VARIATIONS IN PATTERNS OF UTILIZATION AND CHARGES FOR THE CARE OF LOW BACK PAIN IN NORTH CAROLINA, 2000 TO 2009: A STATEWIDE CLAIMS DATA ANALYSIS

Transcription

1 VARIATIONS IN PATTERNS OF UTILIZATION AND CHARGES FOR THE CARE OF LOW BACK PAIN IN NORTH CAROLINA, 2000 TO 2009: A STATEWIDE CLAIMS DATA ANALYSIS Eric L. Hurwitz, DC, PhD, a Dongmei Li, PhD, b Jenni Guillen, MS, c Michael J. Schneider, DC, PhD, d Joel M. Stevans, DC, e Reed B. Phillips, DC, PhD, f Shawn P. Phelan, DC, g Eugene A. Lewis, DC, MPH, h Richard C. Armstrong, MS, DC, i and Maria Vassilaki, MD, MPH, PhD j ABSTRACT Objectives: The purpose of the study was to compare utilization and charges generated by medical doctors (MD), doctors of chiropractic (DC) and physical therapists (PT) by patterns of care for the treatment of low back pain in North Carolina. Methods: This was an analysis of low-back-pain-related closed claim data from the North Carolina State Health Plan for Teachers and State Employees from 2000 to Data were extracted from Blue Cross Blue Shield of North Carolina for the North Carolina State Health Plan using International Classification of Diseases, 9th Revision diagnostic codes for uncomplicated low back pain (ULBP) and complicated low back pain (CLBP). Results: Care patterns with single-provider types and no referrals incurred the least charges on average for both ULBP and CLBP. When care did not include referral providers or services, for ULBP, MD and DC care was on average $465 less than MD and PT care. For CLBP, MD and DC care averaged $965 more than MD and PT care. However, when care involved referral providers or services, MD and DC care was on average $1600 less when compared to MD and PT care for ULBP and $1885 less for CLBP. Risk-adjusted charges (available ) for patients in the middle quintile of risk were significantly less for DC care patterns. Conclusions: Chiropractic care alone or DC with MD care incurred appreciably fewer charges for ULBP than MD care with or without PT care. This finding was reversed for CLBP. Adjusted charges for both ULBP and CLBP patients were significantly lower for DC patients. (J Manipulative Physiol Ther 2016;39: ) Key Indexing Terms: Low Back Pain; Chiropractic; Medical Care; Health Services; Utilization; Healthcare Costs T he numbers of reported cases of neck and back problems have increased dramatically. Martin et al 1 reported 14.8 million cases in 1997 and 21.9 million in 2006, a 67.6% increase in 6 years. The Centers for Disease Control and Prevention reported that back or spine problems are the second most common cause of disability in the United States, and noted a 7.7% increase in a Professor, Office of Public Health Studies, University of Hawai`i at Mānoa, Honolulu, HI. b Associate Professor, Clinical and Translational Science Institute, University of Rochester School of Medicine and Dentistry, Rochester, NY. c Graduate Research Associate, Office of Public Health Studies, University of Hawai`i at Mānoa, Honolulu, HI. d Associate Professor, School of Health and Rehabilitation Sciences, University of Pittsburgh, Pittsburgh, PA. e Assistant Professor, School of Health and Rehabilitation Sciences, University of Pittsburgh, Pittsburgh, PA. f Doctor of Chiropractic, Retired, Pocatello, ID. g Doctor of Chiropractic, Private Practice of Chiropractic, Wake Forest, NC. disability cases due to an aging population. 2,3 Overall, 1% to 2% of adults in the United States are disabled due to back pain. 4 With spine-related disability increasing, the implications on healthcare policy, spending, and identification of cost effective treatment strategies are enormous. The rise in prevalence of back pain and increased utilization of healthcare services are driving the costs of the h Doctor of Chiropractic, Private Practice of Chiropractic, Greensboro, NC. i Doctor of Chiropractic, Private Practice of Chiropractic, Cary, NC. j Research Associate, Division of Epidemiology, Department of Health Sciences Research, Mayo Clinic, Rochester, MN. Submit requests for reprints to: Eric L. Hurwitz, DC, PhD, Professor, Office of Public Health Studies, Department of Public Health Sciences, University of Hawai`i, Mānoa, 1960 East-West Road, Biomedical Sciences D201, Honolulu, HI ( ehurwitz@hawaii.edu). Paper submitted January 23, 2015; in revised form November 1, 2015; accepted January 1, Copyright 2016 by National University of Health Sciences.

2 Journal of Manipulative and Physiological Therapeutics Volume 39, Number 4 Hurwitz et al 253 back pain epidemic upward. Martin et al 5 compared the medical costs of participants with and without spine problems from 1997 to 2005, adjusting for age and gender. Those with spine problems exhibited a 65% higher increase in medical expenditures from $4695 in 1997 to $6096 in 2005 per person. Martin et al also reported that the largest proportion of increasing per user medical expenditures for spine-related problems were for inpatient hospitalizations (37%), outpatient costs (18%), prescription drugs (139%), and emergency room visits (84%). 1 Patients with back pain are most often seen by medical doctors (MD), doctors of chiropractic (DC), physical therapists (PT) and medical specialists to which they are referred. 6 Annual expenditures for MD, DC, and PT care combined have been estimated to range from $84.1 billion to $624.8 billion for low back problems in the United States. 7 Increased costs can be attributed to inflation, an increase in the numbers of office visits to each of these providers, and increases in annual per user expenditures as a result of the mix of services (imaging, specialty interventions, etc). From 1999 to 2008, yearly average inflation-adjusted medical expenditures for patients with primary diagnoses of back or neck conditions rose from $487 to $950 (a 95% increase), mostly due to the steeply rising costs associated with medical specialists and physical therapy services. 8 Chiropractic care experienced a 57% increase in patient visits from 2000 to 2003 in the US, while the mean costs per patient and per chiropractic office visit have remained stable over time. 9 The aim of this study was to assess the utilization and cost of care patterns for low back pain among patients in the North Carolina State Health Plan (NCSHP) for Teachers and State Employees from 2000 to We compared the cost of care of these patterns of care: patients who utilized MDs and DCs alone, MD and DC care in combination with each other, MD or DC care in combination with PT, and/or with additional referred provider care. METHODS This study is a retrospective closed-claim analysis of the NCSHP. These data include claims generated annually by approximately covered beneficiaries (state employees, dependents, and retirees), between the years 2000 and Data were extracted from Blue Cross Blue Shield of North Carolina using an extraction model developed with clinical healthcare analysts from the NCSHP. Cohort Identification and Stratification The low back pain analytic cohort identified all professional and facility claims for a healthcare event with a primary low back pain diagnosis as identified by International Classification of Diseases, 9th Revision (ICD-9) diagnosis codes. The ICD-9 codes used to select the cohort were the most common codes used across all 3 professions (MD, DC, and PT). It was not the intent of this study to include every possible ICD-9 Table 1. Primary Diagnoses (ICD-9 codes) Defining Each Type of Low Back Pain ULBP CLBP Facet joint fixation (718.48) Lumbar spondylosis with myelopathy (721.42) Facet joint swelling (719.08) Degeneration of intervertebral disc (722.52) Lumbar spondylosis (721.3) Disorder of intervertebral disc with myelopathy (722.73) Lumbago (724.2) Lumbar stenosis (724.02) Facet syndrome (724.8) Sciatica (724.3) Muscle spasm (728.85) Neuritis or radiculitis (724.4) Spondylolisthesis (756.12) Compression of spinal nerve root (724.9) Sprain/strain (847.2) Numbness or tingling (782.0) CLBP, complicated low back pain; ULBP, uncomplicated low back pain code utilized by each of the 3 provider types or their specialist referral destinations. It was instead to include the most common codes used by all of the provider types. The codes used by DCs, Subluxation ICD-9 codes, were excluded for a number of reasons. Medical and PT offices rarely use subluxation codes when billing third party payers. These codes are only required when billing traditional Medicare. In these circumstances, Medicare is the primary payer and NCSHP would be secondary. All claims in which NCSHP was the secondary payer were excluded from the analysis. Secondary, tertiary, and quaternary codes were not used because substantial utilization unrelated to the treatment of low back pain came up in the initial extraction. This would have led to overestimation of low back pain charges in our cohort. Therefore, we chose to use the primary diagnosis to identify cases and subsequent claims. According to ICD-9 coding guidelines, the primary diagnosis listed on a claim form should reflect the principal reason for the patient s visit on that date of service. By only using the primary diagnosis to identify claims of interest, our analysis provides estimates that are more conservative by eliminating the scatter of cases where low back pain was only a secondary or tertiary complaint. The low back pain cohort was then stratified into 2 broad categories of low back pain: (1) uncomplicated low back pain (ULBP), and (2) complicated low back pain (CLBP). Table 1 shows the primary diagnoses (ICD-9 codes) used to distinguish between ULBP and CLBP. Our clinical rationale for this stratification was that patients with diagnoses included in the ULBP category were less likely to have radicular complaints than those in the CLBP category and would require fewer healthcare services. Although the reliability of using ICD-9 codes to distinguish between these categories could be argued, all provider patterns were evaluated relative to them under the same assignment. Claim Defined Each claim represents a unique clinical service as defined by an individual allowed Current Procedural

3 254 Hurwitz et al Journal of Manipulative and Physiological Therapeutics May 2016 Terminology code. Medicare and non-north Carolina residents were excluded. Each reporting (fiscal) year represents a benefit year starting in July and ending in June. patterns based on their use of each of these provider types: MD only, DC only, MD-DC, MD-PT, MD-referral, DC-referral, MD-DC-referral, and MD-PT-referral. Provider Categorization The providers were categorized into 4 basic types: (1) Doctor of Chiropractic (DC); (2) Medical/Osteopathic Doctor (MD); (3) Physical Therapist (PT); and (4) Referral providers. The MD group also included additional provider types: General Practice, Internal Medicine, Neurology, Neurosurgery, Obstetrics-Gynecology, Orthopedic Surgery, Osteopathy, Pediatrics, Physical Medicine, General Surgery, Family Practice, or Geriatric Medicine; Nurse Practitioner; Podiatry; Public Health; University/College Infirmary; Urgent Care; and VA/Military Hospital-Professional Staff. Referral providers were hospitals, surgical centers, emergency medicine, and other specialty referral services and providers. Of note is the inclusion of Podiatry in the MD group. This analysis is one in a series of 3 using the NCSHP database to evaluate the patterns of care and costs for 3 common neuromusculoskeletal conditions (low back pain, neck pain, and headache). Podiatrists were included in the MD category because they may use a primary diagnosis of a lumbar radiculopathy when seeing a patient for foot pain or numbness and therefore serve as a possible portal of entry for low back pain patients. Podiatrists would be far less likely to treat neck pain or headache patients. However, they were included in all 3 analyses in order to maintain consistent methodology across the 3 studies. In North Carolina, patients must be referred from a MD or a DC to receive PT care. This places the primary management decisions with the referring medical doctor or chiropractor, with physical therapists providing treatment under the direction of these MD and DC providers. Therefore, the PT category does not stand alone in this analysis. Provider Utilization Patterns Although much interest has developed in the efficacy of the medical home, where patients are encouraged to seek care from their primary care doctors first, instead of self-referring to specialists; not all patients choose this path. Many patients choose to seek care directly from various specialist and non-specialist healthcare professionals. Therefore, this model was intended to reflect how the system currently functions and primary care medical home was not selected as a pattern. Tracking of healthcare utilization and costs becomes very complicated once one begins to follow longitudinally patient treatment through different provider patterns, utilizing a multiple set of tests and interventions (eg, advanced imaging, x-ray, medication, surgery) delivered within heterogeneous settings (eg, office, in- and out-patient facilities). Consequently, patients were classified into care Care Endpoint Once ICD-9 codes were selected and the patterns of care identified, the endpoint for care became the next question. Ideally, this would have been done within the parameters of an episode. We selected not to perform an episodes of care analysis because key parameters such as the standard episode length and clean periods have not been established in the literature. This design decision is in agreement with the epidemiologic model of a long-term condition like LBP, which considers it as an interdependent longitudinal sequence, rather than as a series of separate unrelated events. 10 In addition, back pain may not completely resolve following the episode of care and individuals with persistent pain may use prescription medication, medical and other healthcare services only intermittently. 11 For this reason, we tracked the cost of care for ULBP and CLBP using the primary ICD-9 code through care according to the provider patterns described above, using a fiscal year as the endpoint for all diagnostic codes. Each fiscal year was analyzed separately, so that the date range for any diagnosis within each analysis could be no longer than 1 year. For a diagnosis that spanned more than 1 fiscal year, or that incurred charges over several months in different years, the diagnosis was included in both but would be reported in separate analyses reflecting those years. Tracking of Costs If a patient presented with the primary diagnosis of disorder of intervertebral disc with myelopathy (722.73), in a primary care doctor s office, the aggregate cost of the care under that code would accumulate under the MD-only pattern of care. This would include all diagnostic and treatment services as long as the patient did not receive DC, PT, or referral care services. If that same patient sought care in a DC s office, or had PT, the aggregate costs would accumulate under the MD-DC or MD-PT pattern of care, respectively. It is possible that the patient s diagnosis changed over time. If the diagnosis changed, and the new diagnosis fell within the same clinical category (ULBP or CLBP) the patient would remain in the initially assigned ULBP or CLBP group. However, if the new diagnosis was from the other category (eg, former diagnosis was ULBP and the new diagnosis is CLBP), then the patient would move into the CLBP group and the aggregate costs would accrue to this category. This is how a primary ICD-9 code would be used to track utilization and costs across provider patterns. Cost Reference Points Reference points of average (mean and median) numbers of claims, charges per claim and total allowed charges per

4 Journal of Manipulative and Physiological Therapeutics Volume 39, Number 4 Hurwitz et al 255 patient were used to analyze costs. The total allowed charges were calculated by summing the payments made by both the patient and insurer for a medical service. This dollar amount represents the total amount paid to the provider. Costs were not adjusted for inflation. It was not the purpose of this study to estimate absolute differences in costs over time from , but instead to compare within-year relative differences in costs between patterns of care. For each fiscal year, pharmaceutical claim data were linked with medical claim data based on each patient s unique identity. The major claim types were professional office visit (eg, MD, DC), advanced imaging (magnetic resonance imaging, computerized tomography or similar services), radiology (diagnostic X-ray, arthrography, or similar services), physical therapy (provider specialty is physical therapy or a physical therapy facility with service types belonging to physical therapy), and surgical (services provided by neurosurgeon or orthopedic or general surgeon). The University of Hawaii Human Studies Program approved this study as exempt from federal regulations pertaining to the protection of human research participants, as documented in the Code of Federal Regulations at 45CFR (b)(Exempt Category 4). Statistics We generated frequency distributions of claims and patients for each pattern of care, by year. The number of claims in each provider group for each care pattern was identified by cross tabulation of care pattern and provider type. Within each of the 5 service claim types, care pattern and provider type were cross-tabulated to identify the number of claims in each provider group for each care pattern. The total allowed and per claim medical, pharmaceutical, and combined expenses were then summarized for each patient. The patient-based and claim-based mean and median of medical, pharmaceutical, and combined medical and pharmaceutical expenses were also generated for each care pattern. Scores reflecting risk of expected healthcare cost and utilization relative to that of the overall population were available in years 2006 to Risk scores take into account patient-specific factors that may affect utilization and charges, including age, gender, primary diagnosis, comorbidities, and use of prescription drugs. Essentially, risk scores help define the difficulty of treating a particular patient. Within each type of low back pain (ULBP, CLBP), for patients in the middle quintile of risk, linear regression models were used to fit log10-transformed total allowed charges per patient to examine pairwise differences across the 8 most relevant patterns of care, adjusting for risk score as a continuous variable in the models. Patterns that contained very little utilization were excluded. The log10-based transformation was used because the transformed costs are closer to normal distributions than natural log-transformed cost data. Examining patients in the middle quintile (40th-60th percentiles) removes those at lowest and highest risk and offers apples-to-apples between-pattern comparisons of the average or more typical patient. Linear orthogonal contrasts (ratios) were used to compare differences in charges between DC- and MD-related care patterns. Residual diagnostics were conducted and the normality assumptions of residuals were satisfied. The homoscedasticity assumption was also satisfied for the log transformed cost data. P values, standard errors, and 95% confidence intervals were computed. SAS 9.2 (SAS Institute, Cary, NC) 12 was used for data management and statistical analyses. RESULTS Utilization Uncomplicated Low Back Pain (ULBP). The inclusion criteria were met for ULBP claims, and for CLBP claims. For ULBP, patients increased to patients from 2000 to 2009 (up 136%) meeting the inclusion criteria; for CLBP, 5097 patients increased to patients from 2000 to 2009 (up 177%) meeting the inclusion criteria. The number of ULBP patients in all care patterns increased over the 10-year period; however, gains were greatest among care patterns involving MDs, PTs, and referrals, and lowest among patterns with DCs. Numbers of patients in care patterns with MDs increased from 7375 in 2000 to in 2009 (up 185%), whereas numbers of patients in care patterns with DCs increased from 3390 in 2000 to 5055 in 2009 (up 49%). Complicated Low Back Pain (CLBP). The number of patients with at least 1 claim for CLBP increased from 5097 in 2000 to in 2009 (up 177%). As with ULBP, the number of CLBP patients in all care patterns increased over the 10-year period; however, gains were greatest among care patterns involving MDs, PTs, and referrals, and smallest among DC-care patterns. The numbers of patients in care patterns with MDs increased from 2798 in 2000 to 9122 in 2009 (up 226%), whereas the numbers of patients in care patterns with DCs increased from 1427 in 2000 to 2540 in 2009 (up 78%). Overall Charges Uncomplicated Low Back Pain. Total allowed charges (ie, the sum of patient and payer financial responsibility), for ULBP increased from $ in 2000 to $ in Total charges increased almost threefold (2.95) from 2000 to 2006 before declining by 1 million dollars in 2007, then escalating by 17% from 2007 to 2009 (Fig 1). Average total charges per patient for all care patterns combined increased from $1495 in 2000 to $2396 in 2006 (60% increase), and declined to $2096 in 2007 (12.5% decrease) before climbing up to $2220 in 2008 and $2169 in

5 256 Hurwitz et al Journal of Manipulative and Physiological Therapeutics May Millions of Dollars Complicated LBP Uncomplicated LBP Fig 1. Sum of total allowed charges for all care patterns combined for uncomplicated and complicated low back pain (LBP), by year: North Carolina State Health Plan for Teachers and State Employees, 2000 to Over the decade, average total allowed charges for ULBP increased by 45% (Fig 2). Complicated Low Back Pain. Total allowed charges for CLBP increased from $ in 2000 to $ in There was a threefold increase in total charges from 2000 to 2006 ($ ) and a slight decline to $ in Total charges rose sharply in the last 2 years, however, to $ in 2008 and $ in 2009 (31% increase from 2007 to 2009) (Fig 1). Average total charges per patient for all care patterns combined increased from $3128 in 2000 to $4465 in 2006 (43% increase), dropped to $3768 in 2007 and rose again in 2008 and Over the decade, average total allowed charges for CLBP increased by 38% (Fig 2). Pattern-Specific Charges This section is to report the cost analysis of patterns of provider care or paths through which patients may have been provided care through the system, as well as average (mean and median) numbers of claims, charges per claim and overall allowed charges per patient for these patterns. This section concludes with a presentation of risk-adjusted averages. Uncomplicated Low Back Pain. For all years, care patterns involving multiple types of providers resulted in appreciably greater average charges per patient than care patterns involving single providers. In general, care patterns with MDs and referral providers resulted in greater average charges per patient than other care patterns. MD-only, DC-only, and MD-DC care were consistently the least expensive patterns of care for ULBP (mean [median] total allowed charges per patient in 2009 of $978 [$182], $1165 [$277], and $1667 [$300], respectively). Medical care with physical therapy was more expensive than medical care with chiropractic whether or not care included referral providers. Without referral providers or services involved, medical care with physical therapy was on average more expensive than medical care with chiropractic. With referral providers, medical care with physical therapy ranged from an average $561 (in 2000) to $2508 (in 2008) more expensive than medical care with chiropractic (Table 2). Complicated Low Back Pain. As with ULBP, care patterns involving multiple types of providers resulted in greater average charges than care patterns involving single providers. Patterns with MDs resulted in generally greater average charges; and referrals tended to boost charges. DC-only, MD-only, and MD-PT care were consistently the least expensive patterns of care for CLBP (mean [median] total allowed charges per patient in 2009 of $1394 [$324], $1498 [$250], and $1888 [$335], respectively). Medical care with physical therapy was generally less expensive than medical care with chiropractic when care did not include referral providers (mean total allowed charges in 2009 of $1888 vs. $2642). However, with referral care, the combination of MD and DC care incurred fewer charges than the combination of MD and PT care. Without referral care, MD care with PT was on average $494 (in 2004) to $1567 (in 2006) less expensive than MD

6 Journal of Manipulative and Physiological Therapeutics Volume 39, Number 4 Hurwitz et al Dollars 2500 Uncomplicated LBP Complicated LBP Fig 2. Per-patient mean total allowed charges for all care patterns combined for uncomplicated and complicated low back pain (LBP), by year: North Carolina State Health Plan for Teachers and State Employees, 2000 to Table 2. Number of Patients and Mean (Median) Total Allowed Charges per Patient for Uncomplicated Low Back Pain, by Care Pattern and Year: North Carolina State Health Plan for Teachers and State Employees, Care Pattern DC-Only MD-Only MD-DC MD-PT Year No. Charges No. Charges No. Charges No. Charges (292) (134) (254) (569) (332) (148) (272) (631) (323) (160) (250) (616) (367) (167) (266) (685) (411) (182) (329) (650) (415) (186) (286) (661) (459) (192) (297) (665) (305) (189) (321) (531) (303) (184) (316) (559) (277) (182) (300) (534) DC-Referral MD-Referral MD-DC-Referral MD-PT-Referral Year No. Charges No. Charges No. Charges No. Charges (390) (448) (773) (1196) (436) (493) (756) (1298) (497) (525) (806) (1387) (511) (555) (777) (1289) (607) (580) (916) (1452) (525) (605) (954) (1554) (605) (654) (939) (1470) (595) (579) (769) (1321) (468) (602) (817) (1368) (439) (533) (779) (1282) DC, doctor of chiropractic; MD, medical doctor; PT, physical therapist care with DC care. With referral care, MD care with PT was on average $1270 (in 2009) to $3038 (in 2004) more expensive than MD care with DC care (Table 3). Risk-Adjusted Charges, 2006 to Risk scores take into account patient-specific factors that may affect utilization and charges, including age, gender, primary diagnosis,

7 258 Hurwitz et al Journal of Manipulative and Physiological Therapeutics May 2016 Table 3. Number of Patients and Mean (Median) Total Allowed Charges per Patient for Complicated Low Back Pain, by Care Pattern and Year: North Carolina State Health Plan for Teachers and State Employees, Care Pattern DC-only MD-only MD-DC MD-PT Year No. Charges No. Charges No. Charges No. Charges (369) (205) (490) (401) (442) (233) (466) (447) (419) (236) (350) (406) (478) (237) (384) (507) (535) (246) (479) (399) (536) (245) (535) (522) (633) (252) (586) (412) (409) (236) (414) (436) (397) (250) (477) (318) (324) (250) (510) (335) DC-Referral MD-Referral MD-DC-Referral MD-PT-Referral Year No. Charges No. Charges No. Charges No. Charges (459) (729) (979) (1225) (518) (831) (1047) (1158) (623) (970) (1061) (1420) (710) (883) (1012) (1271) (941) (975) (1311) (1328) (884) (1004) (1224) (1324) (781) (1046) (1241) (1384) (750) (866) (1165) (1230) (788) (880) (1196) (1139) (703) (906) (1158) (1381) DC, doctor of chiropractic; MD, medical doctor; PT, physical therapist comorbidities, and use of prescription drugs. Essentially, risk scores help define the difficulty of treating a particular patient. For ULBP, risk-adjusted mean charges were significantly greater for MD-only vs. DC-only care, MD-PT vs. MD-DC care, and MD-referral vs. DC-referral care. The one exception was for MD-PT vs. MD-DC care in 2007 (P =.0648). Ratios range from 0.24 to 0.67 (ie, total allowed charges on average 33-76% lower for DC patients) (Table 4). For CLBP, risk-adjusted mean charges were significantly greater for the above MD vs. DC patterns, with ratios ranging from 0.21 to 0.50 (ie, total allowed charges on average 50-79% lower for DC patients) (Table 5). For both ULBP and CLBP, risk-adjusted mean charges were consistently greater for MD-PT-referral care vs. MD-DC-referral care. DISCUSSION Low back pain is a highly prevalent condition that has created a substantial economic burden in the United States. Previous research has shown large increases in utilization and expenditures for the management of low back pain nationally throughout the decade. 1,5,8,13 These findings provide further evidence of this upward trend in North Carolina. The numbers of claims and charges increased in the NCSHP population, with claims rising 139% and 181% and total annual allowed charges increasing 242% and 282% for ULBP and CLBP, respectively. Average total allowed charges increased by 45% for ULBP and 38% for CLBP. Utilization of all care patterns increased but rose most dramatically for care involving MDs, PTs, and referral providers. The reasons for these steep increases are not entirely clear and probably represent a combination of factors such as increased prevalence of chronic LBP, changes in beliefs about pain and its management, as well as changes in the treatments used. 4 In our study, patterns of care involving referral providers (eg, radiology, surgical, and acute care facilities) generally incurred the largest charges; however there were important differences found between specific care patterns. When comparing the average charges over the decade for the DC-referral and MD-referral combination, the pattern including DC care incurred fewer total charges per patient. Similar results were found when considering the combination of MD-DC-referral care that incurred fewer total charges per patient than MD-PT-referral care. These cost differentials held in the risk-adjusted models where the mean charges for reflected appreciably lower average charges for patients in DC patterns of care. These findings are consistent with a study by Davis et al., which used data from the Medical Expenditure Panel Survey between 1998 and In that study, the authors suggest that the primary driver of increased spine care expenditures over time was not due to changes in DC or primary care expenditures, but instead resulted from a significant increase in specialty (non-primary care physician) medical care services. 8

8 Journal of Manipulative and Physiological Therapeutics Volume 39, Number 4 Hurwitz et al 259 Table 4. Risk-Adjusted Mean (Standard Error) of Total Allowed Charges per Patient and Cost Ratios With Their 95% Confidence Intervals for Uncomplicated Low Back Pain Among Patients in the Middle Quintile of Risk, by Pattern of Care and Year: North Carolina State Health Plan for Teachers and State Employees, 2006 to 2009 Year Pattern of care DC only $ ($11.74) $ ($12.44) $ ($19.26) $ ($17.70) MD only $ ($26.22) $ ($38.27) $ ($42.13) $ ($33.57) Cost ratio 0.33 (0.27, 0.39) 0.24 (0.21, 0.27) 0.30 (0.27, 0.33) 0.33 (0.31, 0.36) P P b.0001 P b.0001 P b.0001 P b.0001 MD-DC $ ($39.57) $ ($102.38) $ ($91.05) $ ($107.43) MD-PT $ ($98.45) $ ($155.21) $ ($198.06) $ ($158.00) Cost ratio 0.43 (0.26, 0.70) 0.67 (0.44, 1.02) 0.44 (0.32, 0.60) 0.57 (0.43, 0.75) P P =.0006 P =.0648 P b.0001 P b.0001 DC-referral $ ($44.52) $ ($113.19) $ (84.48) $ ($139.46) MD-referral $ ($30.72) $ ($43.39) $ ($52.05) $ ($63.05) Cost ratio 0.43 (0.26, 0.71) 0.59 (0.36, 0.98) 0.44 (0.32, 0.62) 0.53 (0.40, 0.71) P P =.0009 P =.043 P b.0001 P b.0001 MD-DC-referral $268.74($54.71) $ ($86.74) $ ($79.34) $ ($185.82) MD-PT-referral $ ($70.27) $ ($151.89) $ ($150.54) $ ($124.53) Cost ratio 0.65 (0.39, 1.07) 0.41 (0.24, 0.68) 0.34 (0.24, 0.50) 0.95 (0.72, 1.26) P P =.0927 P =.0006 P b.0001 P =.7212 DC, doctor of chiropractic; MD, medical doctor; PT, physical therapist Means adjusted for between-pattern differences in risk scores that reflect measure of risk of expected healthcare cost and utilization relative to that of the overall population taking into account age, sex, primary diagnosis, comorbidities, use of prescription drugs and risk scores; risk scores available only. Table 5. Risk-Adjusted Mean (Standard Error) Total Allowed Charges per Patient and Cost Ratios With Their 95% Confidence Intervals for Complicated Low Back Pain Among Patients in the Middle Quintile of Risk, by Pattern of Care and Year: North Carolina State Health Plan for Teachers and State Employees, 2006 to 2009 Year Pattern of care DC only $ ($14.61) $ ($16.04) $ ($25.88) $ ($25.11) MD only $ ($71.08) $ ($86.30) $ ($106.10) $ ($83.73) Cost ratio 0.21 (0.16, 0.29) 0.21 (0.17, 0.26) 0.27 (0.23, 0.32) 0.29 (0.25, 0.33) P P b.0001 P b.0001 P b.0001 P b.0001 MD-DC $ ($78.60) $ ($156.44) $ ($98.05) $ ($135.65) MD-PT $ (295.51) $ ($418.36) $ ($400.40) $ ($270.83) Cost ratio 0.36 (0.15, 0.83) 0.35 (0.17, 0.71) 0.27 (0.16, 0.46) 0.47 (0.32, 0.69) P P =.0174 P =.0039 P b.0001 P b.0001 DC-referral $208.37($84.87) $ ($121.71) $ ($281.01) $ ($137.76) MD-referral $ ($59.72) $ ($112.35) $ ($114.47) $ ($112.48) Cost ratio 0.44 (0.20, 0.95) 0.41(0.24, 0.71) 0.50 (0.25, 0.98) 0.27 (0.17, 0.42) P P =.0366 P =.0015 P =.045 P b.0001 MD-DC-referral $ ($100.99) $ ($157.12) $618.7 ($145.21) $ ($196.01) MD-PT-referral $ ($118.26) $ ($232.36) $ ($277.23) $ ($310.44) Cost ratio 0.66 (0.31, 1.41) 0.61 (0.30, 1.24) 0.36 (0.21, 0.62) 0.42 (0.27, 0.66) P P =.2827 P =.1752 P =.0003 P b.0001 DC, doctor of chiropractic; MD, medical doctor; PT, physical therapist Means adjusted for between-pattern differences in risk scores that reflect measure of risk of expected healthcare cost and utilization relative to that of the overall population taking into account age, sex, primary diagnosis, comorbidities, use of prescription drugs and risk scores; risk scores available 2006 to 2009 only. When evaluating the patterns of care that did not involve referral providers, we found important differences based on provider composition. On average over the decade, the combination of MD-DC care for the treatment of ULBP incurred fewer total charges per patient than MD-PT care. This relationship was reversed for CLBP, where the combination of MD-DC care incurred greater total charges per patient than the combination of MD-PT care. However, when relevant patient factors where accounted for in the risk-adjusted models, the charges for 2006 to 2009 were found to be less for MD-DC care than for MD-PT care. Similarly, when comparing MD-only to DC-only patterns of care, MD-only care was more expensive than DC-only care for ULBP and CLBP.

9 260 Hurwitz et al Journal of Manipulative and Physiological Therapeutics May 2016 Our findings are consistent with findings from other population-based studies that have investigated the costs associated with low back pain management. Several studies have shown an association between chiropractic care and lower utilization of costly spine services such as all-class medications, diagnostic imaging, spinal injections, and surgery Two of these studies evaluated the cost of low back pain management when DCs were the first provider seen for an episode of care. In both studies when a DC was the portal of entry provider, the total cost was lower when compared to low back pain patients that had other entry points into the healthcare delivery (eg, physical therapy, medical specialty, etc) system. 15 Another study in a commercially insured population showed low back pain episodes initiated with DCs cost 20% less than those initiated with physicians. 17 Our findings and these other studies provide evidence suggesting that DCs can play an important role in the cost-effective management of low back pain populations and can be an integral resource to patients, providers, and policy makers. Patients are now being asked to take on greater financial responsibility through higher premiums, deductibles, and copayments. For example, 23% of employers now offer high deductible benefit plans where the lowest family deductible starts at $ Further evidence of this trend can be seen in the new benefit options brought about by the Affordable Care Act where patients are responsible for 40% of their medical costs in the Bronze Plan. 19 Patients must now make important and informed choices about the costs associated with their use of healthcare services. In most instances, patients are free to access directly primary and specialty medical physicians, physical therapists, and chiropractors. Our findings can help patients to recognize potential economic ramifications when choosing a provider for low back pain management. Providers are also being impacted financially by changes in the healthcare reimbursement system, which is moving away from the traditional fee-for-service model. According to a recent survey on healthcare payment reform, 40% of all commercial in-network payments are now value-oriented, meaning they are tied to provider performance or designed to reduce unnecessary utilization and costs. 20 Given the rate of adoption of these reimbursement methods, it is anticipated that nearly 70% of all payments will be value-based by Importantly, the majority of these value-oriented payments currently put providers at financial risk for their performance. 20 As providers accept more risk, they will start taking a new look at how they can more efficiently manage their patient populations. Our findings suggest that efficiency gains may be obtained through the integration of chiropractors into an inter-professional model of care for the low back pain population. Policy makers are beginning to use levers that incent the use of lower cost services for the management of LBP, such as DC care, to mitigate the rising cost associated with specialty/referral care. As stated earlier, specialty medical services have been identified as a significant driver of spine care costs. However, despite the greater use of specialty services from 1998 and 2008, investigators have noted that self-reported measures of mental health, well-being, physical functioning, and work, school or social limitations had worsened during this time period in the low back pain population. 5 Therefore, it does not appear that the health status of individuals with spine disorders has improved, despite greater use of specialty services with their associated increased cost. 8 Incenting the use of DC care through patient copayment reductions may be one mechanism to curb the high utilization of specialty/referral care. This is necessary because previous studies have shown chiropractic care to be very sensitive to patient cost-sharing. 22,23 The RAND Health Insurance Experiment found that use of chiropractic care (ie, access) was cut in half (odds ratio = 0.49) when patient cost-share was 25% or more of the visit cost. 22 Therefore, the current trend towards higher patient cost-sharing may be driving patients away from the lower cost chiropractic services and towards more expensive patterns of care. In 2007, the North Carolina legislature reversed a mandate requiring that the same copayment amount apply equally to primary medical care and chiropractic care in the NCSHP population. Following the legislative repeal of this mandate, the NCSHP implemented a new policy in October of that year treating chiropractic care as a specialty service. This policy change increased the copayments for DC visits by 20% to 100%, thereby creating a substantial cost-share disparity between primary care and chiropractic. Throughout the decade, the combined charges for complicated and uncomplicated low back pain in 2000 were 36.1 million dollars, which had escalated to 107 million dollars by However, there was a noticeable break in this upward trend the following year when in 2007 the total allowed charges dropped to million dollars. Following the policy change in late 2007, which raised DC copayments to specialty levels, there was resumption in an upward trend of the total allowed charges for LBP. This trend increased upward in 2008 to 118 million dollars, and eventually rose to million dollars in 2009 (Fig 1). Exploring the temporal association of this policy (legislative repeal) decision to raise DC copayments, along with its effects on utilization and charges, was beyond the scope of this current analysis and will be the focus of our future work. Limitations There are methodologic challenges inherent in the analysis of health insurance databases. For this study, these challenges arise in the form of: inability to control for tiering; possible inaccuracy of diagnostic, management and

10 Journal of Manipulative and Physiological Therapeutics Volume 39, Number 4 Hurwitz et al 261 treatment codes; and lack of availability of risk factors for a portion of the analysis. All of these factors could affect the comparability between groups, and potentially create provider under-representation. An additional limitation was that our analysis was limited only to patients from the NCSHP in North Carolina. However, the population demographics of North Carolina during this time frame were approximately the same as national demographics. 24 Therefore, the results may be generalizable to other populations. In addition, as other studies have also mentioned, 1 our estimates might be conservative as they do not account for healthcare use related to comorbidities that might be common among patients with spine disorders (eg, psychological distress, other pain complains, etc). The objective of this study was to compare use and charges of low back pain care in patients with a primary low back pain diagnosis. Study Strengths A major strength of the study was the large amount of low back pain claims made available to us for analysis. The data were from claims generated by approximately persons over the decade in North Carolina, in several different pathways of healthcare services. This study and the series of papers it has generated on the treatment of low back pain, neck pain 25 and headache, 26 provides unique economic examination for healthcare policy makers and legislators. When costs are viewed vertically as if in silos (eg, DC-only costs, MD-only costs), increasing utilization of one particular provider is seen as a net cost increase. However, when costs are viewed across the silos, as this study has done, an increase in utilization of one provider group can result in a net cost decrease given its effect on the patient population. This is an opportunity to view costs laterally versus a confined, vertical analysis. CONCLUSION This study confirms that overall utilization and charges for the treatment of low back pain increased dramatically over the decade in North Carolina. For uncomplicated low back pain without referral care, charges for MD-DC care were on average less than charges for MD-PT care. For complicated low back pain without referral care, average charges are greater for MD-DC care than average charges for MD-PT care. For both uncomplicated and complicated low back pain with additional referral care, charges are consistently less for MD-DC care than charges for MD-PT care. The utilization of chiropractic care appears to play an important role in the management of populations with low back pain. Practical Applications This study found that utilization and expenditures for treatment of low back pain (LBP) in North Carolina have increased dramatically over the decade 2000 to MD and DC care alone with no referrals are the least expensive patterns of low back pain care. DC care alone or with MD care incurs appreciably fewer charges for uncomplicated low back pain than MD care with or without PT care. ACKNOWLEDGMENT We gratefully acknowledge the assistance of Anthony W. Hamm, DC, FACO, and Marc S. Gottlieb, DC. Dr. Hamm assisted with the classification of conditions according to ICD-9 codes and Dr Gottlieb helped with design of the data extraction models. FUNDING SOURCES AND POTENTIAL CONFLICTS OF INTEREST This study was made possible by the North Carolina General Assembly and with grants from NCMIC Research Foundation and Health Network Solutions. Conflicts of interest reported included ELH: financial (personal fees from (a) escrow account managed by Dr Phelan, (b) the RAND Corporation). SPP: institutional (self-managed escrow account); financial (personal fees from self-managed escrow account). RCA: institutional (board member of Health Network Solutions that provided some financial support for the study). RBP: financial (personal fees from NCMIC Foundation). MJS: financial (personal fees from NCMIC for expert testimony and lectures). JMS: financial (personal fees from Landmark Healthcare). DL: financial (personal consulting fees). CONTRIBUTORSHIP INFORMATION Concept development (provided idea for the research): E.L.H., S.P.P., R.C.A. Design (planned the methods to generate the results): E.L.H., S.P.P., R.C.A. Supervision (provided oversight, responsible for orga nization and implementation, writing of the manuscript): E.L.H., S.P.P., R.B.P., E.A.L. Data collection/processing (responsible for experiments, patient management, organization, or reporting data): E.L.H., S.P.P., R.B.P.

11 262 Hurwitz et al Journal of Manipulative and Physiological Therapeutics May 2016 Analysis/interpretation (responsible for statistical analy sis, evaluation, and presentation of the results): E.L.H., S.P.P., D.L., M.J.S., J.M.S., R.B.P. Literature search (performed the literature search): E.L.H., J.G., M.V., J.M.S. Writing (responsible for writing a substantive part of the manuscript): E.L.H., S.P.P., J.G., M.V., M.J.S., J.M.S., R.B.P. Critical review (revised manuscript for intellectual content, this does not relate to spelling and grammar checking): E.L.H., D.L., M.J.S., J.M.S., R.B.P., S.P.P., E.A.L., R.C.A., M.V. REFERENCES 1. 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, Spine 2009;34: McNeil JM, Binette J, CDC. Prevalence of disabilities and associated health conditions among adults - United States, 1999 (Reprinted from MMWR 2001;50:120-5). JAMA 2001; 285: Prevalence and most common causes of disability among adults - United States, Morb Mortal Wkly Rep 2009;58: Friedly J, Standaert C, Chan L. Epidemiology of spine care: the back pain dilemma. Phys Med Rehabil Clin N Am 2010;21: Martin BI, Deyo RA, Mirza SK, et al. Expenditures and health status among adults with back and neck problems. JAMA 2008; 299: Haldeman S, Dagenais S. A supermarket approach to the evidence-informed management of chronic low back pain. Spine J 2008;8: Dagenais S, Caro J, Haldeman S. A systematic review of low back pain cost of illness studies in the United States and internationally. Spine J 2008;8: Davis MA, Onega T, Weeks WB, Lurie JD. Where the United States spends its spine dollars: expenditures on different ambulatory services for the management of back and neck conditions. Spine 2012;37: Davis MA, Sirovich BE, Weeks WB. Utilization and expenditures on chiropractic care in the United States from 1997 to Health Serv Res 2010;45: Dunn KM, Hestbaek L, Cassidy JD. Low back pain across the life course. Best Pract Res Clin Rheumatol 2013;27: Smith M. Identifying Episodes of Back Pain Using Medical Expenditures Panel Survey Data: Patient Experience, Use of Services, and Chronicity. J Manipulative Physiol Ther 2010;33: SAS Institute. SAS (release 9.2) statistical software; 2012 [Cary, NC]. 13. Smith M, Davis MA, Stano M, Whedon JM. Aging baby boomers and the rising cost of chronic back pain: secular trend analysis of longitudinal Medical Expenditures Panel Survey data for years 2000 to J Manipulative Physiol Ther 2013;36: Ivanova JI, Birnbaum HG, Schiller M, Kantor E, Johnstone BM, Swindle RW. Real-world practice patterns, health-care utilization, and costs in patients with low back pain: the long road to guideline-concordant care. Spine J 2011;11: Allen H, Wright M, Craig T, et al. Tracking low back problems in a major self-insured workforce: toward improvement in the patient's journey. J Occup Environ Med 2014;56: Legorreta AP, Metz RD, Nelson CF, Ray S, Chernicoff HO, Dinubile NA. Comparative analysis of individuals with and without chiropractic coverage: patient characteristics, utilization, and costs. Arch Intern Med 2004;164: Liliedahl RL, Finch MD, Axene DV, Goertz CM. Cost of Care for Common Back Pain Conditions Initiated with Chiropractic Doctor Vs Medical Doctor/Doctor of Osteopathy as First Physician: Experience of One Tennessee-Based General Health Insurer. J Manipulative Physiol Ther 2010; 33: Kaiser Family Foundation Employer Health Benefits Survey. Available from: report/2013-employer-health-benefits/. Updated August 20, Accessed October 1, Healthcare.gov. How to pick a health insurance plan: The metal' categories: Bronze, Silver, Gold & Platinum. Available from: Accessed October 1, Catalyst for Payment Reform (CPR). Catalyst for Payment Reform (CPR) National Scorecard on Payment Reform. [Internet]; c2015 [cited 2015 October 1]. Available from: org/images/documents/nationalscorecard2014.pdf. 21. McKesson Health Solutions. McKesson Health Solutions: The State of Value-Based Reimbursement and the Transition from Volume to Value in Available from: mckessonhealthsolutions/images/mhs-2014-signature-research- White-Paper.pdf. Accessed October 1, Shekelle PG, Rogers WH, Newhouse JP. The effect of cost sharing on the use of chiropractic services. Med Care 1996;34: Stevans JM, Zodet MW. Clinical, demographic, and geographic determinants of variation in chiropractic episodes of care for adults using the Medical Expenditure Panel Survey. J Manipulative Physiol Ther 2012;35: Kff.org. North Carolina: Demographics and Economy. [Internet]; c2015 [cited 2015 October 1]Menlo Park, CA: Kaiser Family Foundation; 2013 [updated November 10, 2013]. Available from: demographics-and-the-economy/?state=nc. 25. Hurwitz EL, Li D, Schneider MJ, et al. Variations in patterns of utilization and charges for the care of neck pain in North Carolina, : A statewide claims' data analysis. J Manipulative Physiol Ther 2016;39: Hurwitz EL, Vassilaki M, Li D, et al. Variations in patterns of utilization and charges for the care of headache in North Carolina, : A statewide claims' data analysis. J Manipulative Physiol Ther 2016;39: