back pain may come to
related imaging findings. expect it for future episodes. One trial showed
that patients randomized to routine imaging
Downstream Harms became more likely to believe it was necessary
compared with those randomized to an educa-
Association between imaging and surgery tional intervention without routine imaging, despite
Despite all of the uncertainties related to the inter- no beneficial effects on clinical outcomes.83
pretation of imaging tests, patients and clinicians
frequently view findings on imaging as targets for
surgery or other procedures.78 In fact, the associ- Financial Incentives
ation between rates of advanced spine imaging Financial incentives can influence imaging deci-
and rates of spine surgery is strong.19 One study sions. For example, performance incentives may
showed that variation in rates of spine MRI use be linked to markers of patient satisfaction. At the
accounted for 22% of the variability in overall spine same time, performing more imaging tends to
surgery rates in Medicare beneficiaries, or more be associated with higher patient satisfaction.
than double the variability accounted for by differ- Randomized trials show that patients express
ences in patient characteristics.11 In one study, more satisfaction with their care when they un-
patients randomized to rapid MRI had twice the dergo routine lumbar imaging compared with
number of lumbar operations as those receiving no routine imaging,75 and when they undergo
plain radiographs, although small numbers made advanced imaging instead of radiography,58 even
the difference only marginally statistically signifi- when their clinical outcomes are no better. A study
cant.58 Another study found that for work-related of Medicare beneficiaries found earlier use of
acute low back pain, MRI within the first month imaging and more advanced imaging when clini-
was associated with a more than eightfold in- cian incentives were based on patient satisfaction.9
crease in risk for surgery and more than a fivefold Therefore, financial incentives based on patient
increase in subsequent total medical costs satisfaction can encourage overuse of imaging.
compared with propensity-matched controls who From a health systems perspective, financial
did not undergo early MRI.12 incentives may exist for using imaging units. A
top-of-the-line MRI unit can cost $2 million or
more to purchase, and approximately $800,000
WHY ISNT CURRENT PRACTICE CONSISTENT
a year to operate.84,85 At the same time, advanced
WITH THE EVIDENCE?
imaging offers a high profit margin. Relative to
Patient Expectations
actual costs, Medicare provides far greater reim-
One reason that current practice is not con- bursement for MRI (reimbursement-to-cost ratio,
sistent with the evidence is patient expectations.79 2.3) than for conventional radiography (reimburse-
Patients want a specific diagnosis to explain their ment-to-cost ratio, 0.9).86
 Appropriate Use of Lumbar Imaging 577

Research shows that greater availability of Time

imaging units seems to correlate with increased
Another reason that back imaging is overused may
use. The number of MRI scanners in the United
be that clinicians are frequently overworked and
States more than tripled from 2000 to 2005, from
pressed for time, and ordering an imaging test
7.6 to 26.6 per million people.84 In 2006, about
seems more expedient than explaining why imag-
7000 sites in the United States offered MRI
ing is not necessary.47,91 In patients who have
studies,87 which translates into a number of MRIs
strong notions about the need for back imaging,
per capita almost twice as high as in any other
the perceived (or real) time savings may be partic-
industrialized country, and more than four times
ularly high.
as high as in Canada or the United Kingdom.87 In
2006, estimates showed that almost as many
RECOMMENDATIONS ON IMAGING USE
MRI machines were available in western Pennsyl-
When to Image
vania (N 5 140) as in all of Canada (N 5 151).85
One study found that each new MRI unit added Routine imaging in low-risk patients does not
within a geographic area was associated with improve patient outcomes but increases costs
approximately 40 additional lumbar MRIs over and exposes patient to harms, including unneces-
a 5-year period.84 sary radiation exposure and invasive treatments,
Imaging self-referral, or acquisition of imaging and the deleterious effect of likely labeling that
equipment and billing for imaging tests by non- person as a patient with a degenerative spinal
radiologist physicians, is also associated with disorder. Several professional societies have is-
increased imaging use. A 2009 report from the sued practice guidelines and standards to help
Medicare Payment Advisory Commission found address overuse of low back imaging. In 2007,
physician ownership or investment in imaging the American College of Physicians (ACP) and
facilities associated with higher use rates after the American Pain Society (APS) published a joint
adjustment for potential confounders.88 A study clinical practice guideline on diagnosis and treat-
of Medicare claims found that acquisition of ment of low back pain.8 The key recommendations
MRI equipment by primary care physicians in- regarding diagnostic imaging were:
creased MRI use rates from 11% to 17%
(P<.001) in the following 180 days.84 In ortho-  Do not routinely obtain imaging or other
pedic surgery practices, the MRI rates increased diagnostic tests in patients with nonspecific
from 22% to 27% (P<.001). An earlier study of low-back pain
workmans compensation cases found more in-  Perform diagnostic imaging and testing
appropriate imaging requests when physicians when severe or progressive neurologic defi-
self-referred.89 cits are present or when serious underlying
conditions are suspected
 Evaluate patients with persistent low back
Defensive Medicine pain and signs or symptoms of radiculop-
Overuse of back imaging could also be related to athy or spinal stenosis who are candidates
perceived liability risks of missing a serious diag- for surgery or epidural steroid injection.
nosis. Defensive medicine refers to the alteration In 2009, the American College of Radiology
of clinical behavior because of concerns over published consensus-based criteria on appropri-
malpractice liability. Inappropriate imaging seems ateness of imaging for various low back pain
to be a very common form of defensive medicine. scenarios that were largely consistent with the
In one study, more than 90% of physicians from ACP/APS guidelines.92 For uncomplicated low
six specialties in Pennsylvania reported defensive back pain with or without radiculopathy, imaging
medicine practices.90 Almost half of those with was deemed inappropriate in the absence of the
positive responses reported use of imaging in clin- following red flags:
ically unnecessary circumstances as their most
recent defensive act. Defensive medicine prac-  Recent significant trauma or milder trauma
tices are probably more likely when there is a high- at age older than 50 years
er likelihood (or a perceived higher likelihood) of  Unexplained weight loss
a legal claim related to the back pain, or when  Unexplained fever
patients express dissatisfaction. Low back pain  Immunosuppression
imaging is often a routine part of the evaluation in  History of cancer
workmans compensation and disability cases,  Intravenous drug use
despite the absence of evidence that it improves  Prolonged use of corticosteroids or
outcomes in these settings. osteoporosis
578 Chou et al

 Age older than 70 years symptoms, or risk factors for cauda equina syn-
 Focal neurologic deficit with progressive or drome or vertebral infection. Although these con-
disabling symptoms ditions are rare, the prevalence of suggestive
 Duration longer than 6 weeks. findings or risk factors for them is low,94 and timely
diagnosis and treatment may prevent serious
Subsequently, the ACP published more detailed sequelae related to compression of the spinal
guidance on use of lumbar imaging (Table 4).93 cord (which typically ends at the L1 or L2 level)
All of the guidelines are consistent in recommend- or cauda equina. Clinical findings for cauda equina
ing immediate imaging in patients with signs, syndrome or vertebral infection include new

Table 4
Suggestions for imaging in patients with acute low back pain

Imaging Action Suggestions for Initial Imaging

Immediate imaging
Radiography (with erythrocyte Major risk factors for cancer (new onset of low back
sedimentation rate) pain with history of cancer, multiple risk factors
for cancer, or strong clinical suspicion for cancer)
MRI Risk factors for spinal infection (new onset of low
back pain with fever and history of intravenous
drug use or recent infection)
Risk factors for or signs of cauda equina syndrome
(new urinary retention, fecal incontinence, or
saddle anesthesia)
Severe neurologic deficits (progressive motor
weakness or motor deficits at multiple neurologic
levels)
Defer imaging after a trial of therapy
Radiography (with or without erythrocyte Weaker risk factors for cancer (unexplained weight
sedimentation rate) loss or age >50 y)
Radiography Risk factors for or signs of ankylosing spondylitis
(morning stiffness that improves with exercise,
alternating buttock pain, awakening because of
back pain during the second part of the night, or
younger age [2040 y])
Risk factors for vertebral compression fracture
(history of osteoporosis, use of corticosteroids,
significant trauma, or older age [>75 y for men
or >65 y for women])
MRI Signs and symptoms of radiculopathy (back pain
with leg pain in an L4, L5, or S1 nerve root
distribution or positive result on straight leg raise
or crossed straight leg raise test) in patients who
are candidates for surgery or epidural steroid
injection
Risk factors for or symptoms of spinal stenosis
(radiating leg pain, older age, or
pseudoclaudication) in patients who are
candidates for surgery
No imaging No criteria for immediate imaging and back pain
improved or resolved after a 1-month trial
of therapy
Previous spinal imaging with no change in clinical
status

Reproduced from Chou R, Qaseem A, Owens DK, et al. Diagnostic imaging for low back pain: advice for high-value health
care from the American College of Physicians. Ann Intern Med 2011;154:1819; with permission.
 Appropriate Use of Lumbar Imaging 579

urinary retention, saddle anesthesia, fecal inconti- values.54,95 For example, one study of 1172
nence, or fever (especially in patients with risk consecutive patients with acute back pain pre-
factors for bacteremia). Urinary retention is the senting to primary care found that approximately
most common finding for cauda equina syndrome. one-quarter were older than 55 years, ap-
Without urinary retention, the likelihood of cauda proximately one-quarter had morning back stiff-
equina syndrome is on the order of 1 in 10,000.51 ness, and approximately one-third had pain that
Imaging is also recommended for patients with improved with exercise.95 All of these are consid-
severe or progressive neurologic deficits, defined ered risk factors for cancer or inflammatory back
as objective motor weakness at a single level, or pain, but no cases of cancer and only two inflam-
deficits at multiple spinal levels. matory conditions were identified in this cohort.
In patients with findings or risk factors for other A more efficient strategy proposed by the ACP is
specific conditions, such as cancer (age >50 to perform imaging based on the estimated preva-
years, failure to improve after 1 month, weight lence of the condition before risk factor assess-
loss), vertebral compression fracture (older age, ment (the pretest probability) and how strongly
history of or risk factors for osteoporosis, signifi- the risk factors predict the condition (Table 5).93
cant trauma), ankylosing spondylitis (morning stiff- For instance, the prevalence of cancer in a primary
ness, improvement with exercise, younger age, care population is approximately 0.7%.55 A pre-
chronic symptoms), herniated disc (radiculopathy, vious history of cancer (not including nonmelano-
positive straight leg raise), or symptomatic spinal ma skin cancer) is the strongest risk factor for
stenosis (older age, leg pain, pseudoclaudication), spinal tumor (positive likelihood ratio, 15).55 Unex-
optimal diagnostic strategies are less clear. The plained weight loss, failure to improve after 1
traditional recommendation has been to act on month, and age older than 50 years are weaker
all risk factors by obtaining imaging. However, risk factors (positive likelihood ratio, 2.73.0).
the low frequency of these conditions and the Based on these likelihood ratios, the probability
low specificity of most risk factors would result in of cancer if a history of prior cancer is present
high imaging rates with low positive predictive would increase to approximately 9%, or high

Table 5
Using likelihood ratios to update estimated probabilities of spinal tumor in patients with low back pain

Example: 57-Year-Old Man Presents to Primary

Step Care Clinic with 3 Days of Acute Low Back Pain
1. Estimate the pretest probability of A study of patients seen in a primary care
persistent low back pain (the clinic estimated an overall prevalence of low
probability before assessing for back pain of 0.7%55
risk factors)
2. Convert the pretest probability to Pretest probability for cancer 5 0.7% or 7/1000;
pretest odds pretest odds 5 7/(1000 7)a 5 7/993
3. Determine which risk factors for Example A: assume that the patient has a personal
cancer are present, and the associated history of prostate cancer; the likelihood ratio for
likelihood ratio spinal tumor in persons with such a history is 1555
Example B: assume that the patient has a no risk
factors for cancer other than age >50 y; the
likelihood ratio for spinal tumor in patients
aged >50 years is 2.755
4. Multiple the pretest odds by the Example A: posttest odds for spinal tumor 5 7/993 
likelihood ratio to determine the 15 or 105/993
posttest odds Example B: posttest odds for spinal tumor 5 7/993 
2.7 or 19/993
5. Convert the posttest odds to posttest Example A: posttest probability for spinal tumor 5
probability 105/(105 1 993)b 5 105/1098 5 9.6%
Example B: Posttest probability for spinal tumor 5
19/(191993) 5 19/1012 5 1.9%

a
Converting probability to odds: probability x/y 5 odds x/(y x).
b
Converting odds to probability: odds x/y 5 probability x/(x 1 y).
580 Chou et al

enough to warrant immediate imaging (a strong clinical suspicion persists despite initial negative
clinical suspicion for cancer would give a similar tests. For persistent radicular symptoms or spinal
result).94 In patients with any one of the other three stenosis without severe neurologic compromise,
risk factors, the posttest probability only increases advanced imaging should be performed after at
marginally, to 1% to 2%.55 Deferring imaging least a 1-month trial of therapy if patients are inter-
would be reasonable in most cases, unless symp- ested in and candidates for surgery or an epidural
toms fail to improve after several weeks.54,96 For steroid injection.8
patients without signs of neurologic compromise
but with minor risk factors for vertebral compres- Repeat Imaging
sion fracture, ankylosing spondylitis, herniated
disc, or spinal stenosis, a trial of therapy before Repeat back imaging is common. In one study of
imaging is also warranted according to the ACP patients with low back pain in North Carolina
criteria, because delaying imaging in these cir- who had received an MRI or CT of the back,
cumstances is unlikely to result in missed thera- more than half reported a second advanced imag-
peutic opportunities. For herniated disc or spinal ing test within the previous year.40 Although
stenosis, none of the trials showing benefits of evidence on the effects of repeat imaging on
surgery have enrolled patients with fewer than 6 patient outcomes is not available, prospective
to 12 weeks of symptoms, and no serious neuro- studies of repeat imaging indicate that new MRI
logic complications were seen among patients findings are uncommon.60,63 Rather, typical find-
not randomized to immediate surgery.97100 ings are progression of already identified degener-
Diagnostic rules based on the evaluation of ative changes and, in some cases, improvement
multiple risk factors could help better inform and regression over time. Therefore, periodically
imaging decisions but require further develop- performing repeat imaging on a routine basis is
ment. For vertebral compression fracture, one extremely unlikely to be an effective or informative
study evaluated a diagnostic rule based on four approach. Rather, repeat imaging should only be
risk factors (female sex, age >70 years, significant performed for new or changed clinical features,
trauma, or prolonged use of corticosteroids).94 It such as acute or progressive neurologic symp-
found a likelihood ratio of 1.8 (95% CI, 1.12.0) if toms or recent trauma.
one of four risk factors was present, and 15
(95% CI, 7.225) if two or more were present. Patient Education
However, these rules require more external
Patient expectations regarding back imaging are
validation before they can be recommended for
frequently discordant with the evidence.79 How-
general use.
ever, most patients do not want tests that are
unnecessary, costly, or potentially harmful. The
Choice of Imaging Procedure
ACP guidelines recommend education to help
The ACR appropriateness criteria recommend bring patient expectations more in line with the
lumbar plain radiography for the initial evaluation evidence.93 Explaining that risk factor assessment
of low back pain in patients with recent trauma or is sensitive for identifying worrisome conditions
history of osteoporosis, and in persons aged 70 such as cancer or infection, acute low back pain
years or older,92 which is consistent with ACP is highly likely to improve in the first 4 weeks,
recommendations93 on imaging. Both societies and imaging can be performed later if symptoms
recommend lumbar spine and pelvis plain radiog- fail to improve may help reassure some patients
raphy for evaluation of suspected ankylosing spon- that they have been appropriately assessed and
dylitis.92,93 Both also recommend that advanced that the problem is not being simply dismissed.
imaging be reserved for situations in which findings In fact, effective education may be less burden-
are more likely to influence clinical decision making, some than often assumed. One randomized trial
as in patients with major trauma or severe neuro- found that a brief educational intervention re-
logic compromise (objective or progressive motor garding back imaging took less than 5 minutes
weakness, deficits at multiple levels, or suspected and resulted in similar patient satisfaction with
cauda equina syndrome) and in those with risk overall care (and similar clinical outcomes) com-
factors for vertebral infection.93 When cancer is pared with routinely performing lumbar spine plain
not strongly suspected and in the absence of radiography.83 Supplementing face-to-face infor-
neurologic signs, obtaining initial imaging with mation with patient handouts, self-care education
lumbar radiography plus an erythrocyte sedimen- books,101 online materials,93 or other methods
tation rate is a reasonable approach,96 although could be an efficient strategy to reinforce or
advanced imaging may be appropriate when high expand on key educational points.
 Appropriate Use of Lumbar Imaging 581

Effecting Change unnecessary downstream testing and treat-

ments.108 Conversely, the liberal application of
Changing clinician behaviors is challenging. In
terms such as herniation to small or trivial interver-
general, research suggests that active and individ-
tebral disc findings that are unlikely to be clinically
ualized methods are more effective for changing
relevant may imply a diagnosis that supports
clinician behaviors than passive ones, such as
surgical interventions.108 One study found in-
simply providing guidelines.102 Similarly, a system-
clusion of an epidemiologic statement in lumbar
atic review of interventions for increasing appro-
imaging reports describing the prevalence of
priate low back pain imaging found varying
common findings on lumbar spine MRI in asymp-
effects from distribution of educational mate-
tomatic adults was associated with a trend toward
rials.103 Prior authorization requirements for ad-
decreased likelihood of repeat imaging (odds ratio
vanced spine imaging have been imposed by
[OR], 0.21; 95% CI, 0.031.7) and decreased the
many health insurers and are more effective at
likelihood of opioid prescription use (OR, 0.29;
reducing inappropriate imaging than more passive
95% CI, 0.110.77), although no difference was
approaches, but are often viewed as onerous by
seen in the rate of surgical consultations or sur-
physicians.87 As a potentially more acceptable
geries.109 Studies show that primary care clini-
alternative, one randomized trial found that an
cians prefer a more active role from radiologists
intervention consisting of an educational session
in interpreting imaging findings, such as radiology
by local clinical leaders followed by clinician audit
reports that include management recommenda-
and feedback was more effective than no interven-
tions, such as suggestions for potential treat-
tion for reducing inappropriate lumbar imaging.104
ments and referrals.110,111 Additional research is
Another promising method is computer-based
needed to understand effective methods for accu-
decision support that provides information at the
rately describing imaging findings, their potential
time of ordering, such as whether the patient has
clinical significance, and appropriate management
undergone a recent imaging study, and compares
options.
ordering patterns to those of peers.87 Radiologists
Efforts to decrease inappropriate imaging must
could play an important role in making decision
also address financial incentives that influence
support effective through providing mandatory
overuse. For example, clinician incentive pro-
consultation when provider requests for imaging
grams that are based on patient satisfaction may
are inconsistent with guidelines.105 One retrospec-
have a tendency to reward unnecessary testing
tive study showed that one-quarter of imaging
and therefore be counterproductive.9 Rather,
requests for MRI or CT were inappropriate and
financial incentives should be based on how
could have potentially been avoided with decision
well clinicians adhere to guidelines. Other efforts
support, including radiology consultation.106 A
to curb overuse include payment reductions by
prerequisite, of course, is that radiologists must
Medicare to limit rewards for imaging self-
be familiar with and able to discuss guideline
referral, and federal legislation mandating disclo-
recommendations with referring physicians.
sure of physician ownership interests.87,88,112,113
The most effective interventions for reducing
inappropriate imaging may be multifactorial. One
study found that implementing an intervention SUMMARY
that included a requirement for providers to iden-
tify an approved lumbar imaging indication before Strong evidence shows that routine back imaging
ordering an advanced imaging study, institu- does not improve patient outcomes, exposes
tional education programs on appropriate imag- patients to unnecessary harms, and increases
ing, periodic audits and feedback for providers costs. However, imaging practices remain incon-
who ordered imaging tests inconsistent with the sistent with evidence-based guidelines and use
criteria, and availability of same- or next-day phys- continues to rise. Diagnostic imaging studies
ical therapy and consultation for patients in whom should only be performed in selected higher-risk
imaging was not indicated reduced the rate of patients who have severe or progressive neuro-
lumbar MRI for low back pain by 23% (relative logic deficits or are suspected of having a serious
risk, 0.77; 95% CI, 0.670.87).107 or specific underlying condition, and advanced
When imaging tests are performed, radiologists imaging with MRI or CT should be reserved for
play an important role in how findings are commu- patients with a suspected serious underlying
nicated. For example, describing uncomplicated condition or neurologic deficits, or who are can-
degenerative findings as common or incidental didates for invasive interventions. To be most
age-related phenomena rather than as abnormali- effective, efforts to reign in imaging should be
ties could potentially reduce labeling effects and multifactorial and address clinician behaviors,
582 Chou et al

patient expectations and education, and financial 14. Hall FM. Overutilization of radiological examina-
incentives. Radiologists can play a critical role in tions. Radiology 1976;120:4438.
these efforts through providing consultative ex- 15. Deyo RA, Mirza SK, Turner JA, et al. Overtreating
pertise in decision support programs related to chronic back pain: time to back off? J Am Board
appropriateness of imaging requests, and accu- Fam Med 2009;22:628.
rately reporting the prevalence and potential clin- 16. Spelic DC, Kaczmarek RV, Hilohi MC, et al. Nation-
ical significance (or insignificance) of imaging wide surveys of chest, abdomen, lumbosacral spine
findings. radiography, and upper gastrointestinal fluoroscopy:
a summary of findings. Health Phys 2010;98:498514.
REFERENCES 17. Deyo RA. Cascade effects of medical technology.
Annu Rev Public Health 2002;23:2344.
1. Deyo RA, Mirza SK, Martin BI. Back pain preva- 18. Cahill KS, Chi JH, Day A, et al. Prevalence, compli-
lence and visit rates. Spine 2006;31:27247. cations, and hospital charges associated with use
2. Hart LG, Deyo RA, Cherkin DC. Physician office of bone-morphogenetic proteins in spinal fusion
visits for low back pain: Frequency, clinical evalua- procedures. JAMA 2009;302(1):5866.
tion, and treatment patterns from a U.S. national 19. Verrilli D, Welch HG. The impact of diagnostic
survey. Spine 1995;20:119. testing on therapeutic interventions. JAMA 1996;
3. Koes BW, van Tulder MW, Thomas S. Diagnosis 275:118991.
and treatment of low back pain. BMJ 2006; 20. Friedly J, Chan L, Deyo R. Increases in lumbosa-
332(7555):14304. cral injections in the Medicare population. Spine
4. Luo X, Pietrobon R, Sun SX, et al. Estimates and 2007;32:175460.
patterns of direct health care expenditures among 21. Cherkin DC, Deyo RA, Loeser JD, et al. An interna-
individuals with back pain in the United States. tional comparison of back surgery rates. Spine
Spine 2004;29(1):7986. 1994;19:12016.
5. Martin BI, Deyo RA, Mirza SK, et al. Expenditures 22. Brox JI, Sorensen R, Friis A, et al. Randomized clin-
and health status among adults with back and ical trial of lumbar instrumented fusion and cogni-
neck problems. JAMA 2008;299:65664. tive intervention and exercises in patients with
6. Stewart WF, Ricci JA, Chee E, et al. Lost productive chronic low back pain and disc degeneration.
time and cost due to common pain conditions in Spine 2003;28(17):191321.
the US workforce. JAMA 2003;290:244354. 23. Fairbank J, Frost H, Wilson-MacDonald J, et al.
7. Loeser JD, Volinn E. Epidemiology of low back Randomised controlled trial to compare surgical
pain. Neurosurg Clin N Am 1991;2:71325. stabilisation of the lumbar spine with an intensive
8. Chou R, Qaseem A, Snow V, et al. Diagnosis and rehabilitation programme for patients with chronic
treatment of low back pain: a joint clinical practice low back pain: the MRC spine stabilisation trial.
guideline from the American College of Physicians BMJ 2005;330(7502):1233.
and the American Pain Society. Ann Intern Med 24. Fritzell P, Hagg O, Wessberg P, et al. 2001 Volvo
2007;147:47891. award winner in clinical studies: Lumbar fusion
9. Pham HH, Landon BE, Reschovsky JD, et al. versus nonsurgical treatment for chronic low back
Rapidity and modality of imaging for acute low pain: a multicenter randomized controlled trial from
back pain in elderly patients. Arch Intern Med the Swedish Lumbar Spine Study Group. Spine
2009;169:97281. 2001;26(23):252132.
10. Chou R, Fu R, Carrino JA, et al. Imaging strategies 25. Carragee EJ, Hurwitz EL, Weiner BK. A critical
for low-back pain: systematic review and meta- review of recombinant human bone morphogenetic
analysis. Lancet 2009;373(9662):46372. protein-2 trials in spinal surgery: emerging safety
11. Lurie JD, Birkmeyer NJ, Weinstein JN. Rates of concerns and lessons learned. Spine J 2011;
advanced spinal imaging and spine surgery. Spine 116(6):47191.
2003;28(6):61620. 26. Gibson J, Waddell G. Surgery for degenerative
12. Webster BS, Cifuentes M. Relationship of early lumbar spondylosis: updated Cochrane review.
magnetic resonance imaging for work-related Spine 2005;30(20):231220.
acute low back pain with disability and medical 27. Maghout-Juratli S, Franklin GM, Mirza SK, et al.
utilization outcomes. J Occup Environ Med 2010; Lumbar fusion outcomes in Washington state
52:9007. workers compensation. Spine 2006;31(23):271523.
13. Owens DK, Qaseem A, Snow V, et al. High- 28. Freburger JK, Holmes GM, Agans RP, et al. The
value, cost conscious care: concepts for clini- rising prevalence of chronic low back pain. Arch
cians to evaluate benefits, harms, and costs of Intern Med 2009;169:2518.
medical interventions. Ann Intern Med 2010;154: 29. Carey TS, Garrett J. Patterns of ordering diagnostic
17480. tests for patients with acute low back pain. The
 Appropriate Use of Lumbar Imaging 583

North Carolina Back Pain Project. Ann Intern Med 45. Guyatt GH, Tugwell PX, Feeny DH, et al. A frame-
1996;125:80714. work for clinical evaluation of diagnostic technolo-
30. Cherkin DC, Deyo RA, Wheeler K, et al. Physician gies. Can Med Assoc J 1986;134:58794.
variation in diagnostic testing for low back pain. 46. Sackett DL, Haynes RB. The architecture of diag-
Who you see is what you get. Arthritis Rheum nostic research. BMJ 2002;324:53941.
1994;37:1522. 47. Schers H, Wensing M, Huijsmans Z, et al. Imple-
31. Di Iorio D, Henley E, Doughty A. A survey of mentation barriers for general practice guidelines
primary care physician practice patterns and on low back pain. Spine 2001;26:E34853.
adherence to acute low back problem guidelines. 48. Department of Clinical Epidemiology, and Bio-
Arch Fam Med 2000;9:101521. statistics, McMaster University Health Sciences
32. Webster BS, Courtney TK, Huang YH, et al. Survey Centre. How to read clinical journals: VII. To under-
of acute low back pain management by specialty stand an economic evaluation. Can Med Assoc J
group and practice experience. J Occup Environ 1984;130:15429.
Med 2006;48:72332. 49. Pengel LH, Herbert RD, Maher CG, et al. Acute low
33. Wennberg DE. Variation in the delivery of health back pain: systematic review of its prognosis. BMJ
care: the stakes are high. Ann Intern Med 1998; 2003;327:3237.
128:8668. 50. Vroomen P, de Krom M, Knottnerus J. Predicting
34. Fisher ES, Bynum JP, Skinner JS. Slowing the the outcome of sciatica at short-term follow-up. Br
growth of health care costslessons from regional J Gen Pract 2002;52:11923.
variation. N Engl J Med 2009;360:84952. 51. Deyo RA, Rainville J, Kent DL. What can the history
35. Fisher ES, Wennberg DE, Stukel TA, et al. The and physical examination tell us about low back
implications of regional variations in Medicare pain? JAMA 1992;268:7605.
spending. Part 2: health outcomes and satisfaction 52. Jarvik JG, Deyo RA. Diagnostic evaluation of low
with care. Ann Intern Med 2003;138:28898. back pain with emphasis on imaging. Ann Intern
36. Ivanova JI, Birnbaum HG, Schiller M, et al. Real-world Med 2002;137:58697.
practice patterns, health-care utilization, and costs 53. Underwood MR, Dawes P. Inflammatory back pain
in patients with low back pain: the long road in primary care. Br J Rheumatol 1995;34:10747.
to guideline-concordant care. Spine J 2011;11: 54. Suarez-Almazor ME, Belseck E, Russell AS, et al.
62232. Use of lumbar radiographs for the early diagnosis
37. Williams CM, Maher CG, Hancock MJ, et al. Low of low back pain. JAMA 1997;277:17826.
back pain and best practice care. A survey of 55. Deyo R, Diehl A. Cancer as a cause of back pain:
general practice physicians. Arch Intern Med frequency, clinical presentation, and diagnostic
2010;170:2717. strategies. J Gen Intern Med 1988;3:2308.
38. Webster BS, Courtney TK, Huang YH, et al. Brief 56. van Tulder MW, Assendelft WJ, Koes BW, et al.
report: physicians initial management of acute Spinal radiographic findings and nonspecific low
low back pain versus evidence-based guidelines. back pain: a systematic review of observational
J Gen Intern Med 2005;20:11325. studies. Spine 1997;22(4):42734.
39. Smith-Bindman R, Miglioretti DL, Larson EB. Rising 57. Chou D, Samartzis D, Bellabarba C, et al. Degener-
use of diagnostic medical imaging in a large ative magnetic resonance imaging changes in
integrated health system. Health Aff 2008;27: patients with chronic low back pain. Spine 2011;
1491502. 36:54353.
40. Carey TS, Freburger JK, Holmes GM, et al. A long 58. Jarvik JG, Hollingworth W, Martin B, et al. Rapid
way to go. Practice patterns and evidence in magnetic resonance imaging vs radiographs for
chronic low back pain care. Spine 2009;34:71824. patients with low back pain: a randomized
41. Weiner DK, Kim YS, Bonino P, et al. Low back pain controlled trial. JAMA 2003;289:28108.
in older adults? are we utilizing healthcare 59. Boden SD, David DO, Dina TS, et al. Abnormal
resources wisely? Pain Med 2006;7:14350. magnetic-resonance scans of the lumbar spine in
42. Friedman BW, Chilstrom M, Bijur PE, et al. Diag- asymptomatic subjects. J Bone Joint Surg 1990;
nostic testing and treatment of low back pain in 72:4038.
United States emergency departments. Spine 60. Jarvik JJ, Hollingworth W, Heagerty P, et al. The
2010;35:E140611. longitudinal assessment of imaging and disability
43. U.S. Department of Health and Human Services. of the back (LAIDBACK) study. Spine 2001;26:
Hospital compare. Available at: http://www.hospital 115866.
compare.hhs.gov/hospital-search.aspx. Accessed 61. Jensen MC, Brant-Zawadzki MN, Obuchowski N,
April 22, 2012. et al. Magnetic resonance imaging of the lumbar
44. Fryback DG, Thornbury JR. The efficacy of diag- spine in people without back pain. N Engl J Med
nostic imaging. Med Decis Making 1991;11:8894. 1994;331:6973.
584 Chou et al

62. Wiesel SW, Tsourmas N, Feffer HL, et al. 1984 Volvo 77. Chou R, Shekelle P. Will this patient develop persis-
Award in clinical sciences. A study of computer- tent disabling low back pain? JAMA 2010;303:
assisted tomography. I. The incidence of positive 1295302.
CAT scans in an asymptomatic group of patients. 78. Rhodes LA, McPhillips-Tangum CA, Markham C,
Spine 1984;9:54951. et al. The power of the visible: the meaning of diag-
63. Carragee E, Alamin T, Cheng I, et al. Are first-time nostic tests in chronic back pain. Soc Sci Med
episodes of serious LBP associated with new MRI 1999;48:1189203.
findings? Spine J 2006;6:62435. 79. Verbeek J, Sengers M, Riemens L, et al. Patient
64. Jarvik JG, Hollingworth W, Heagerty PJ, et al. expectations of treatment for back pain. Spine
Three-year incidence of low back pain in an initially 2004;29:230918.
asymptomatic cohort. Spine 2005;30:15418. 80. McPhillips-Tangum CA, Cherkin DC, Rhodes LA,
65. Nachemson A. The lumbar spine: an orthopaedic et al. Reasons for repeated medical visits among
challenge. Spine 1975;1:5971. patients with chronic back pain. J Gen Intern
66. Halpin SF, Yeoman L, Dundas DD. Radiographic Med 1998;13:28995.
examination of the lumbar spine in a community 81. Wilson IB, Dukes K, Greenfield S, et al. Patients
hospital: an audit of current practice. BMJ 1991; role in the use of radiology testing for common
303:8135. office practice complaints. Arch Intern Med 2001;
67. Rockey P, Tompkins R, Wood R, et al. The useful- 161:25663.
ness of x-ray examinations in the evaluation of 82. Campbell EG, Regan S, Gruen RL, et al. Profes-
patients with back pain. J Fam Pract 1978;7: sionalism in medicine: results of a national survey
45565. of physicians. Ann Intern Med 2007;147:795802.
68. Gillan MG, Gilbert FJ, Andrew JE, et al. Influence of 83. Deyo RA, Diehl AK, Rosenthal M. Reducing roent-
imaging on clinical decision making in the treat- genography use. Can patient expectations be
ment of lower back pain. Radiology 2001;220: altered? Arch Intern Med 1987;147:1415.
3939. 84. Baras JD, Baker LC. Magnetic resonance imaging
69. Kleinstuck F, Dvorak J, Mannion AF. Are structural and low back pain care for Medicare patients.
abnormalities on magnetic resonance imaging Health Aff 2009;28:w113340.
a contraindication to the successful conservative 85. Snowbeck C. Region still rich with MRI machines.
treatment of chronic nonspecific low back pain? Pittsburgh: Pittsburgh Post-Gazette; 2005.
Spine 2006;31:22507. 86. Gray DT, Deyo RA, Kreuter W, et al. Population-
70. Fazel R, Krumholz HM, Wang Y, et al. Exposure based trends in volumes and rates of ambulatory
to low-dose ionizing radiation from medical im- lumbar spine surgery. Spine 2006;31:195763.
aging procedures. N Engl J Med 2009;361: 87. Iglehart JK. Health insurers and medical-imaging
84957. policya work in progress. N Engl J Med 2009;
71. Berrington de Gonzalez A, Mahesh M, Kim KP, 360:10307.
et al. Projected cancer risks from computed tomo- 88. Medicare Payment Advisory Commission. Report to
graphic scans performed in the United States in the congress: aligning incentives in the Medicare
2007. Arch Intern Med 2009;169:20717. program. Available at: http://www.medpac.gov/
72. Fisher ES, Welch HG. Avoiding the unintended documents/Jun09_EntireReport.pdf. Accessed April
consequences of growth in medical care. JAMA 6, 2012.
1999;281:44653. 89. Swedlow A, Johnson G, Smithline N, et al.
73. Haynes RB, Sackett DL, Taylor DW, et al. Increased Increased costs and rates of use in the California
absenteeism from work after detection and labeling workers compensation system as a result of self-
of hypertensive patients. N Engl J Med 1978;299: referral by physicians. N Engl J Med 1992;327:
7414. 15026.
74. Gilbert FJ, Grant AM, Gillan MG, et al. Low back 90. Studdert DM, Mello MM, Sage WM, et al. Defensive
pain: influence of early MR imaging or CT on treat- medicine among high-risk specialist physicians in
ment and outcomemulticenter randomized trial. a volatile malpractice environment. JAMA 2005;
Radiology 2004;231:34351. 293:260917.
75. Kendrick D, Fielding K, Bentley E, et al. Radiog- 91. Anonymous Is time management an important
raphy of the lumbar spine in primary care patients cause of excessive imaging?. Backletter 2009;
with low back pain: randomised controlled trial. 24:50.
BMJ 2001;322:4005. 92. Davis PC, Wippold FJ II, Brunberg JA, et al. ACR
76. Leeuw M, Goossens ME, Linton SJ, et al. The fear- appropriateness criteria on low back pain. J Am
avoidance model of musculoskeletal pain: current Coll Radiol 2009;6:4017.
state of scientific evidence. J Behav Med 2007; 93. Chou R, Qaseem A, Owens DK, et al. Diagnostic
30:7794. imaging for low back pain: advice for high-value
 Appropriate Use of Lumbar Imaging 585

health care from the American College of Physi- 104. Schectman JM, Schroth WS, Verme D, et al.
cians. Ann Intern Med 2011;154:1819. Randomized controlled trial of education and
94. Henschke N, Maher CG, Refshauge KM. Screening feedback for implementation of guidelines for
for malignancy in low back pain patients: a system- acute low back pain. J Gen Intern Med 2003;18:
atic review. Eur Spine J 2007;16:16739. 77380.
95. Henschke N, Maher CG, Refshauge KM, et al. 105. Khorasani R. Can radiology professional society
Prevalence of and screening for serious spinal guidelines be converted to effective decision sup-
pathology in patients presenting to primary care port? J Am Coll Radiol 2010;7:5612.
settings with acute low back pain. Arthritis Rheum 106. Lehnert BE, Bree RL. Analysis of appropriateness
2009;60:307280. of outpatient CT and MRI referred from primary
96. Joines JD, McNutt RA, Carey TS, et al. Find- care clinics at an academic medical center: how
ing cancer in primary care outpatients with low critical is the need for improved decision support?
back pain: a comparison of diagnostic strategies. J Am Coll Radiol 2010;7:1927.
J Gen Intern Med 2001;16:1423. 107. Blackmore CC, Mecklenburg RS, Kaplan GS.
97. Peul W, van den Hout W, Brand R, et al. Prolonged Effectiveness of clinical decision support in control-
conservative care versus early surgery in patients ling inappropriate imaging. J Am Coll Radiol 2011;
with sciatica caused by lumbar disc herniation: 8:1925.
two year results of a randomised controlled trial. 108. Breslau J, Seidenwurm D. Socioeconomic aspects
BMJ 2008;336:13558. of spinal imaging: impact of radiological diagnosis
98. Weinstein J, Tosteson T, Lurie J, et al. Surgical on lumbar spine-related disability. Top Magn Reson
versus nonsurgical therapy for lumbar spinal Imaging 2000;11:21823.
stenosis. N Engl J Med 2008;358(8):794810. 109. McCullough BJ, Johnson GR, Martin BI, et al.
99. Weinstein JN, Lurie JD, Tosteson TD, et al. Surgical Lumbar spine MR imaging and reporting: do
versus nonsurgical treatment for lumbar degenera- epidemiological data in reports affect clinical
tive spondylolisthesis. N Engl J Med 2007;356(22): management? Radiology 2012;262(3):9416.
225770. 110. Eskander MG, Leung A, Lee D. Style and content
100. Weinstein JN, Lurie JD, Tosteson TD, et al. Sur- of CT and MR imaging lumbar spine reports: radiol-
gical vs nonoperative treatment for lumbar disk ogist and clinician preferences. Am J Neuroradiol
herniation: the Spine Patient Outcomes Research 2010;31:18427.
Trial (SPORT) observational cohort. JAMA 2006; 111. Grieve FM, Plumb AA, Khan SH. Radiology report-
296(20):24519. ing: a general practitioners perspective. Br J Radiol
101. Burton AK, Waddell G, Tillotson KM, et al. Informa- 2010;83:1722.
tion and advice to patients with back pain can have 112. Hillman BJ, Goldsmith J. Imaging: the self-
a positive effect. A randomized controlled trial of referral boom and the ongoing search for effec-
a novel educational booklet in primary care. Spine tive policies to contain it. Health Aff 2010;29:
1999;24:248491. 22316.
102. Grimshaw JM, Shirran L, Thomas R, et al. Chang- 113. Report to Congressional Requesters. Medicare
ing provider behavior: an overview of system- part B imaging services. Rapid spending growth
atic reviews of interventions. Med Care 2001; and shift to physician offices indicate need for
39(Suppl 2):II245. CMS to consider additional management prac-
103. French SD, Green S, Buchbinder R, et al. Interven- tices, GAO-08-452. Washington, DC: United States
tions for improving the appropriate use of imaging Government Accountability Office; 2008. Available
in people with musculoskeletal conditions. Co- at: http://www.gao.gov/new.items/d08452.pdf. Ac-
chrane Database Syst Rev 2010;1:CD006094. cessed April 6, 2012.