Spine (Phila Pa 1976). 2009 (Sep 1); 34 (19): 2077–2084 ~ FULL TEXT
Brook I. Martin, MPH; Judith A. Turner, PhD; Sohail K. Mirza, MD, MPH;
Michael J. Lee, MD; Bryan A. Comstock, MS; Richard A. Deyo, MD, MPH
Departments of Health Services,
University of Washington,
STUDY DESIGN: Analysis of nationally representative survey data for spine-related health care expenditures, utilization and self-reported health status.
OBJECTIVE: To study trends from 1997 to 2006 in per-user expenditures for spine-related inpatient, outpatient, pharmacy, and emergency services; and to compare these trends to changes in health status.
SUMMARY OF BACKGROUND DATA: Although prior work has shown overall spine-related expenditures accounted for $86 billion in 2005, increasing 65% since 1997, the study did not report per-user expenditures. Understanding population-level per-user expenditure for specific services relative to changes in the health status may help assess the value of these services.
METHODS: We analyzed data from the Medical Expenditure Panel Survey, a multistage survey sample designed to produce unbiased national estimates of health care utilization and expenditure. Spine-related hospitalizations, outpatient visits, prescription medications and emergency department visits were identified using ICD-9-CM diagnosis codes. Regression analyses controlling for age, sex, comorbidity, and time (years) were used to examine trends from 1997 to 2006 in inflation-adjusted per-user expenditures, and utilization, and self-reported health status.
RESULTS: An average of 1774 respondents with spine problems was surveyed per year; the proportion suggested an increase in the number of people who sought treatment for spine problems in the United States from 14.8 million in 1997 to 21.9 million in 2006. From 1997 to 2006, the mean adjusted per-user expenditures were the largest component of increasing total costs for inpatient hospitalizations, prescription medications, andemergency department visits, increasing 37% (from $13,040 in 1997 to $17,909 in 2006), 139% (from $166 to $397), and 84% (from $81 to $149), respectively. A 49% increase in the number of patients seeking spine-related care (from 12.2 million in 1997 to 18.2 million in 2006) was the largest contributing factor to increased outpatient expenditures. Population measures of mental health and work, social, and physical limitations worsened over time among people with spine problems.
CONCLUSION: Expenditure increases for spine-related inpatient, prescription, and emergency services were primarily the result of increasing per-user expenditures, while those related to outpatient visits were primarily due to an increase in the number of users of ambulatory services.
From the FULL TEXT Article:
Treatment for back and neck problems accounted for approximately $86 billion in healthcare expenditure in the United States in 2005.  Sustained growth in expenditures for surgical and nonsurgical treatments for spinal disorders has raised questions about the value of common and often expensive interventions. Value in healthcare can be viewed as improvement in patient outcomes per unit of cost. At the population level, value can be assessed by comparing the relative improvement in the health of the affected population to the investment of resources. To get an estimate of value for spine expenditures at the population level, it is important to understand changes in spine expenditures relative to changes in the health status of individuals with spine problems.
We previously demonstrated that although total health care expenditures in US adults with spine problems increased 65% between 1997 and 2005, measures of mental health, physical functioning, and work and social limitations in this population worsened during this interval.  We calculated those expenditure estimates as the mean expenditure for the sample multiplied by the population size. Such per-capita estimates distribute the costs for a particular service over the population of interest whether or not the service was used. Per-capita estimates of expenditures are useful for policy makers, but are less informative to patients than are per-user estimates of expenditures. Per-user estimates distribute expenditures for a service over the population of service users, and therefore reflect the average individual financial burden for using a specific service.
In this report, we extend our previous analysis by estimating per-user expenditures (i.e., expenditures among those who sought health care services for spine problems) in 4 major components of health care:
(1) inpatient care,
(2) outpatient care,
(3) pharmaceutical care, and
(4) emergency department care.
We also extend our estimates an additional year by including 2006 data. Specifically, we:
(1) document changes from 1997 to 2006 in expenditures in these 4 components of health care
among those who sought services for spine problems;
(2) identify the contributions of treatment prevalence, per-user cost, and mean number of visits
to expenditures within each component; and
(3) describe the changes in from 1997 to 2006 on indicators of health status among patients
with spine problems.
Materials and Methods
We used the same general methods as described in our earlier report.  In brief, we examined data from 1997 to 2006 using the Medical Expenditure Panel Survey (MEPS), maintained by the Agency for Healthcare Research and Quality. [2-4] MEPS is a nationally representative annual cross-sectional survey of household medical utilization that is supplemented by provider and employer records. The survey uses a multistage sampling design and includes variables to account for clustering, stratification, oversampling, and nonresponse. These variables allow researchers to produce unbiased national estimates of expenditures, utilization, and health status. The survey is used widely in research on national utilization trends and policy.
Respondents with medical expenditures for common back problems were selected using the 3-digit International Classification of Disease, version 9, Clinical Modification (ICD-9-CM; codes 720-724, 737, 805-806, 839, 846-847). The use of ICD-9-CM codes for capturing spine conditions has been previously validated and is commonly used for reporting health care utilization.  Patient-reported conditions were translated into ICD-9-CM codes by MEPS personnel. A high level of agreement (at 3-digit ICD level) between patient descriptions of conditions and physician-provided information has been demonstrated.  We excluded patients younger than age 18 years from all analyses. However, because we were interested in national levels of utilization for back pain over a broad range of causes, we did not exclude people who had comorbid conditions or were pregnant.
We report the total number of outpatient visits (both office-based clinical visits and hospital outpatient visits), hospitalizations, emergency department visits, and medication prescriptions. Outpatient visits include visits to both physicians and nonphysician providers, such as nurses, chiropractors, and physical therapists. Hospital charges included all services incurred during a stay, including direct hospital care, diagnostic tests and procedures, imaging studies, and laboratory work. Events were counted only if there was an associated charge (a small number of visits involved zero expenditure because they were for phone calls or for follow-up visits in a prepaid insurance program).
Expenditures are defined as the sum of payments for care received for each outpatient visit, inpatient stay, prescription medication, and emergency room visit, including out-of-pocket payments and payments made by private insurance, Medicaid, Medicare, and other sources. Excluded are expenditures for over-the-counter medications, medical equipment, or services provided by free-standing radiology clinics. We relied on a direct method for capturing expenditures by averaging the aggregated expenditures within each service category for every respondent with a service that was directly linked to a spine-related primary diagnosis. This method is a conservative estimate of expenditures because it does not account for the excess utilization related to comorbidity (e.g., other pain complaints, psychological distress) frequently observed among people with spine problems. An alternative approach is the incremental method which defines spine expenditures as the difference in mean expenditures between those with and without spine problems. The incremental approach yields higher estimates because it included the excess expenditure associated with non-spine comorbidity among people with spine problems.  We did not use the incremental method of analysis because we were interested only in medical care that was specifically identified as being spine-related.
Multiple self-reported indicators of health were included for the respondents and are commonly used in spine research. SF-12 Physical Component Summary (PCS) and Mental Component Summary (MCS) scores are available from 2000 to 2006 (version 1 scores were transformed to the version 2 equivalents); norm-based scoring was used, with lower scores indicating worse health. [8, 9] We report the proportion of respondents with scores less than or equal to 40 (equivalent to 1 standard deviation or more below the population mean). Respondents also rated their physical and mental health as excellent, very good, good, fair, or poor. We dichotomized respondent ratings as fair-poor versus good-excellent. We also dichotomized respondent ratings of social activity limitations; work, school, or home activity limitations; and physical functioning limitations as some versus none. We also calculated reported the proportion of patients who reported that they received help or supervision with instrumental activities of daily living, such as taking medications, preparing light meals, doing laundry, or going shopping. Finally, we calculated the proportion of respondents who had any comorbidity based on Quan's enhanced comorbidity score. 
For each service category, we calculated the total number of respondents who reported health care utilization for spine problems, the proportion of all spine patients who used the service, the total number of service events, the mean number of events per user, the mean per-user expenditures, and the total national expenditures. We used descriptive statistics to summarize the health status measures using the responses of all respondents who reported spine problems. To account for changing population demographics over time, we adjusted the estimates for age, sex, and comorbidity using a 2-part generalized linear regression model with a gamma distribution and log link function. In a 2-part model, estimates of expenditures within each service are produced by multiplying the likelihood of receiving the service by the magnitude of the expenditures among those who received the service. Total expenditures for spine-related events were then calculated by summing the adjusted survey-weighted estimates of spine-related expenditures within each service category. We also included time (year) in the model to examine trends in expenditures across years. Expenditures from 1997 to 2005 were inflated to match their 2006 equivalents based on the Medical Consumer Price Index. 
We used logistic regression to produce estimates of the age-, sex-, comorbidity-, and time (year)-adjusted proportion of people with spine problems who reported poor/fair physical health, poor/fair mental health, and work, social, and physical limitations, as well as with MCS or PCS scores less than or equal to 40 in each year. All regression analyses incorporated weighting and design variables to account for multistage sampling methods. All analyses were performed using STATA, version 10.0 (Stata Corp, College Station, TX)
On average, 1774 respondents with medical expenditures for spine problems were surveyed per year. There was an increase in the estimated number (and treated prevalence) of people reporting care for spine problems in the United States from 14.8 million (10.8% of the population) in 1997 to 21.9 million (13.5%) in 2006 (Table 1). The average age of people who reported spine care increased slightly from 1997 to 2006 (47.7-50.4 years), as did the proportion who relied only on public health insurance (13.3%-15.6%). Racial composition showed a decrease in the proportion who described themselves as white.
Expenditures and Utilization
Combining all health care service categories and adjusting for age, sex, comorbidity, and inflation, the national health expenditures among US adults with spine problems increased from $19.4 billion in 1997 to $35.1 billion in 2006, an 82% increase. Figures 1 and 2 illustrate the percent increase in per-user expenditures and total national expenditures for spine problems for each year relative to 1997 for each of the 4 service categories. Although there were increases in all categories, per-user, and total national pharmacy expenditures increased dramatically relative to the other service categories. However, emergency expenditures also increased relative to inpatient and outpatient expenditures.
Change in per-user expenditures among
spine patients for 4 service categories relative to 1997
(MEPS, 1997-2006) (age, sex, and inflation adjusted).
Change in total national expenditures for spine
problems for 4 major service categories relative to 1997
(MEPS, 1997-2006) (age, sex, and inflation adjusted).
The annual per-user expenditure for hospitalizations increased an average of 3.7% per year, from $13,040 in 1997 (95% CI: 12,370-13,710) to $17,909 in 2006 (95% CI: 16,763-19,054; Table 2). In 2006, inpatient expenditures accounted for 29.3% of total national spine-related expenditures. Although the total expenditures for hospitalizations increased 53.0% from 1997 to 2006, the proportion of respondents with spine problems who had any hospitalization decreased slightly (3.5%-2.6%) over time. The number of users of inpatient services, total number of hospitalizations, and the mean number of hospitalizations per user remained generally stable over time.
The number of respondents with spine problems reporting outpatient visits increased 4.7% per year on average during the study period (Table 3). The proportion of spine patients who reported any outpatient visit remained fairly stable over time, whereas the mean number of visits per-user increased 0.8% per year on average and the mean per-user expenditure increased 1.9% per year. These changes contributed to a 6.7% average annual increase in total national outpatient expenditures, from $10.4 billion in 1997 (95% CI: $9.6 billion-$11.2 billion) to $18.4 billion in 2006 (95% CI: $16.6 billion-$20.1 billion), the largest absolute increase among all service categories. In 2006, outpatient expenditures accounted for 52.3% of total spine-related expenditures, representing the largest of all service categories.
The average annual per-user expenditures for prescription medication increased from 1997 to 2006 more than any other service category, an average of 10.2% per year, from $166 per patient in 1997 (95% CI: $164-$167) to $397 in 2006 (95% CI: $393-$402) (Table 4). Total national pharmacy expenditures for spine problems increased an average of 14.4% per year from 1997 to 2006 (or 232% overall from 1997 to 2006), the greatest proportional increase of all service categories. In 2006, prescription expenditures accounted for 12.2% of the total spine-related medical expenditures. In addition to the per-user expenditure for prescription drugs, the total number of patients reporting prescriptions and the number of prescriptions per patient also increased, while the proportion of all spine patients receiving prescriptions remained fairly stable. From 1997 to 2006, there was a 139% increase in per-user expenditure for prescriptions, compared to a 39.9% increase in number of users reporting a prescription. This included a 660% increase in expenditures for opioid medications among people with spine problems, from $246 million in 1997 to $1.9 billion in 2006.
Expenditures for spine-related emergency visits in 2006 accounted for 6.2% ($2.2 billion, 95% CI: $2.0-$2.4 billion) of the total spine-related expenditures, the smallest of all service categories (Table 5). However, total spine-related emergency expenditures increased 9.9% per year on average from 1997 to 2006, mostly due to a 7.1% per year average increase in per-user costs. Among patients with spine problems, the proportion reporting any emergency visits remained stable while the number of patients reporting emergency care increased 2.8% per year and the mean number of visits per patient increased 1.2% per year.
The proportion of patients who reported any limitation in physical functioning increased steadily and significantly over time, from 20.6% in 1997 to 28.3% in 2006 (Table 6). In addition, the proportion of respondents who reported poor mental health social limitation, and work/school/home limitation also increased steadily and significantly from 1997 to 2006. The proportion of patients with SF-12 PCS and MCS scores indicating poor health status, the proportion rating their physical health as fair or poor, and the proportion with limitations in instrumental activities of daily living rating did not change significantly over time among respondents with spine problems.
National expenditures for spine problems increased 82%, or an average of 7.0% per year, from 1997 to 2006. Paradoxically, measures of self-reported mental and physical health and activity limitations among those with spine problems worsened, and the percentage of respondents with spine problems who reported work, social and physical functioning limitations increased substantially during this period.
The estimated number and treated prevalence (those who sought care) of people reporting care for spine problems in the United States increased from 1997 to 2006. The treated prevalence of 13.5% in 2006 is substantially lower than the population prevalence of people with spine problems reported in the National Health Interview Survey (26%).  The National Health Interview Survey may have been more likely to include people with spine problems who do not seek medical care. Also from 1997 to 2006, we observed a decrease in the proportion of respondents with spine problems who described their race as white. At least partially, this may be an artifact of changes in the survey question; respondents were not allowed to select multiple racial categories until 2002.
The rise in per-user expenditure increases outweighed the rise in treated prevalence for inpatient, pharmacy, and emergency visits. For example, from 1997 to 2006, there was a 139% increase in per-user expenditures for prescriptions compared to a 40% increase in the number of users reporting prescriptions. In contrast, increases in spine-related outpatient expenditures were primarily the result of an increase in the number of people with spine problems seeking care as opposed to an increase in per-user costs, although per-user costs and the number of visits per user also increased slightly from 1997 to 2006. Within inpatient, pharmacy and emergency visit categories, increasing per-user costs may indicate greater use of expensive technology and treatments.  Consequently, a greater level of scrutiny, such as broader use of chronic care models and technology assessments, may be important for slowing expenditure growth in back pain. Rising per-user costs may also affect accessibility of spine care. [14, 15]
Pharmacy costs accounted for the greatest increase in both per-user and total national expenditures for spine patients. The number of patients reporting prescriptions and the number of prescriptions per patient increased over the study time period. Use and expenditures for prescriptions of opioid medications rose dramatically over the study period. Benefits and harms of analgesic medications, especially opioids, need to be better understood. [16-18] Recent reports of rising morbidity and mortality associated with increased opioid use are alarming. [19, 20] Although the data we analyzed were cross-sectional and do not allow for inferences about causal effects, the data are not consistent with increased use of opioids resulting in improved mental health and physical and role function. Other studies have also raised concerns about effects of opioid use on functional outcomes among patients with chronic pain. [21-23]
Most inpatient care for back pain is related to surgery, and increased per-user expenditures may be related to greater use of invasive procedures, increased use of spinal fixation devices, increased use of biologic materials, and a shift to ambulatory care for less expensive procedures, such as discectomy.  Our estimates of the number of hospitalizations are similar to those from the Health Care Utilization Project (HCUP) for discectomy, laminectomy, and fusion surgery.  Combined, these 3 procedures account for about 83% of all inpatient spine operations.  However, our study may underestimate the expenditures and utilization of inpatient services due to limitations in the use of the 3-digit ICD-9-CM coding algorithm. Specifically, commonly used codes for spondylolisthesis and spondylolysis (ICD-9-CM 738.4, 756.11, and 756.12) could not be included in our study because at the 3-digit level they could not be distinguished from non-spine related events. In addition, some fusion procedures are likely to be excluded from the analysis because they are coded using non-spine-specific diagnosis codes such as arthropathies, not elsewhere classified (716); joint disorder, not elsewhere classified (719); or orthopedic aftercare (V54). As a result of missing some fusion surgeries, this study likely underestimates the expenditures associated with inpatient utilization.
Our earlier studies, as well as data from the HCUP, suggest that the incidence and cost of fusion surgery have rapidly increased in recent years. [26, 27] For example, publicly available data from HCUP show that the mean cost of spinal fusion surgery increased from $35,000 in 1997 to $65,000 in 2006 after adjusting for inflation.  We hypothesize that the decrease in the total number of respondents reporting any spine-related hospitalization since 2004 is due to an increase in the use of ambulatory surgery for discectomy. This explanation is consistent with the observed increase in the number of patients reporting any outpatient visit and the increased inpatient per-user expenditures (as more expensive inpatient procedures would become more predominant in determining mean per-user costs). The decrease in inpatient visits may also partly be due to a growing shift from decompression surgery to fusion, and since fusion may be less well captured in MEPS (as previously explained) such a shift would result in an apparent decrease in overall spine-related hospitalizations. Despite the decrease in the number of hospitalizations since 2004, increased per-user expenditures outweighed the decrease in the number of users. Thus, there was no decrease in overall inpatient expenditure during this period.
These data represent a conservative estimate of total costs because they consider only a patient's primary diagnosis and do not account for the expenditures associated with comorbid conditions that are common among spine patients. Our previous study reported spine-related expenditures using an incremental approach that included the additional costs associated with comorbidities. In that study, we reported that national expenditures associated with spine problems totaled $86 billion in 2005, an increase of 65% since 1997.  The methods for estimating overall expenditures in this report represent a slight improvement over those in our previous report. In addition to incorporated another year of data (2006) advances in statistical software allowed inclusion of additional cases (previously a small number of cases were dropped because they were sampled from a stratum with a single sampling unit, which prevented calculating confidence intervals). Our estimates from 1997 to 2005 in this report are similar to those of the direct method in our earlier report after we take into consideration differences due to adjustments for inflation between the 2005 and 2006 study.
Our analysis has several limitations. Although the MEPS sampling design allows for unbiased estimates of national trends, the data are cross-sectional. Unlike prospective longitudinal data, within group changes over time in costs and outcomes cannot be inferred from our data. Furthermore, annual estimates of expenditures, utilization, and health status may vary due to sampling error; although long-term trends are generally robust. The available data lack detailed clinical information such as specific diagnosis, disease severity, disease duration, and spinal level(s). Although some MEPS respondents had secondary spinal diagnoses along with primary diagnosis of other conditions, we considered only the primary diagnosis for including patients in the analysis sample; this may underestimate both prevalence and utilization for spine problems. We did not evaluate specific treatments and functional outcome from specific interventions cannot be determined on the basis of these data.
Our findings suggest that the expenditure increases for spine problems are primarily the result of increasing per-user expenditures for inpatient, prescription, and emergency services, as well as an increasing number of people with spine problems using outpatient services. Increased use of imaging, surgical technology, chiropractic care, spinal injections, and prescription medications have all likely contributed to the observed increases in expenditures for spine problems. [29-32] The discordance in population-level changes in health status versus spine expenditures raises concerns about value for therapies for spine problems. A greater level of scrutiny may be needed in evaluating established treatments and introducing expensive new spinal technologies. Efforts to control costs must also assure patient safety and effectiveness. Analysis of observational data, comparative-effectiveness, and cost-effectiveness studies, clinical treatment guidelines, chronic care model approaches, and provider profiling may all have a role in guiding health policy to align value in clinical practice. Patients, providers, and insurers share the common goal of slowing the growth rate of medical expenditure without harming patient care.