Accepted Manuscript A Systematic Review of Diagnostic Imaging Utilization for Low Back Pain in the United States Simon Dagenais, DC, PhD, MSc Erin K. Galloway, BA Darren M. Roffey, PhD PII:
S1529-9430(13)01626-4
DOI:
10.1016/j.spinee.2013.10.031
Reference:
SPINEE 55625
To appear in:
The Spine Journal
Received Date: 8 June 2012 Revised Date:
23 August 2013
Accepted Date: 23 October 2013
Please cite this article as: Dagenais S, Galloway EK, Roffey DM, A Systematic Review of Diagnostic Imaging Utilization for Low Back Pain in the United States, The Spine Journal (2013), doi: 10.1016/ j.spinee.2013.10.031. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
A SYSTEMATIC REVIEW OF DIAGNOSTIC IMAGING UTILIZATION FOR LOW BACK PAIN IN THE UNITED STATES
RI PT
Simon Dagenais, DC, PhD, MSc 1; Erin K. Galloway, BA 1; Darren M. Roffey, PhD 1,2,3
Palladian Health, West Seneca, NY 14224, USA
2.
Clinical Epidemiology Program, Ottawa Hospital Research Institute, Ottawa, Ontario
SC
1.
K1Y 4E9, Canada
University of Ottawa Spine Program, The Ottawa Hospital, Ottawa, Ontario K1Y 4E9, Canada
M AN U
3.
Please address correspondence and reprint requests to:
2732 Transit Road,
TE D
Simon Dagenais, DC, PhD, MSc
West Seneca, NY, 14224, USA
EP
Phone: (716) 712-2700;
AC C
Email:
[email protected] ACCEPTED MANUSCRIPT
ABSTRACT
2
BACKGROUND CONTEXT: Various studies have previously reported on the rising
3
utilization and costs of diagnostic imaging for low back pain (LBP) in the United States (US).
4
However, it is unclear if the methods used in these studies allowed for meaningful comparisons,
5
or if the reported utilization data can be used to develop evidence-based benchmarks.
6
PURPOSE: The primary purpose of this study was to review previous estimates of the
7
utilization of diagnostic imaging for LBP reported in the US.
8
STUDY DESIGN/SETTING: Systematic review of published literature.
9
METHODS: A search through May 2012 was conducted using keywords and free text terms
M AN U
SC
RI PT
1
related to health services and LBP in Medline and Health Policy Reference; results were
11
screened for relevance independently, and full-text studies were assessed for eligibility. Only
12
studies published in English since the year 2000 reporting on utilization of diagnostic imaging
13
for LBP using claims data from the US were included. Reporting quality was assessed using a
14
modified Downs and Black tool for observational studies. This study was funded by Palladian
15
Health. Study authors were paid consultants and shareholders of Palladian Health when this
16
study was conducted.
17
RESULTS: The search strategy yielded 1102 citations, of which 7 met the criteria for eligibility.
18
Studies reported utilization from commercial health plans (n=4) and Medicare (n=3), with
19
sample sizes ranging from 13,760 to 740,467 members with LBP from specific states or across
20
the US. The number of diagnostic codes used to identify nonspecific LBP ranged from 2 to 66;
21
other heterogeneity was noted in the methods used across these studies. In commercial health
22
plans, utilization of x-rays was 12.0% to 32.2% of patients with LBP, magnetic resonance
23
imaging (MRI) was used in 16.0% to 21.0%, computed tomography (CT) was used in 1.4% to
AC C
EP
TE D
10
1
ACCEPTED MANUSCRIPT
3.0%, and MRI and/or CT was used in 10.9% to 16.1%. Findings in Medicare populations were
2
22.9% to 48.2% for x-rays, 11.6% for MRI, and 10.4% to 16.3% for MRI and/or CT.
3
CONCLUSIONS: The reported utilization of diagnostic imaging for LBP varied across the
4
studies reviewed; differences in methodology made meaningful comparisons difficult.
5
Standardizing methods for performing and reporting analyses of claims data related to utilization
6
could facilitate efforts by third-party payers, health care providers, and researchers to identify
7
and address the perceived overuse of diagnostic imaging for LBP.
SC
8
10
Word count
M AN U
9
RI PT
1
377
11
Keywords
13
Low back pain; diagnostic imaging; x-rays; magnetic resonance imaging; computed tomography;
14
utilization; claims data
AC C
EP
15
TE D
12
2
ACCEPTED MANUSCRIPT
INTRODUCTION
2
Low back pain (LBP) is experienced by 25-50% of the adult population in the United States (US)
3
each year, making it one of the most common reasons for seeking health care from a variety of
4
clinicians, including primary care providers (PCPs), nonsurgical specialists, spine surgeons,
5
allied health providers, as well as complementary and alternative medicine (CAM) providers (1-
6
3). The costs associated with health care services for spine pain (primarily LBP) in the US rose
7
from $45.9 billion in 1997 to $102.6 billion in 2004, an annualized growth rate of over 12% (4).
8
Yet despite the increased resources allocated to health services for LBP, their value has been
9
questioned due to rising chronicity, disability, and unexplained variations in care (4-8).
M AN U
SC
RI PT
1
10
One of the contributors to the growing costs of LBP is the increased use of medical technologies,
12
particularly advanced diagnostic imaging such as magnetic resonance imaging (MRI) and
13
computed tomography (CT), which now account for nearly 14% of Medicare part B (medical
14
insurance) expenditures (7;9-12). Use of spinal MRI in Medicare increased by 83% from 1993 to
15
1998, while lumbar MRI increased by 300% between 1994 and 2006 (7;10). Numerous factors
16
have been offered to explain the rise of diagnostic imaging for LBP, including changing
17
demographics, increased care seeking and patient expectations about LBP, increased physician
18
ownership of imaging facilities, and reimbursement fee-for-service payment models (10;11;13-
19
16). The supply of imaging equipment may also play a role, as the number of MRI scanners in
20
the US increased from 7.6 to 26.6 per 1 million people between 2000 and 2005, and ongoing
21
utilization is necessary to recoup initial and operating expenses (11;17).
AC C
EP
TE D
11
22
3
ACCEPTED MANUSCRIPT
Researchers have observed large geographic variations in the utilization of diagnostic imaging
2
for LBP across the US that seemingly cannot be attributed to clinical need alone, leading to
3
concerns that these services are not always clinically necessary, compounded by the subsequent
4
utilization of related health services (10;18-20). For example, MRI frequently detects anatomical
5
irregularities that are not responsible for the symptoms of LBP observed, but which may
6
nevertheless trigger a cascade of additional diagnostic testing or surgical procedures aimed at
7
correcting anomalies on diagnostic imaging (17;20;21). Concerns have also been voiced about
8
other harms associated with diagnostic imaging, including the ionizing radiation of spinal x-rays
9
and CT that increase with repeated exposure (22).
M AN U
SC
RI PT
1
10
Reports of increased utilization, high costs, and potentially inappropriate use of diagnostic
12
imaging for LBP have fueled interest in this topic by third-party payers, including commercial,
13
employer-sponsored, and government-sponsored health plans, as well as health services
14
researchers and health administrators eager to identify potential targets for cost containment,
15
particularly when evidence-based clinical practice guidelines (CPGs) are available to guide
16
appropriate utilization (13). For example, the proportion of patients with LBP receiving
17
diagnostic imaging has been endorsed as a measure of health care quality by the National
18
Committee for Quality Assurance (NCQA), is currently incorporated into the Healthcare
19
Effectiveness Data and Information Set (HEDIS) measures reported by a majority of health
20
plans, and will likely be included in quality measures for patient-centered medical homes
21
(23;24).
AC C
EP
TE D
11
22
4
ACCEPTED MANUSCRIPT
Efforts to develop programs targeting the potential overuse of health services often begin by
2
analyzing claims data, which are readily available to third-party payers, administrators, and
3
researchers, are directly relevant to the populations of interest, and offer relatively large sample
4
sizes (25). Findings from such analyses can then be compared to various proposed benchmarks
5
to determine if overuse is occurring, which can prompt administrative, reimbursement, or policy
6
changes that can affect patients with LBP and spine care clinicians. However, it is unclear if
7
meaningful benchmarks related to the utilization of diagnostic imaging for LBP can in fact be
8
developed based on existing studies reporting on utilization of these health services.
SC
RI PT
1
M AN U
9
The primary objective of this study was therefore to conduct a systematic review to identify,
11
appraise, summarize, and synthesize studies reporting on the utilization of diagnostic imaging for
12
LBP in the US. Secondary objectives were to compare the methods used for such analyses, and
13
describe factors that have been reported to influence utilization of diagnostic imaging for LBP.
TE D
10
14
METHODS
16
Information sources and search
17
Medline was searched through May 2012 using the Medical Subject Headings (MeSH) and free
18
text terms developed by the Cochrane Back Review Group to identify studies related to LBP,
19
with additional terms related to utilization, claims, and third-party reimbursement (Appendix 1)
20
(26). A broad search strategy was used to identify studies related to any health service for LBP;
21
findings not related to diagnostic imaging will be reported in future studies. The Medline search
22
strategy was modified for the Health Policy Reference Center (HPRC) database, using both
AC C
EP
15
5
ACCEPTED MANUSCRIPT
1
subject headings and free-text to search abstracts of academic journals. Author files and
2
references from relevant studies were also scanned to uncover additional studies.
3
Eligibility criteria
5
Studies were included if they reported on utilization of any diagnostic imaging for LBP in adults
6
(aged 18 or older) in the US based on third-party payer records (e.g. claims). Studies were
7
excluded if they were published prior to 2000, in languages other than English, relied on self-
8
reported utilization (e.g. surveys), did not report primary data (e.g. reviews), did not report the
9
International Classification of Diseases-Version 9 (ICD-9) diagnostic codes used in the analysis,
SC
M AN U
10
RI PT
4
or could not attribute utilization to LBP (e.g. use of lumbar spine MRI for any indication).
11
Screening
13
Search results were screened by two reviewers to determine eligibility; if eligibility could not be
14
determined from the search records, the full-text article was obtained. Reviewer consistency was
15
examined and calibrated using the first 50 search records. Conflicts were resolved by discussion
16
until consensus was achieved, or by using a third reviewer. Studies meeting eligibility criteria
17
were assessed for reporting quality, and relevant data were abstracted.
EP
AC C
18
TE D
12
19
Quality assessment
20
Since authors could not locate a validated tool to assess the reporting quality of utilization
21
studies, one was developed by modifying the Downs and Black checklist for observational
22
studies (Appendix 2) (27;28). Reporting quality was assessed by determining if the following
23
items were present: 1. type of third-party payer, 2. type of health plan management, 3. date/year 6
ACCEPTED MANUSCRIPT
of data analyzed, 4. time interval for utilization, 5. general eligibility criteria, 6. LBP eligibility
2
criteria, 7. total population size, 8. LBP population size, 9. current procedural terminology (CPT)
3
codes used to identify health services, and 10. utilization reported separately for each type of
4
health service. Quality was assessed independently by two reviewers, who discussed potential
5
conflicts until consensus was reached.
RI PT
1
6
Data items and data collection process
8
Data used to assess reporting quality were also abstracted into evidence tables by two reviewers
9
independently and discussed until consensus was reached. Utilization data for each type of
M AN U
SC
7
diagnostic imaging (e.g. x-rays, MRI, CT, MRI and/or CT) were abstracted using raw data (i.e.
11
numerator and denominator) if available, or rates otherwise, for all claimants and various
12
subgroups of interest (e.g. etiology of LBP, duration of LBP, severity of LBP, clinician specialty,
13
current/previous utilization of other health services).
TE D
10
14
Funding and potential conflicts of interest
16
This study was funded by Palladian Health, a company that manages specialty health benefits
17
such as chiropractic, physical therapy, physical fitness, and complementary and alternative
18
medicine on behalf of health plans and insurers. Although Palladian Health does not currently
19
manage the utilization of diagnostic imaging services related to low back pain (other than x-rays
20
performed by chiropractors), it may offer such services in the future. Study authors are paid
21
consultants (SD, DMR, EKG) and shareholders (SD) of Palladian Health.
AC C
EP
15
22
7
ACCEPTED MANUSCRIPT
RESULTS
2
Literature search
3
The literature search returned 1102 citations, including 897 from MEDLINE and 205 from
4
HPRC, of which 24 were duplicates, yielding 1078 unique citations. Upon screening the
5
abstracts and search records, 1010 were deemed not relevant, and full-text articles were obtained
6
for 68 studies, 62 of which were identified from Medline, and 6 from HPRC. Based on full-text
7
articles, 61 studies were excluded, most commonly because they did not state the ICD-9
8
diagnosis codes used to identify claims related to LBP (n=18), did not identify LBP as the
9
indication for utilizing various health services (n=9), or were not related to diagnostic imaging
SC
M AN U
10
RI PT
1
(n=8) (Figure 1). Seven studies met the stated eligibility criteria (11;13;20;29-32).
11
Study characteristics
13
The seven included studies reported on utilization in commercial (n=4) and government-
14
sponsored (i.e. Medicare) health plans (n=3) at the regional (n=2), state (n=2), or national (n=3)
15
level, with population sample sizes ranging from 165,569 to over 8 million individuals (mean
16
1,678,102; median 453,265) (Table 1). One study analyzed data from a staff-model health
17
maintenance organization (HMO) for patients of all ages, including those who may have been
18
enrolled in managed Medicare health plans (30). As utilization was reported for all ages and only
19
21.2% of patients were older than 65, findings are presented along with other commercial health
20
plans rather than Medicare. The number of years of data analyzed ranged from 1 to 8 (mean 3.9;
21
median 3), while utilization was measured over a period that varied from 6 months to 36 months.
22
While the stated eligibility criteria for each study differed, most focused on measuring utilization
23
of diagnostic imaging for patients with nonspecific (i.e. mechanical) LBP, generally defined by
AC C
EP
TE D
12
8
ACCEPTED MANUSCRIPT
excluding serious spinal pathology such as cancer, trauma, or infection. Four of the studies
2
required that individuals be enrolled for a minimum amount of time – from 6 to 24 months –
3
before and/or after the initial claim identifying LBP, termed the “index” visit. The number of
4
individuals with LBP in the studies included varied from 13,760 to 740,467 (mean 152,706;
5
median 34,408), representing 2.6% to 8.3% of all individuals in a health plan. Five studies
6
reported on the utilization of x-rays for LBP, four studies on MRI for LBP, two studies on CT for
7
LBP, and four studies reported on MRI and/or CT for LBP; findings are presented below by type
8
of modality.
SC
RI PT
1
M AN U
9
Reporting quality
11
The mean number of items reported in each study varied from 5 to 9 (mean 7; median 7) of the
12
10 assessed (Table 2). The most commonly reported item was the date/year of data analyzed
13
(n=7), followed by type of third-party payer (n=6), utilization for a given time interval (n=6),
14
general eligibility criteria (n=6), and number of individuals with LBP (n=6). The least commonly
15
reported items were total population size (n=3), and utilization reported separately for each type
16
of health service (n=3).
EP
17
TE D
10
Diagnostic codes
19
A total of 76 unique ICD-9 diagnosis codes were used by the studies included to identify health
20
care utilization related to LBP (Table 3). The number of codes used in each study varied from 2
21
to 69, and the most common were 724.2 lumbago and 724.5 unspecified backache, both used in
22
all 7 included studies. Three studies cited a list of 66 ICD-9 diagnosis codes developed by
23
Cherkin et al. to identify claims for health services related to mechanical LBP (e.g. herniated
AC C
18
9
ACCEPTED MANUSCRIPT
disc, degenerative changes, spinal stenosis, instability, closed fractures, nonspecific backache,
2
failed spine surgery, and other diagnoses) that excluded serious spinal pathology (e.g.
3
malignancy, infection) (33). One study cited a list of ICD-9 diagnosis codes related to LBP that
4
were developed by the NCQA for use in HEDIS measures (13;34).
5
RI PT
1
Utilization of spinal x-rays
7
Three of the five studies measuring utilization of x-rays reported the CPT codes used to identify
8
relevant claims; one included only lumbar x-rays, two also included full-spine and sacral x-rays,
9
while one also included cervical spine x-rays (11;13;32). In commercial health plans, utilization
M AN U
SC
6
of x-rays was 12.0% over 24 months, 30.9% over 12 months, and 32.2% over 12-36 months
11
among those with mechanical LBP. Utilization of x-rays appeared to increase with the number of
12
episodes of LBP, from 6.3% for those with only 1 episode of LBP (i.e. 30-day period with claim
13
for utilization of health services related to LBP) to 30.9% for those with 6 or more episodes of
14
LBP (30). Similarly, utilization increased with duration of LBP, with higher rates reported for
15
those with (44.0%) than without (29.9%) chronic LBP (i.e. more than 3 months) (32). Use of x-
16
rays also varied by etiology of LBP, being highest among those with no neurologic findings
17
(32.8%) and lowest among those with other types of LBP (16.8%), and slightly lower among
18
those with nonspecific LBP (27.4%) (31;32). Use of x-rays was similar for those with (31.4%) or
19
without (29.0%) claims for opioid analgesics, and higher among those with (76.1%) than without
20
(31.3%) claims for spinal surgery (31;32). The mean time interval between the index claim for
21
LBP and subsequent utilization of x-rays was 34.7 days (32).
AC C
EP
TE D
10
22
10
ACCEPTED MANUSCRIPT
In Medicare populations, utilization of x-rays was 22.9% over 36 months for acute
2
uncomplicated LBP and 48.2% over 12 months for nonspecific LBP (11;13). For those who
3
received only x-rays, utilization was slightly lower among those with (19.2%) than without
4
(23.5%) dual Medicare-Medicaid eligibility, but did not differ for those above (23.3%) or below
5
(22.5%) median income; utilization was notably higher among patients of clinicians whose
6
compensation was partially related to patient satisfaction incentives (46.9%) (13).
RI PT
1
SC
7
Utilization of spinal MRI
9
Two of the four studies measuring utilization of MRI for LBP reported the CPT codes used to
10
identify relevant claims; one included only lumbar MRI, while the other also included cervical
11
and thoracic MRI, with or without contrast (20;32). In commercial health plans, utilization of
12
MRI was 16.0% over 12 months, 18.0% over 12-36 months, and 21.0% over 19 months among
13
those with mechanical LBP (29;31;32). Utilization of MRI was lower for those with nonspecific
14
LBP (8.5%), and higher for those with (33.3%) than without (14.9%) chronic LBP (32). Use of
15
MRI was higher among those with (75.7%) than without (16.8%) claims for spinal surgery (32).
16
The mean time interval between the index claim for LBP and subsequent utilization of MRI was
17
63.6 days (32). In Medicare, utilization of MRI was 11.6% over 6 months in those with
18
nonspecific LBP, higher among patients who consulted with an orthopedist (20.1%) than with a
19
PCP (10.5%), and higher among those clinicians with (11.4%) than without (20.6%) ownership
20
in a MRI facility (20).
TE D
EP
AC C
21
M AN U
8
22 23
11
ACCEPTED MANUSCRIPT
Utilization of spinal CT
2
Only one of the two studies measuring utilization of CT for LBP reported the CPT codes used to
3
identify relevant claims, which included lumbar CT with or without contrast (31;32). In
4
commercial health plans, utilization of CT was 3.0% over 12 months and 1.4% over 12-36
5
months for those with mechanical LBP (31;32). Utilization of CT was lower for those with
6
nonspecific LBP (0.4%), and higher for those with (3.5%) chronic LBP than non-chronic LBP
7
(1.0%) (32). Use of CT was much higher among those with (14.0%) than without (1.2%) claims
8
for spinal surgery (32). The mean time interval between the index claim for LBP and subsequent
9
utilization of CT was 81.3 days (32). No studies were identified that reported utilization of CT
SC
for LBP in a Medicare population.
11
M AN U
10
RI PT
1
Utilization of spinal MRI and/or CT
13
Two of the four studies measuring utilization of MRI and/or CT for LBP reported the CPT codes
14
used to identify relevant claims; one included only lumbar MRI or CT, while the other also
15
included cervical and thoracic MRI, with or without contrast (11;13). In commercial health
16
plans, utilization of MRI and/or CT was 10.9% over 24 months and 16.1% over 12-36 months
17
for nonspecific LBP (30;31). Utilization of MRI and/or CT appeared to increase with the number
18
of episodes of LBP, from 4.6% for those with only 1 episode of LBP (i.e. 30-day period with
19
claim for utilization of health services related to LBP) to 37.3% for those with 6 or more
20
episodes of LBP (30). Utilization of MRI and/or CT was higher for those with (24.2%) than
21
without (11.1%) claims for opioids, higher for those with acquired (20.4%) than congenital
22
(3.4%) conditions related to LBP, and similar for those with (17.5%) and without (17.3%)
23
neurologic findings (31). In Medicare, utilization of MRI and/or CT was 16.3% over 12 months
AC C
EP
TE D
12
12
ACCEPTED MANUSCRIPT
in those with nonspecific LBP and 10.4% over 36 months for uncomplicated acute LBP (11;13).
2
Utilization of MRI and/or CT was lower among those with (7.3%) than without (11.0%) dual
3
Medicare-Medicaid eligibility, and lower for those below (9.8%) than above (11.3%) median
4
income (13). Use of MRI and/or CT was much higher among patients of clinicians whose
5
compensation was partially related to patient satisfaction incentives (17.1%) (13).
RI PT
1
6
DISCUSSION
8
The search strategy identified over 1,000 citations related to utilization of health services for
9
LBP - the majority from Medline - indicating some level of interest in this topic by researchers.
10
Many of those studies were excluded from this review because they did not identify the ICD-9
11
diagnosis codes used to determine that utilization was in fact related to LBP. Although
12
assumptions can be made about the likely indication for spinal diagnostic imaging in the absence
13
of ICD-9 diagnosis codes, it is difficult to compare and interpret findings from studies that did
14
not adequately describe the population of interest. For example, a national sample of Medicare
15
fee-for-service (FFS) part B data reported that utilization of spinal CT varied from 0.3% to 2.4%
16
of all enrollees in different regions across the country, while another study reported that use of
17
spinal CT occurred in 1.1% of all Medicare patients, which was lower than the 1.4% and 3.0%
18
reported in studies that focused on those with LBP (10;21). Similarly, spinal MRI was used by
19
0.8% to 5.9% of all Medicare members, compared to 11.6% of those with LBP (10;21).
20
Measuring the reported utilization of health services without analyzing specific ICD-9 diagnosis
21
codes related to their use makes it challenging for stakeholders to interpret findings, identify
22
changes in use over time, and ascertain the appropriateness according to evidence-based CPGs,
23
particularly for relatively high-cost health services such as MRI and/or CT for LBP (35).
AC C
EP
TE D
M AN U
SC
7
13
ACCEPTED MANUSCRIPT
Although all of the studies included in this review seemed intent on measuring utilization of
2
diagnostic imaging for nonspecific LBP (i.e. excluding LBP from serious spinal pathology and
3
specific causes), wide variation was noted in the 76 different ICD-9 diagnosis codes used to
4
identify this clinical construct. For instance, one study included only two ICD-9 codes (724.2
5
lumbago and 724.5 unspecified backache), while three studies cited a list of nearly 70 diagnoses
6
to identify administrative claims related to mechanical LBP (11;29-31). However, it was unclear
7
exactly which ICD-9 codes had been used in some studies since ranges (e.g. 737.10-737.30)
8
were reported rather than listing each diagnosis to minimize confusion (33). Some of the studies
9
that cited the list of ICD-9 diagnosis codes proposed by Cherkin et al. listed 53 codes, while
SC
M AN U
10
RI PT
1
others listed 66, making it unclear if the original list had been interpreted correctly (31).
11
Other differences noted in the studies reviewed that may have influenced utilization included
13
patient eligibility criteria, duration of observation, CPT codes, and other requirements. For
14
example, one study included only patients seen by PCPs participating in a community-based
15
survey, while another focused on comparing utilization for patients seen by PCPs or orthopedic
16
surgeons (13;20). One Medicare study reported utilization of MRI within 6 months of the index
17
visit for LBP, while another Medicare study measured utilization over a 36-month period,
18
making it difficult to compare findings when duration varied 6-fold (20). The studies measuring
19
utilization of x-rays for LBP differed in their inclusion of cervical, full-spine, and sacral films,
20
which may have impacted their results (11;13;32). Studies also differed in the length of time
21
required without any claims for LBP to identify the index visit, which varied from 6 to 12
22
months in three studies and was not specified in four studies.
AC C
EP
TE D
12
14
ACCEPTED MANUSCRIPT
Some of the different methods used to identify claims related to spine pain in administrative
2
databases were recently examined by researchers at the Department of Veterans Affairs (VA)
3
(36). The authors found five previous studies with ICD-9 diagnosis codes related to different
4
types of LBP and neck pain (including the one by Cherkin et al.), as well as two classification
5
systems from the Agency for Healthcare Research and Quality (AHRQ) Healthcare Cost and
6
Utilization Project (HCUP). Data from the VA were then used to compare the number of patients
7
identified as having spine pain in the period 2002-2009 using different methods, as well as a
8
more comprehensive approach combining all diagnosis codes. Although considerable overlap
9
was found in these methods, discrepancies in their scope affected the number of patients
M AN U
SC
RI PT
1
identified. Authors recommended that future studies on utilization of health services for LBP
11
should specify clear objectives, ICD-9 diagnosis codes, exclusions (e.g. co-morbidities, serious
12
pathology), and other methods to facilitate comparison and interpretation. Enhanced
13
standardization for studies related to utilization of health services for LBP could potentially build
14
on previous reporting guidelines for retrospective database studies (37).
TE D
10
15
Standardizing the methods used in studies focused on measuring utilization of health services for
17
LBP may help mitigate some of the limitations associated with administrative data, but their
18
accuracy nevertheless depends on the information submitted by clinicians and administrators
19
(32). For example, spinal x-rays taken for symptoms of LBP in a patients with a history of
20
prostate cancer could be coded with the primary ICD-9 diagnosis code “724.2 lumbago”, which
21
would have been captured in many of the studies reviewed, while those coded “233.3 carcinoma
22
in situ of prostate” would not have been included.
AC C
EP
16
23
15
ACCEPTED MANUSCRIPT
Retrospective studies of claims data also contain sparse information that may influence
2
utilization and appropriateness, such as duration of LBP, symptom severity, functional
3
limitations, co-morbidities, severity and distribution of neurologic findings, clinical presentation,
4
patient preferences, beliefs, and expectations, and health plan management (13;25;30). For
5
example, utilization of x-rays was lower in patients with (22.6%) and without neurologic
6
findings (32.8%) (31). This finding may be viewed positively and suggests that clinicians were
7
aware that x-rays are unable to identify the source of neurologic compression responsible for the
8
patient’s clinical presentation, and therefore did not order them (31). Conversely, this may also
9
suggest that utilization of x-rays was too high for LBP without neurologic findings. Similarly, a
M AN U
SC
RI PT
1
lower observed rate of MRI utilization in those with nonspecific LBP (8.5%) than any type of
11
LBP (18.0%) may suggest that clinicians were aware of recommendations from evidence-based
12
CPGs against advanced diagnostic imaging for nonspecific LBP (32). However, an audit of
13
clinical records previously reported that 26% of x-rays and 66% of MRI and/or CT ordered for
14
acute LBP were deemed inappropriate according to criteria outlined in evidence-based CPGs,
15
making this hypothesis unlikely (38).
TE D
10
EP
16
Numerous factors may influence the use of diagnostic imaging for LBP, including patient
18
expectations, beliefs, and preferences. For example, some patients may believe that diagnostic
19
imaging is necessary to provide a diagnosis and validate their symptoms, and will perceive its
20
absence as an indication of poor quality care (39). Clinicians may therefore feel pressure to order
21
diagnostic imaging to ensure patient satisfaction (40). One study that examined indirect financial
22
incentives available to clinicians based on patient satisfaction reported that utilization of x-rays
23
for LBP was higher when those incentives were present (13). The role of other financial
AC C
17
16
ACCEPTED MANUSCRIPT
incentives was also explored in another study, which compared utilization of MRI for LBP in
2
physicians with or without an ownership interest in imaging equipment and reported that use was
3
nearly twice when this potential financial incentive was present (20). The role of these financial
4
and non-clinical factors should be explored further to understand how they may influence
5
utilization of diagnostic imaging for LBP.
RI PT
1
6
Another aspect that may sway clinicians to order diagnostic imaging is fear of litigation if they
8
fail to identify rare but serious spinal pathology related to LBP, such as spinal tumor, infection,
9
fracture, or cauda equina syndrome (41). Such fear may lead to the practice of defensive
M AN U
SC
7
medicine whereby clinicians err on the side of overusing diagnostic imaging, despite its high
11
costs and potential harms, to mitigate their legal liability (39). A study that recently examined
12
use of diagnostic imaging among orthopedists in Pennsylvania attributed 19% of utilization and
13
35% of costs to defensive medicine (42). The most common imaging modality ordered for
14
defensive reasons was MRI, for which 39% of orders were classified as defensive. Orthopedists
15
who had been sued for medical malpractice in the past five years had a higher proportion of
16
diagnostic imaging for defensive purposes (25%) than those who had not been sued (15%).
17
Although malpractice tort reform has been proposed as a potential solution to address this
18
problem, its impact on the utilization of diagnostic imaging for LBP remains unknown (43).
EP
AC C
19
TE D
10
20
Despite the limitations of claims data, they are often used to develop programs aimed at
21
improving the quality of health care, or at reducing costs associated with unnecessary utilization.
22
For example, the NCQA credentialed clinicians on a voluntary basis for its Back Pain
23
Recognition Program (BPRP), and one of the 13 clinical measures was related to diagnostic 17
ACCEPTED MANUSCRIPT
imaging, which specified that a maximum 50% of patients with LBP should receive diagnostic
2
imaging within 6 weeks of onset in the absence of red flags for serious spinal pathology (44).
3
Although the validity of this 50% threshold is difficult to evaluate, it was the only numerical
4
benchmark uncovered on this topic. Given that the NCQA recently announced it was retiring its
5
BPRP, it may be beneficial for other researchers or organizations to suggest an alternative
6
standard and method of measurement that could be used in future studies on this topic to
7
facilitate interpretation and comparison of findings in different populations and settings (45).
SC
RI PT
1
8
Other methods proposed to address potential overuse of health services include those targeting
M AN U
9
patient demand, such as decision aids, shared decision making, and public education campaigns.
11
For instance, the American Board of Internal Medicine (ABIM) Foundation recently launched its
12
Choosing Wisely® campaign to help patients choose health care that is supported by scientific
13
evidence and clinically necessary (46). This campaign is supported by various professional
14
societies, including the American Academy of Family Physicians (AAFP) and the American
15
College of Physicians (ACP). The use of diagnostic imaging for LBP without red flags was the
16
first item on the list from the AAFP, while diagnostic imaging for non-specific LBP was the
17
second item on the ACP list (46). Future studies should measure the impact of this campaign on
18
patient demand and subsequent utilization of diagnostic imaging for LBP.
EP
AC C
19
TE D
10
20
Clinician behavior can also be targeted through informal or continuing medication education
21
(CME) to share findings from recent CPGs, given that audits of clinical records reported that
22
diagnostic imaging guidelines were followed in only 2% of patients with chronic LBP (47). A
23
study previously reported that education of PCPs about CPGs for the assessment and 18
ACCEPTED MANUSCRIPT
management of LBP through CME presentations reduced utilization of MRI by 28% and CT by
2
81% without decreasing patient satisfaction with care (48). However, these findings were
3
obtained in a group practice HMO shortly after the educational intervention, and may not be
4
applicable in other settings. Education can also be enhanced through informal decision aids or
5
more formal clinical decision support (CDS) systems that provide feedback to clinicians on the
6
perceived appropriateness of the health service being ordered at the point of entry. For example,
7
a decision instrument based on five clinical criteria had a 99.0% sensitivity to detect cervical
8
spine injury later confirmed by radiographs for patients following blunt trauma; similar
9
instruments could likely be developed for LBP (49). Another study examined the effects of a
10
mandatory CDS prior to obtaining authorization for lumbar MRI in a large multidisciplinary
11
health network (12). Utilization of lumbar MRI for patients with LBP decreased from 12.7% in
12
the year before CDS was implemented to 9.3% the following year, a net reduction of 27%.
M AN U
SC
RI PT
1
TE D
13
The availability of other health services, including chiropractic and complementary and
15
alternative medical therapies such as massage or acupuncture, has also been found to impact
16
utilization of diagnostic imaging for LBP. One study compared utilization of spinal MRI and x-
17
rays among 700,000 health plan members with chiropractic benefits and 1 million members
18
without this coverage through their managed care plan (50). Utilization of MRI for patients with
19
LBP was considerably lower among patients with (4.3%) than without (6.9%) chiropractic
20
coverage; similar findings were also reported for x-rays (1.8% vs. 2.3%). Authors postulated that
21
chiropractic care may have been substituted for other health services, resulting in decreased use
22
of diagnostic imaging and other health services, though this hypothesis could not be confirmed.
AC C
EP
14
23
19
ACCEPTED MANUSCRIPT
Study limitations
2
This review has a number of limitations that should be acknowledged, including the limited
3
search strategy and eligibility criteria that limited the studies reviewed to those published in
4
English since the year 2000 based on third-party claims data for payers in the US that specified
5
the ICD-9 diagnosis codes used in their analyses. This narrow focus may have excluded
6
potentially relevant studies that may nevertheless have reported interesting findings. In addition,
7
all of the studies included in this review were based on claims data with imperfect accuracy,
8
incomplete clinical information, and may be influenced by other factors (e.g. administrative,
9
financial, or other requirements). Inferences about the appropriateness, clinical necessity, or
M AN U
SC
RI PT
1
perceived overuse of health services based solely on claims data are therefore speculative. It is
11
also difficult for third-party payers to discuss the topic of health care utilization without raising
12
the possibility that financial conflicts of interest are influencing their opinions. Nevertheless, this
13
review reports on a topic that appears to be of interest to many in the spine community, and
14
whose importance may continue to grow.
TE D
10
15
CONCLUSIONS
17
The reported utilization of various diagnostic imaging modalities for LBP, including x-rays, CT,
18
and MRI, varied across the studies reviewed. However, differences in the methodology used by
19
these studies impeded interpretation and meaningful comparisons, making it difficult to
20
determine if previously reported utilization of diagnostic imaging for LBP is appropriate.
21
Standardizing the methods for performing and reporting analyses of claims data related to
22
utilization could facilitate efforts by third-party payers, health care providers, and researchers to
23
identify and address the perceived overuse of diagnostic imaging for LBP in future studies.
AC C
EP
16
20
ACCEPTED MANUSCRIPT
1
REFERENCES
2
1.
study of prevalence and care-seeking. Spine (Phila Pa 1976). 1996;21:339-44. 2.
national surveys, 2002. Spine (Phila Pa 1976). 2006;31:2724-7. 3.
of chronic low back pain. Spine J. 2008;8:1-7.
7
4.
back and neck problems. JAMA. 2008;299:656-64.
9 10
5.
Carey TS, Freburger JK, Holmes GM, et al. A long way to go: practice patterns and evidence in chronic low back pain care. Spine (Phila Pa 1976). 2009;34:718-24.
11 12
Martin BI, Deyo RA, Mirza SK, et al. Expenditures and health status among adults with
M AN U
8
Haldeman S, Dagenais S. A supermarket approach to the evidence-informed management
SC
5 6
Deyo RA, Mirza SK, Martin BI. Back pain prevalence and visit rates: estimates from U.S.
RI PT
3 4
Carey TS, Evans AT, Hadler NM, et al. Acute severe low back pain. A population-based
6.
Martin BI, Turner JA, Mirza SK, Lee MJ, Comstock BA, Deyo RA. Trends in health care expenditures, utilization, and health status among US adults with spine problems, 1997-
14
2006. Spine (Phila Pa 1976). 2009;34:2077-84. 7.
off? J Am Board Fam Pract. 2009;22:62-8.
16 17
Deyo RA, Mirza SK, Turner JA, Martin BI. Overtreating chronic back pain: time to back
8.
EP
15
TE D
13
Deyo RA, Mirza SK, Martin BI, Kreuter W, Goodman DC, Jarvik JG. Trends, major medical complications, and charges associated with surgery for lumbar spinal stenosis in
19
older adults. JAMA. 2010;303:1259-65.
20
9.
AC C
18
United States Government Accountability Office. Medicare Part B Imaging Services:
21
Rapid spending growth and shift to physician offices indicate need for CMS to consider
22
additional management practices. Washington: U.S. Government Accountability Office.
23
2008; Report No.: GAO-08-452. 21
ACCEPTED MANUSCRIPT
1
10.
Rao VM, Parker L, Levin DC, Sunshine J, Bushee G. Use trends and geographic variation
2
in neuroimaging: nationwide medicare data for 1993 and 1998. Am J Neuroradiol.
3
2001;22:1643-9.
healthcare resources wisely? Pain Med. 2006;7:143-50.
5
12.
controlling inappropriate imaging. J Am Coll Radiol. 2011;8:19-25.
7 8
13.
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;169:972-81.
9 10
Blackmore CC, Mecklenburg RS, Kaplan GS. Effectiveness of clinical decision support in
SC
6
Weiner DK, Kim YS, Bonino P, Wang T. Low back pain in older adults: are we utilizing
RI PT
11.
14.
M AN U
4
Swedlow A, Johnson G, Smithline N, Milstein A. Increased costs and rates of use in the
11
California workers' compensation system as a result of self-referral by physicians. N Engl J
12
Med. 1992;327:1502-6.
Emergency Department database analysis. J Emerg Med. 2004;26:37-45.
14
16.
ill: a review of the literature. Health Policy. 1997;42:101-15.
16 17
17.
Baras JD, Baker LC. Magnetic resonance imaging and low back pain care for Medicare patients. Health Aff. 2009;28:w1133-40.
18 19
de Boer AG, Wijker W, de Haes HC. Predictors of health care utilization in the chronically
EP
15
Isaacs DM, Marinac J, Sun C. Radiograph use in low back pain: a United States
TE D
15.
18.
AC C
13
Freeborn DK, Shye D, Mullooly JP, Eraker S, Romeo J. Primary care physicians' use of
20
lumbar spine imaging tests: effects of guidelines and practice pattern feedback. J Gen
21
Intern Med. 1997;12:619-25.
22
ACCEPTED MANUSCRIPT
1
19.
French SD, Green S, Buchbinder R, Barnes H. Interventions for improving the appropriate
2
use of imaging in people with musculoskeletal conditions. Cochrane Database of
3
Systematic Reviews. 2010;(1):CD006094. 20.
Shreibati JB, Baker LC. The relationship between low back magnetic resonance imaging,
RI PT
4 5
surgery, and spending: impact of physician self-referral status. Health Serv Res.
6
2011;46:1362-81.
surgery. Spine (Phila Pa 1976). 2003;28:616-20. 22.
a large integrated health system. Health Aff (Millwood). 2008;27:1491-502.
10 11
23.
National Committee for Quality Assurance. Back pain recognition program requirements. Washington: National Committee for Quality Assurance; 2007.
12 13
Smith-Bindman R, Miglioretti DL, Larson EB. Rising use of diagnostic medical imaging in
M AN U
8 9
Lurie JD, Birkmeyer NJ, Weinstein JN. Rates of advanced spinal imaging and spine
SC
21.
24.
Rosenthal MB, Abrams MK, Bitton A. Recommended core measures for evaluating the
TE D
7
14
patient-centered medical home: Cost, utilization, and clinical quality. The Commonwealth
15
Fund. 2012. Report No.: 1601.
2001;10:389-92.
17 18
Strom BL. Data validity issues in using claims data. Pharmacoepidemiol Drug Saf.
26.
EP
25.
van Tulder M, Furlan A, Bombardier C, Bouter L; Editorial Board of the Cochrane
AC C
16
19
Collaboration Back Review Group. Updated method guidelines for systematic reviews in
20
the cochrane collaboration back review group. Spine (Phila Pa 1976). 2003;28:1290-9.
21
27.
Santaguida PL, Gross A, Busse J, et al. Evidence report on complementary and alternative
22
medicine in back pain utilization report. Rockville: Agency for Healthcare Research and
23
Quality. 2009. Report No.: 09-E006. 23
ACCEPTED MANUSCRIPT
1
28.
Downs SH, Black N. The feasibility of creating a checklist for the assessment of the
2
methodological quality both of randomised and non-randomised studies of health care
3
interventions. J Epidemiol Community Health. 1998;52:377-84. 29.
Rhee Y, Taitel MS, Walker DR, Lau DT. Narcotic drug use among patients with lower
RI PT
4 5
back pain in employer health plans: a retrospective analysis of risk factors and health care
6
services. Clin Ther. 2007;29:2603-12. 30.
Ritzwoller DP, Crounse L, Shetterly S, Rublee D. The association of comorbidities,
SC
7
utilization and costs for patients identified with low back pain. BMC Musculoskelet
9
Disord. 2006;7:72.
10
31.
M AN U
8
Vogt MT, Kwoh CK, Cope DK, Osial TA, Culyba M, Starz TW. Analgesic usage for low
11
back pain: impact on health care costs and service use. Spine (Phila Pa 1976).
12
2005;30:1075-81. 32.
Ivanova JI, Birnbaum HG, Schiller M, Kantor E, Johnstone BM, Swindle RW. Real-world
TE D
13 14
practice patterns, health-care utilization, and costs in patients with low back pain: the long
15
road to guideline-concordant care. Spine J. 2011;11:622-32. 33.
Cherkin DC, Deyo RA, Volinn E, Loeser JD. Use of the International Classification of
EP
16
Diseases (ICD-9-CM) to identify hospitalizations for mechanical low back problems in
18
administrative databases. Spine (Phila Pa 1976). 1992;17:817-25.
19
34.
National Committee for Quality Assurance. HEDIS 2006 technical specifications. Washington: National Committee for Quality Assurance; 2005.
20 21
AC C
17
35.
Dagenais S, Tricco AC, Haldeman S. Synthesis of recommendations for the assessment and
22
management of low back pain from recent clinical practice guidelines. Spine J.
23
2010;10:514-29. 24
ACCEPTED MANUSCRIPT
1
36.
Sinnott PL, Siroka AM, Shane AC, Trafton JA, Wagner TH. Identifying neck and back
2
pain in administrative data: defining the right cohort. Spine (Phila Pa 1976) 2012;37:860-
3
74. 37.
report of the ISPOR Task Force on Retrospective Databases. Value Health. 2003;6:90-7.
5 6
Motheral B, Brooks J, Clark MA, et al. A checklist for retrospective database studies -
RI PT
4
38.
Schroth WS, Schectman JM, Elinsky EG, Panagides JC. Utilization of medical services for the treatment of acute low back pain: conformance with clinical guidelines. J Gen Intern
8
Med. 1992;7:486-91. 39.
Chou R, Qaseem A, Owens DK, Shekelle P. Diagnostic imaging for low back pain: advice
M AN U
9
SC
7
10
for high-value health care from the American College of Physicians. Ann Intern Med.
11
2011;154:181-9. 40.
and overuse. Arch Intern Med. 2009;169:921-3.
13 14
Deyo RA. Imaging idolatry: the uneasy intersection of patient satisfaction, quality of care,
41.
TE D
12
Anderson PA, Sasso RC, Riew KD. Comparison of adverse events between the Bryan artificial cervical disc and anterior cervical arthrodesis. Spine (Phila Pa 1976).
16
2008;33:1305-12. 42.
prospective practice audit in Pennsylvania. J Bone Joint Surg Am. 2012;94:e18.
18 19
43.
Epstein NE. It is easier to confuse a jury than convince a judge: the crisis in medical malpractice. Spine (Phila Pa 1976). 2002;27:2425-30.
20 21
Miller RA, Sampson NR, Flynn JM. The prevalence of defensive orthopaedic imaging: a
AC C
17
EP
15
44.
Hilde G, Hagen KB, Jamtvedt G, Winnem M. WITHDRAWN: Advice to stay active as a
22
single treatment for low-back pain and sciatica. Cochrane Database of Systematic Reviews
23
2006;(2):CD003632. 25
ACCEPTED MANUSCRIPT
1
45.
National Committee for Quality Assurance. NCQA to retire back pain recognition program
2
[Internet]. Washington: National Committee for Quality Assurance; 2011 [cited 2012 June
3
1]. Available from: http://www.ncqa.org/tabid/137/Default.aspx 46.
American Academy of Family Physicians. Choosing wisley: five things physicians and
RI PT
4
patients should question [Internet]. Philadelphia: American Board of Internal Medicine
6
(ABIM) Foundation; 2012 [cited 1 June, 2012]. Available from:
7
http://choosingwisely.org/wp-content/uploads/2012/04/Five-Things.pdf 47.
retrospective study of 132 patients in the Haute-Vienne district of France. Joint Bone
9
Spine. 2002;69:589-96.
10 11
Hourcade S, Treves R. Computed tomography in low back pain and sciatica. A
M AN U
8
SC
5
48.
Klein BJ, Radecki RT, Foris MP, Feil EI, Hickey ME. Bridging the gap between science and practice in managing low back pain. A comprehensive spine care system in a health
13
maintenance organization setting. Spine (Phila Pa 1976). 2000;25:738-40.
14
49.
TE D
12
Hoffman JR, Mower WR, Wolfson AB, Todd KH, Zucker MI. Validity of a set of clinical criteria to rule out injury to the cervical spine in patients with blunt trauma. National
16
Emergency X-Radiography Utilization Study Group. N Engl J Med. 2000;343:94-9.
17
50.
EP
15
Legorreta AP, Metz RD, Nelson CF, Ray S, Chernicoff HO, Dinubile NA. Comparative analysis of individuals with and without chiropractic coverage: patient characteristics,
19
utilization, and costs. Arch Intern Med. 2004;164:1985-92.
AC C
18
26
ACCEPTED MANUSCRIPT
1
FIGURE 1. Flow diagram of studies in this review
AC C
EP
TE D
M AN U
SC
RI PT
2
27
ACCEPTED MANUSCRIPT
Table 1. Study characteristics LBP sample Rate
Data Utilization
Type of imaging
Inclusion
Illinois 165,569 18-64 Colorado >410,000 ≥18
13,760 8.3%
2002-2004 19 months
MRI
1. LBP of mechanical causes
1. infection 2. accidental causes
16,567 4.0%
1996-2001 24 months
x-rays MRI/CT
1. major existing conditions (e.g. neoplasm, infection, pregnancy, trauma)
x-rays MRI CT MRI/CT x-rays MRI CT
1. LBP diagnosis in 1997 or 1998 2. first contact in ambulatory or hospital setting 3. no LBP in previous 12 months 4. enrolment for 24 months after index LBP 1. LBP due to mechanical causes
1. LBP diagnosis during 2004-2006 2. coverage >6 months before index LBP claim 3. coverage >12 months after index LBP claim
1. other LBP diagnosis during observation period
Exclusion
RI PT
Commercial (29) Employersponsored (30) 1 Staff model HMO
Location Sample size Age
SC
Reference Plan type
Pennsylvania 255,958 ≥18
17,148 6.7%
2001 12 months
(32) NR
National ≥8,000,000 18-64
211,551 2.6%
2004-2006 12-36 months
Pennsylvania NR ≥65
34,408 2 NR
2000-2002 12 months
x-rays MRI/CT
1. nonspecific LBP
1. lumbar spinal stenosis
National 496,529 ≥65
35,039 7.1%
2000-2002 36 months
x-rays MRI/CT
1. uncomplicated acute LBP 2. seen by PCP 4 participating in Community Tracking Study Physician Surveys 3. enrolled for >6 months after LBP
National NR ≥65
740,467 6 NR
1. LBP in previous 6 months 2. care in >1 state 3. hospice or nursing homes 4. neurologic deficit 5. end-stage renal disease 6. trauma, cancer, infection 7. index visit by radiologist 1. LBP in previous 12 months 2. Medicare eligible