WHITE PAPER

GEOGRAPHIC VARIATION IN SPENDING AND ULTILIZATION AMONG THE COMMERCIALLY INSURED

WILLIAM D. MARDER, PHD GINGER S. CARLS, PHD EMILY EHRLICH, MPH RAY FABIUS, MD, CPE, FACPE JULY 27, 2011

TABLE OF CONTENTS
EXECUTIVE SUMMARY ...................................................................................................................................2 INTRODUCTION ...............................................................................................................................................3 Explaining Variation ....................................................................................................................................3 METHODS.........................................................................................................................................................4 FINDINGS .........................................................................................................................................................5 Summary of Findings ..................................................................................................................................5 What Are the 10 Highest and Lowest Medical Spending MSAs? ..............................................................5 Is the pattern of geographic variation the same across different age groups and types of spending?..7 Are high medical spending MSAs also high spending for drugs, inpatient, and outpatient medical care? ............................................................................................................................................. 10 Are high spending areas for seniors also high spending for adults and children? ................................. 11 CONCLUSIONS ............................................................................................................................................... 12 APPENDIX ...................................................................................................................................................... 12 REFERENCES ................................................................................................................................................. 13

Geographic Variation in Spending and Utilization Among the Commercially Insured 1

EXECUTIVE SUMMARY
Variation in healthcare utilization and spending across the United States has been carefully documented for the Medicare population. Less is known about variation among the privately insured population of the U.S. This paper uses the Thomson Reuters MarketScan Research Databases to examine variation in spending for enrollees with employer-sponsored health insurance (private insurance). We found substantial spending variation across metropolitan areas for those with private insurance, but different patterns than have been described in the literature based typically on Medicare Fee-for-Service (FFS) Parts A and B. We also found the lowest medical care spending metropolitan statistical area (MSA) is Ogden-Cleareld, Utah and the highest spending MSA is Anderson, Ind. Previous research has highlighted McAllen, Texas as the highest medical care spending MSA for Medicare, but our research found that this MSA is among the 10 lowest spending MSAs for commercially insured enrollees. In conclusion, we determined that a better understanding of variation in healthcare spending among the privately insured is needed to establish the appropriate policy response and avoid unintended policy effects.

2 Geographic Variation in Spending and Utilization Among the Commercially Insured

INTRODUCTION
Variation in healthcare utilization and spending across the United States is well documented. Beginning with Jack Wennbergs study of variation in the 1970s, researchers have observed variation in healthcare utilization and spending across geographic regions.1 And although there is speculation as to why there is variation, there is general consensus in the eld that unexplained geographic variation in use and healthcare spending cannot be fully explained by differences in demographics (age and gender) and available measures of health status or risk adjustment.2 Furthermore, it is implied that regions with high spending represent overuse, creating unnecessary spending in the healthcare system. Much of the literature on geographic variation in spending and utilization is based on data from the Medicare program (Parts A and B). While it is not debated that there is considerable variation in healthcare utilization and spending in a public insurance program such as Medicare, this experience of variation may or may not be unique to seniors and/or public programs. Making policy decisions based solely on Medicare evidence is risky since there are important differences between programs for the over-65 population and under-65 population and their healthcare needs. Evidence from private insurance and the under-65 population is needed. This paper provides a descriptive look at geographic variation observed for the privately insured population of the U.S. We do not resolve controversies about overuse or underuse; rather, we confront the reality that patterns of variation may not be the same across different insurance arrangements. Policy responses should take this into account to avoid unintended effects. Existing explanations for geographic variation may need to be modied to account for both the patterns observed in Medicare, as well as patterns observed in commercially insured populations. EXPLAINING VARIATION A policy response to geographic variation in healthcare spending will depend on the source of this variation. Several explanations have been proposed to account for geographic variation in healthcare. The answer may lie in some combination of these hypotheses. Supply of Healthcare Clinical Uncertainty. Clinical uncertainty allows for provider-driven inuences on care.1 Variations in provider-driven demand may be due to differences in local practice culture and training, as well as nancial incentives.3, 4, 5, 6 Variations in Local Delivery System Efciency. The availability of primary care physicians and specialists (especially under-supply of primary care physicians and over-supply of specialists), integrated delivery systems, and other aspects of the healthcare system may lead to differences in efciency and use of healthcare resources.7, 8 Healthcare Demand. Some level of variation may be due to differences in underlying health, health behaviors, and preferences for healthcare across regions that cannot be completely accounted for using current methods.8 Market Structure, Pricing, and Competition. Differences in pricing mechanisms market prices versus administrative prices and hospital competition may explain some observed geographic patterns and differences in Medicare and commercial insurance.9, 10 Healthcare Fraud and Abuse. Fraud and abuse may explain some of the geographic variation in the Medicare population where fraudulent billing schemes are well documented. 11

Geographic Variation in Spending and Utilization Among the Commercially Insured 3

METHODS
Using the MarketScan Research Databases, we explored patterns of geographic variation in 2009 spending for three age groups of enrollees with commercial insurance provided by an employer (typically large- to medium-sized rms): Children (age 0-17 years) Adults (age 18-64 years), and Seniors (age 65 years and over) This analysis represents the experience of 5.7 million continuously-enrolled children, 16 million continuously-enrolled adults, and 1.8 million continuously-enrolled seniors. The seniors in our analysis are a subset of Medicare beneciaries with supplemental insurance from their previous employers, and are more likely to be healthier and have higher incomes than the full set of Medicare enrollees. In our analysis, we used the geographic unit of metropolitan statistical area (MSA), representing cities in the United States. We included the 382 MSAs with at least 100 enrollees in each of the age categories. Spending was measured by allowed charges, which includes spending by both patients and the insurance plan. We explored geographic variation across MSAs and across age groups in: Total spending Medical care spending Inpatient medical care spending Outpatient medical care spending Outpatient prescription drug spending

We provide descriptive results unadjusted for demographic characteristics, pricing, or health status. A revised version of this paper will explore the impact of demographic and other adjustments. When available, we compare our results to similar unadjusted estimates reported in the literature. This paper is organized as follows. First we describe geographic variation in spending by answering the following questions: What are the 10 highest and 10 lowest medical care spending MSAs for all age groups and how do these MSAs rank by age group? Is the pattern of geographic variation the same across populations and type of spending? Since these analyses suggested that geographic spending patterns varied by age group and type of spending, we measured how different or similar patterns were by age group and type of spending. We analyzed the following correlations: Medical-Drug spending for all ages and by age group Inpatient-Outpatient medical spending for all ages and by age group Child-Adult total healthcare, medical, and drug spending Adult-Senior total healthcare, medical, and drug spending Child-Senior total health, medical, and drug spending

4 Geographic Variation in Spending and Utilization Among the Commercially Insured

FINDINGS
SUMMARY OF FINDINGS

Our analysis revealed ve main ndings: 1. Signicant geographic variation in total healthcare spending occurred in the under-65 population, as well as the over 65 population. 2. The location of the highest and lowest medical care spending MSAs varied signicantly by age group and by type of spending and was substantively different from previously published results found in the Medicare FFS population. 3. Variation in drug spending was typically higher than variation in medical care spending, except for seniors with supplemental insurance from a previous employer. 4. Inpatient and outpatient medical spending appeared to be strongly positively correlated with each other, and medical and outpatient drug spending were weakly correlated. 5. Medical care spending was weakly correlated between age groups, while drug spending was more strongly correlated.
WHAT ARE THE 10 HIGHEST AND LOWEST MEDICAL SPENDING MSAS?

Ogden-Cleareld, Utah has the lowest medical care spending of $2,623, which was 36 percent of the spending in the highest spending MSA (Anderson, Ind. $7,231) (Table 1). McAllen, Texas has previously been identied as the highest medical care spending area or among the highest medical spending areas for Medicare recipients, driven by home healthcare spending.11,12 In the commercially insured population, we nd similar high spending for seniors, but not for adults and children (Table 1). Among commercially insured seniors, McAllen, Texas had medical spending that was 129.7 percent of the national average and ranked in the 91st percentile. However, among commercially insured adults, McAllen, Texas had medical spending in the lowest percentile, with spending of only 70.2 percent of the national average among adults. El Paso, Texas has been highlighted as a city with similar demographics as McAllen, Texas, but Medicare spends half as much in El Paso as in McAllen.12 For seniors with commercial insurance, medical care spending in these two MSAs was very similar ($12,663 in El Paso and $12,484 in McAllen, both in 91st percentile). A study of Blue Cross Blue Shield, under-65, enrollees that compared medical care spending in McAllen and El Paso found that McAllen actually had lower spending in the under-65 population.13 For adults with commercial insurance in the data reported here, we also nd lower medical care spending in McAllen ($2,944) compared with El Paso ($3,644). Miami, Fla. has also been highlighted as one of the highest medical care spending MSAs for Medicare recipients, with some speculation that this may be in part due to Medicare fraud and abuse in this region.11 The spending in Miami among the commercially insured was mixed. For seniors, Miamis medical spending was 135.5 percent of the national average and ranked high, in the 93rd percentile. However, among commercially insured adults, Miamis medical spending was relatively lower, at 109.0 percent of the national average and ranking in the 73rd percentile. Commercially insured children in Miami had high medical spending that was 133.1 percent of the national average (95th percentile).

Another commonly used geographic level of analysis is the hospital referral region (HRR), used by the Dartmouth Atlas. HRRs are based on the hospital referral regions for the Medicare population and may not be applicable to the under-65 population included in our analysis.

Geographic Variation in Spending and Utilization Among the Commercially Insured 5

TABLE 1. LOWEST AND HIGHEST MEDICAL SPENDING MSAS FOR ALL AGES
10 LOWEST SPENDING MSAS

MSA

Medical Spending, Annual $ All $2,623 $2,719 $2,762 $2,916 $2,919 $2,942 $2,950 $2,979 $2,996 $3,029

Percentile Rank for Medical Care Spending All 1 1 2 2 2 2 3 3 3 3 Children 14 27 10 3 10 12 32 46 57 40 Adults 3 2 2 3 4 5 1 6 2 9 Seniors 17 25 20 9 73 56 91 30 81 72

Medical Spending as Percent of National Average All 63.9% 66.2% 67.3% 71.0% 71.1% 71.7% 71.9% 72.6% 73.0% 73.8% Children 74.8% 81.7% 73.5% 63.7% 72.6% 74.1% 83.4% 89.1% 93.5% 86.8% Adults 75.3% 71.2% 71.4% 74.3% 77.1% 77.4% 70.2% 79.4% 72.9% 81.7% Seniors 77.8% 82.6% 80.4% 70.3% 108.4% 95.9% 129.7% 84.4% 113.5% 107.7%

Age Groups Ogden-Clearfield, UT Dubuque, IA FayettevilleSpringdale-Rogers, AR-MO Fort Smith, AR-OK Laredo, TX Amarillo, TX McAllen-EdinburgMission, TX Salt Lake City, UT Fargo, ND-MN Sioux City, IA-NE-SD

10 HIGHEST SPENDING MSAS

MSA

Medical Spending, Annual $ All $7,231 $7,168 $6,528 $6,312 $6,128 $6,123 $6,061 $5,977 $5,976 $5,931

Percentile Rank for Medical Care Spending All 100 100 99 99 99 99 98 98 98 98 Children 74 18 96 31 90 78 24 45 1 49 Adults 100 90 99 83 84 90 80 98 85 99 Seniors 59 83 99 85 50 65 78 39 70 33

Medical Spending as Percent of National Average All 176.2% 174.7% 159.1% 153.8% 149.3% 149.2% 147.7% 145.7% 145.6% 144.5% Children 102.0% 77.5% 135.9% 83.4% 117.2% 105.7% 79.6% 89.0% 54.8% 89.9% Adults 163.2% 122.7% 142.6% 115.6% 117.0% 121.3% 112.5% 137.5% 117.7% 152.7% Seniors 97.2% 116.6% 187.8% 119.0% 92.0% 101.7% 112.1% 88.0% 106.0% 85.2%

Age Groups Anderson, IN Punta Gorda, FL Racine, WI Naples-Marco Island, FL Ocean City, NJ Barnstable Town, MA Flint, MI Lake Havasu CityKingman, AZ Ocala, FL Carson City, NV

Percentile Rank is ranking from 1 to 100 where 1 = MSAs with medical care spending in the lowest 1% of MSAs and 100 is MSAs with the highest 1% in spending. National Average Annual Medical Care spending was $4,104 overall, $1,920 children, $4,197 adults, and $9,622 seniors. Source: Thomson Reuters MarketScan Research Databases, 2009

6 Geographic Variation in Spending and Utilization Among the Commercially Insured

IS THE PATTERN OF GEOGRAPHIC VARIATION THE SAME ACROSS DIFFERENT AGE GROUPS AND TYPES OF SPENDING?

To further explore geographic variation in spending, we characterized spending by MSAs using a common measure of the ratio of high-spending to low-spending areas (Figure 1) as well as display annual spending for each MSA in maps for each age group (Figure 2-4). For our ratio of high to low spending areas, high spending MSAs were at the 90th percentile and low spending MSAs were at the 10th percentile across the country (Figure 1). We found that geographic variation in drug spending was typically larger than variation in medical care spending, except for seniors. Drug spending for all age groups in MSAs at the 90th percentile was twice that of spending for MSAs at the 10th percentile (2.06 times as high) but medical spending was only 60 percent higher in the 90th percentile compared with the 10th (1.6 times as high). For seniors, this ratio was higher for medical care than drugs (1.40 for drugs, 1.74 for medical care). Geographic variation in medical care spending was larger among seniors with supplemental insurance than found in previous research based on seniors enrolled in Medicare (Medicare Parts A and B). Previous research found that medical care spending in MSAs at the 90th percentile had spending 55 percent higher than MSAs at the 10th percentile, compared with 74 percent in these data (both unadjusted estimates).11
FIGURE 1. VARIATION IN TOTAL HEALTHCARE SPENDING, BY AGE GROUP Ration of Spending in MSAs ay 90th and 10th Percentiles

2.50 2.00
1.63 1.60 2.06 1.57 1.62 1.97 1.81 1.44 1.46 1.51 1.74 1.40

1.50 1.00 0.50 0.00

All Ages Total Spending

Children

Adults

Seniors Drug Spending

Medical Care Spending

Source: Thomson Reuters MarketScan Research Databases, 2009

Geographic Variation in Spending and Utilization Among the Commercially Insured 7

The following maps illustrate total healthcare spending (medical and drug) per enrollee with commercial health insurance for children ages 0 to 17; adults ages 18 to 64; and seniors ages 65 and older.
FIGURE 2. REGIONAL VARIATION IN TOTAL HEALTHCARE SPENDING, CHILDREN AGES 0 TO 17

Metropolitan Statistical Areas (MSA), ages 0 to 17

$1,236 - $1,778 $1,779 - $1,940 $1,941 - $2,070 $2,071 - $2,326 $2,327 - $3,604

Source: Thomson Reuters MarketScan Research Databases, 2009

ii The data were unadjusted and location was based on the residence of the patient. The un-shaded areas are non-metropolitan areas and metropolitan areas without data. Non-metropolitan areas include both micropolitan and non-CBSA (often called rural) counties.

8 Geographic Variation in Spending and Utilization Among the Commercially Insured

FIGURE 3. REGIONAL VARIATION IN TOTAL HEALTHCARE SPENDING, ADULTS AGES 18 TO 64

Metropolitan Statistical Areas (MSA), ages 18 to 64

$3,417 - $4,625 $4,626 - $5,025 $5,026 - $5,383 $5,384 - $5,881 $5,882 - $8,946

Source: Thomson Reuters MarketScan Research Databases, 2009 FIGURE 4. REGIONAL VARIATION IN TOTAL HEALTHCARE SPENDING, SENIORS AGE 65 AND OLDER

Source: Thomson Reuters MarketScan Research Databases, 2009

Geographic Variation in Spending and Utilization Among the Commercially Insured 9

ARE HIGH MEDICAL SPENDING MSAS ALSO HIGH SPENDING FOR DRUGS, INPATIENT, AND OUTPATIENT MEDICAL CARE?

We investigated correlations between different types of medical spending (inpatient and outpatient) and medical and drug spending across MSAs. A strong correlation in spending is indicated by a correlation coefcient close to +1 or -1, and no discernable pattern of correlation is indicated by a correlation coefcient of 0. As approaches are developed to address variation, it is important to understand whether spending across service types is a complement or substitute. Positive correlations show that service types are complements, and as spending in an MSA for one category increases, the other also increases. In this case, approaches to reduce spending in one service type may also reduce spending in the complementary service type. Negative correlations between the types of services suggest that the service types are substitutes and as spending in an MSA for one category of service increases, the other declines. We found: Spending for inpatient and outpatient medical care was strongly positively correlated, suggesting that they are complements (Figure 5). The correlation was strongest among seniors (correlation=0.73). This means that in areas where there is high inpatient usage there is also high outpatient usage, especially for seniors. The hypothesis that high outpatient utilization can reduce inpatient service needs is not supported by our study. Outpatient drug and medical spending were weakly correlated for seniors and children (correlation=0.11) suggesting that some drug and medical spending is a substitute of medical care spending, and some is a complement (Figure 5), similar to ndings from an earlier study.14 Drug and medical spending was moderately correlated among adults (correlation=0.34).

FIGURE 5. CORRELATION BETWEEN TYPES OF SPENDING, BY AGE GROUP 1.00

Correlation Coefficient

0.75

0.65

0.70 0.51

0.73

0.50 0.36 0.25 0.11 0.00 0.11 0.34

All Ages Medical-Drug

Children

Adults

Seniors

Inpatient-Outpatient

Source: Thomson Reuters MarketScan Research Databases, 2009

10 Geographic Variation in Spending and Utilization Among the Commercially Insured

ARE HIGH SPENDING AREAS FOR SENIORS ALSO HIGH SPENDING FOR ADULTS AND CHILDREN?

Using the same correlation measure described previously, we found that medical care spending was weakly correlated between age groups, while drug spending was more strongly correlated (Figure 6). This means that MSAs that had high medical care spending in one age group were not typically high spending for other age groups. Medical spending by adults and seniors was weakly correlated (0.06) across MSAs, as was medical spending by children and seniors (0.10). Prescription drug spending by children and adults was strongly correlated across MSAs (0.58), with more moderate correlations between adults and seniors (0.49) and children and seniors (0.29).
FIGURE 6. CORRELATION IN SPENDING BETWEEN AGE GROUPS, BY TYPE OF SERVICE 1.00

Correlation Coefficient

0.75 0.58 0.50 0.34 0.25 0.10 0.00 0.11 0.06 0.10 0.38 0.29 0.49

Total Child-Adult Adult-Senior

Medical Child-Senior

Drug

Source: Thomson Reuters MarketScan Research Databases, 2009

Geographic Variation in Spending and Utilization Among the Commercially Insured 11

CONCLUSIONS
In this analysis using the Thomson Reuters MarketScan Research Databases, we found that signicant variation in healthcare spending exists in the commercially insured population, but not in the same way for seniors in previously published research for Medicare Parts A and B. The reasons for these differences must be understood to generate effective policies that use resources efciently without compromising healthcare quality. Important differences exist in variation in spending for medical services compared to outpatient drug spending, which may be driven by practice patterns, pricing, and market structure. Future research should be conducted to determine whether the patterns we observed persist after adjustments for demographic characteristics and health status have been made. What is clear is that signicant variation in healthcare spending across our country exists, regardless of whether you look at age bands, medical, or drug expenditures.

APPENDIX
Two Thomson Reuters MarketScan Research Databases were used in this study: the Commercial Claims and Encounters (Commercial) Database and the Medicare Supplemental and Coordination of Benets (COB) Database. The Thomson Reuters MarketScan Commercial Database is composed of data contributed by medium and large employers and health plans. It contains the longitudinal administrative claims experience for tens of millions of individuals with employer-sponsored insurance annually. The data include active employees, early retirees, COBRA continues, and their dependents, covered under a variety of fee-for-service and capitated health plans: exclusive provider organizations, preferred provider organizations (PPOs), pointof-service (POS) plans, and health maintenance organizations (HMOs). More than 2000 employer-health plan combinations are represented in these data annually. The commercial database captures enrollment, inpatient, outpatient, and outpatient pharmacy information starting in the early 1990s. The commercial database contains fully adjudicated claims and full payment information, including the plan/employer liability, patient out-of-pocket costs, and any coordination of benets, or secondary payer/payment. The Thomson Reuters Medicare Supplemental Database contains the healthcare experience of individuals with Medicare supplemental insurance paid for by employers. Because it covers an older population than the Commercial database, the Medicare database is a key data source. Both the Medicare-covered portion of payment (represented as COB) and the employer-paid portion are included in this database.

12 Geographic Variation in Spending and Utilization Among the Commercially Insured

REFERENCES
1 2 3 4

10

11 12

13

14

Wennberg J, Gittelsohn A. Small area variations in healthcare delivery. Science December 14 1973; 182(117):1102-8. PMID 4750608. Gold M. Geographic variation in Medicare per capita spending: Should policymakers be concerned? Research Synthesis Report No 6, July 2004. Robert Wood Johnson Foundation. Last accessed May 5, 2011 at [Link]. Cutler DM, Sheiner L. The Geography of Medicare. American Economic Review 1999; 89(2):228-233. Fisher ES, Wennberg DE, Stukel TA, Gottlieb DJ, Lucas FL, Pinder EL. The implications of regional variations in Medicare spending. Part 1: The content, quality, and accessibility of care. Annals of Internal Medicine 2003; 138(4):273-287.1. Leape LL, Park RE, Soloman DH, Chassin MR, Kosecoff J, Brooke RH. Does inappropriate use explain small-area variations in the use of healthcare services? Journal of the American Medical Association 1990; 263(5):669-672. Dartmouths Center for the Evaluative Clinical Sciences, The Dartmouth Institute for Health Policy and Clinical Practice, [Link] last accessed May 6, 2011 Baicker K, Chandra A. Medicare spending, the physician workforce, and beneciaries quality of care. Health Affairs 2004;23(3):291-304. Congressional Budget Ofce. Geographic variation in health care spending. Washington, DC: Congressional Budget Ofce; February 2008. Chernew ME, Sabik LM, Chandra A, Gibson T, Newhouse JP. Geographic correlation between large-rm commercial spending and Medicare spending. The American Journal of Managed Care 2010; 16(2):131-138. Government Accountability Ofce. Federal Employees with Health Benets Program: Competition and Other Factors Linked to Wide Variation in Health Care Prices. GAO-05-856. Washington, DC: Government Accountability Ofce 2005. [Link] last accessed July 27, 2011. Gawande, A. The cost conundrum. The New Yorker. June 2009. Medicare Payment Advisory Commission (MedPac). Report to the Congress: Variation in Medicare Service Use. January 2011. Washington, DC. Franzini L, Mikhail OI, Skinner JS. McAllen and El Paso Revisited: Medicare Variations Not Always Reected In the Under-Sixty-Five Population. Health Affairs December 2010; 29(12): 2302-2308. Zhang Y, Baicker K, Newhouse JP. Geographic variation in Medicare drug spending. New England Journal of Medicine July 29, 2010; 363(5):405-409

Geographic Variation in Spending and Utilization Among the Commercially Insured 13

ABOUT THOMSON REUTERS Thomson Reuters is the worlds leading source of intelligent information for businesses and professionals. We combine industry expertise with innovative technology to deliver critical information to leading decision makers in the financial, legal, tax and accounting, healthcare and science and media markets, powered by the worlds most trusted news organization. With headquarters in New York and major operations in London and Eagan, Minnesota, Thomson Reuters employs 55,000 people and operates in over 100 countries. Thomson Reuters 777 E. Eisenhower Parkway Ann Arbor, MI 48108 USA [Link] 2011 Thomson Reuters. All rights reserved. Thomson Reuters and the Kinesis logo are trademarks of Thomson Reuters. MarketScan is a registered trademark of Thomson Reuters (Healthcare) Inc. H PRO MDS 1106 9938 HFMA MC

14 Geographic Variation in Spending and Utilization Among the Commercially Insured