|
H E A L T H T R A C K I N G :
T R E N D S
22 August 2006
The Rise In Spending Among
Medicare Beneficiaries:
The Role Of Chronic Disease Prevalence
And Changes In Treatment Intensity
Increasing numbers of beneficiaries being treated for five
or more conditions a year are driving Medicare spending upward.
by Kenneth E. Thorpe and David H. Howard
ABSTRACT:
We examine the impact of the rise in treated disease prevalence on the growth in Medicare beneficiaries’ health care spending. Virtually all of this spending growth is associated with patients who are under medical management for five or more conditions. This is traced to both a rise in true disease prevalence and changes in clinical treatment thresholds. Using the metabolic syndrome as a case study, we find that the share of patients treated with medications has increased 11.5 percentage points in less than ten years. This raises important questions about the “fit” of how Medicare pays for services for complex medical management. [Health Affairs 25 (2006): w378–w388; 10.1377/hlthaff.25.w378]
Medicare spending is projected to nearly triple from 3 percent of U.S. gross domestic product (GDP) in 2006 to 8.8 percent by 2030.1 The rising share of resources consumed by Medicare has accelerated policymakers’ interest in introducing new approaches for slowing growth in Medicare spending. Understanding the factors accounting for the recent and projected growth is important for crafting effective solutions for the future. This paper examines the impact of the rise in treated disease prevalence and spending per treated case on the growth in health care costs incurred by Medicare beneficiaries. Our analysis focuses on total health care spending for out-of-institution services received by Medicare beneficiaries, regardless of the source of payment.2 We believe that a clearer understanding of the factors driving the rise in total spending will prove critical for projecting future spending among the Medicare population and advancing future Medicare reforms.
Prior research. Previous studies have attempted to quantify the factors responsible for the rise in spending over time. They fall into two broad groups. The first set of papers tracks changes in Medicare spending over time by provider (hospitals, prescription drugs, physician services, and so on).3 Some of them are limited in that they track Medicare spending only. Medicare finances approximately half of all spending among Medicare beneficiaries, and, until recently, it did not offer an outpatient prescription drug benefit. Understanding the societal implications of population aging requires tracking all sources of spending.4
A second set of studies estimates the impact of disability, obesity, and other risk factors on Medicare spending during the 1990s.5 These results have been used to project future Medicare spending based on changes in obesity, disability, and the burden of disease over time. One important result from this work is the documented decline in disability rates among the elderly.6 Other factors held constant, a reduction in disability would reduce annual per capita Medicare spending. At the same time, however, reductions in disability have improved the life expectancy of the elderly. On balance, several analysts have found little difference in lifetime health care spending after age seventy among both the disabled and the nondisabled.7 Thus, continued reductions in disability among the elderly might not curb growth in Medicare spending during the next two decades.
A second result from this body of work focuses on the rapid rise in spending among the nondisabled elderly. Although disabled Medicare beneficiaries incur higher spending than their nondisabled peers do, the difference has declined dramatically over time. Between 1992 and 2000, spending per nondisabled beneficiary increased 82 percent, compared with 58 percent for those with one or two activities of daily living (ADL) limitations and 44 percent for those with five or more such limitations.8 Factors accounting for these differences in trends are not well understood.
Study goals. In this paper we examine the trends in Medicare beneficiaries’ annual health care spending. We provide further evidence on the sources of spending growth by documenting increases in treated disease prevalence and in the number of beneficiaries treated for multiple conditions. We also present new data concerning changes in treatment intensity.
Previous work has shown that increases in treated disease prevalence during the 1990s account for a large share of the growth in spending by private health insurers.9 Several factors could account for the rise in the prevalence of treated disease. These include increases in the population prevalence of disease, more-aggressive treatment of asymptomatic or mildly symptomatic patients, better detection of disease, innovation and new technologies that allow the treatment of conditions previously left untreated, and declining mortality rates.10 In contrast, increases in spending per treated case are largely driven by the introduction of new technologies for treating patients. Our analysis focuses broadly on the role of treated prevalence and spending per treated case in accounting for the rise in Medicare beneficiaries’ total health care spending, regardless of the source of payment. We do not attempt to identify the relative importance of all of the factors listed above.
Metabolic syndrome. Using metabolic syndrome as a case study, we examine two key areas related to the rise in prevalence: the rise in obesity, and more-aggressive use of medications to treat asymptomatic or mildly symptomatic patients. Metabolic syndrome is characterized by the presence of three of the following five conditions: abnormal levels of glucose, low high-density lipoprotein (HDL) cholesterol, elevated blood pressure, high triglyceride levels, and abdominal obesity.11 People with metabolic syndrome are at increased risk of carotid and coronary disease.12 Treatment includes changes in lifestyle such as weight management through increased physical activity and improved nutrition. However, if these prove ineffective, medications such as diuretics, statins, and antidiabetic medications may be prescribed.
Study Data And Methods
Data for the study were drawn from three sources: the 1987 National Medical Expenditure Survey (NMES), the 2002 Medical Expenditure Panel Survey (MEPS), and the National Health and Nutrition Examination Survey (NHANES III for 1988–1994 and NHANES 1999–2002). All are nationally representative surveys of the noninstitutionalized population.13 All estimates from each data set used each survey’s sampling weights (using the svymean command in STATA version 8) to produce nationally representative estimates.
NMES and MEPS data. We used NMES and MEPS data to examine changes in treated disease prevalence and spending. Both surveys contain detailed information on self-reported medical conditions, health insurance status, patient demographics, health care spending (excluding nursing home and other institutional care), and use of medical care services. We included any respondent reporting Medicare as a source of coverage (regardless of age) enrolled in Medicare for six months or more.14 Key results were similar when we restricted the analysis to elderly or nonelderly beneficiaries. We measured total annual spending for all sources of payment, including Medicare, private insurance, and out-of-pocket spending.
We identified respondents’ medical conditions using the Clinical Classifications System (CCS) groups developed by the Agency for Healthcare Research and Quality (AHRQ).15 The medical conditions and procedures associated with a health care visit were collected through the MEPS Household Component (MEPS-HC). The medical condition responsible for the visit was recorded by the MEPS interviewer as verbatim text and then coded by professional coders to fully specified International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes, including medical condition and V codes (a supplementary classification of factors influencing health status and contact with health services). We counted the number of treated conditions based on a list of 259 CCS conditions in 1987, 1997, and 2002, following our previous work in calculating condition-specific health care spending and attributing changes in total spending to each condition.16 We adjusted the 1987 spending data from charges to payments using the methods detailed by AHRQ.17 Costs were stated in terms of 2002 dollars using the GDP deflator.18
NHANES data. We used data from NHANES to examine the clinical and treated prevalence of the conditions that the metabolic syndrome comprises. These data permit standardized clinical measurement of height and weight, blood pressure, cholesterol, fasting glucose levels, and triglyceride levels across time. Using examination and laboratory files from NHANES, we estimated the number of Medicare beneficiaries with metabolic syndrome, as defined by the presence of three of the following five conditions: (1) abdominal obesity, defined as waist circumference exceeding 40 inches in men and 35 inches in women; (2) fasting plasma triglycerides that exceed 150 mg/dl; (3) HDL cholesterol under 40 mg/dL in men and under 50 mg/dL in women; (4) blood pressure of 130/85 mm Hg or higher; and (5) fasting blood glucose of 110 mg/dL or higher.
We also examined the number of beneficiaries under treatment for these conditions—for example, those reporting the use of an antihypertensive, antidiabetic (insulin or oral agents), or statins, among other therapeutic interventions. We tabulated the number of beneficiaries with metabolic syndrome who were treated for zero conditions, one condition, two conditions, and three or more conditions.
Study Results
Total spending by condition. Exhibit 1 presents trends in total spending by medical condition. The top ten conditions accounted for two-thirds of the growth in spending incurred by Medicare beneficiaries between 1987 and 2002. The three conditions associated with metabolic syndrome—hypertension, diabetes, and hyperlipidemia—collectively accounted for 16.1 percent of the rise in spending.19
Following earlier work, we also decomposed the rise in spending into the proportions attributable to changes in treated prevalence, spending, and enrollment.20 Controlling for enrollment, the rise in treated prevalence accounted for most of the rise in spending for six of the top ten conditions. These results are similar to those found among all adults regardless of source of coverage.21
Since the treated prevalence of heart disease in the Medicare population was relatively constant over time (27.0 percent in 1987, compared with 27.8 percent in 2002), most of the spending increase reported in Exhibit 1 is attributable to an increase in spending per treated case. In contrast, the treated prevalence of mental disorders increased sharply, rising from 7.9 percent in 1987 to 19.0 percent in 2002, and so spending increases are partly attributable to growth in the number of treated cases.
Age-adjusted treated prevalence. Exhibit 2 reports changes in beneficiaries’ age-adjusted treated disease prevalence during 1987–2002. These changes provide insight into the factors accounting for the rise in spending by medical condition shown in Exhibit 1. Sizable increases in treatment rates were observed for several conditions, including hyperlipidemia (twenty percentage points) and mental disorders (eleven percentage points). The substantial growth in treated prevalence of mental disorders reflects, in part, the introduction of new therapeutic treatments such as selective serotonin reuptake inhibitors (SSRIs). Increases in diabetes and obesity levels might also have played a role. Both conditions are associated with depression. Indeed, diabetics are twice as likely as nondiabetics to be depressed.22
Number of conditions per beneficiary. The number of medical conditions treated per Medicare beneficiary has risen sharply over time (Exhibit 3). In 1987, 31 percent of Medicare beneficiaries received treatment for five or more conditions.23 This group accounted for about half of total spending. Ten years later, nearly 40 percent of beneficiaries were treated for five or more conditions, accounting for 65 percent of overall spending. And just five years later, more than half of all Medicare beneficiaries were treated for five or more conditions, accounting for three-fourths of total spending. Virtually all of the spending growth since 1987 can be traced to patients treated for five or more conditions. And in 2002, 92.9 percent of health care spending was incurred by beneficiaries with three or more conditions during the year.
Explaining the trends. What accounts for these trends? Clearly, increases in obesity levels play a role. Many obese people have multiple morbidities such as hyperlipidemia, diabetes, and hypertension.24 The share of obese Medicare beneficiaries in the NMES/MEPS data sets increased from 9.4 percent in 1987 to 22.5 percent in 2002. Data from NHANES III and NHANES 1999–2002 indicate that the share of beneficiaries who are obese increased from 21.7 percent in 1988–94 to 29.5 percent in 1999–2002.25
Obesity. Overall, the prevalence of obesity among Medicare beneficiaries has doubled since 1987, but the share of spending incurred by obese beneficiaries has almost tripled—from 9.4 percent to nearly 25 percent of total spending (Exhibit 4). Thus, a rise in the share of obese Medicare beneficiaries combined with a higher share treated for five or more conditions over time accounts for about fifteen percentage points of the rise in spending.26
The share of normal-weight beneficiaries (as defined by body mass index, or BMI) treated for five or more conditions increased from 11.5 percent of all beneficiaries in 1987 to 16 percent in 2002, even as the overall share of the normal-weight group declined. The overall share of spending associated with this group (normal weight treated for five or more conditions) increased from 19.6 percent in 1987 to 24.1 percent in 2002. Treatment for hyperlipidemia, mental disorders, and osteoporosis and other bone disorders accounted for the largest increment in treated prevalence among normal-weight beneficiaries treated for five or more conditions (tabulations not shown). Similar trends were observed among overweight beneficiaries (Exhibit 4).
Other factors. Although obesity clearly plays a role in the growth of treated prevalence, increases in the share of nonobese beneficiaries treated for five or more medical conditions show that there are other factors at work. One explanation is that physicians are more aggressively treating healthier Medicare beneficiaries over time.27 To document this trend, we tabulated trends in self-reported health status using the 1987 NMES and 2002 MEPS among Medicare beneficiaries treated for five or more conditions over time. The results (not shown) reveal that 33 percent of those treated for five or more conditions in 1987 reported being in excellent or good health. By 2002, however, nearly 60 percent of Medicare beneficiaries treated for five or more conditions reported being in excellent or good health. Thus, either treatment is diffusing to healthier patients, treatments are improving health outcomes, or both are occurring. Physicians might also be more aggressively treating Medicare beneficiaries for asymptomatic or mildly symptomatic conditions over time. Metabolic syndrome provides an interesting case study to examine these trends.
Metabolic syndrome. The data presented in Exhibit 5 provide new information concerning the prevalence of metabolic syndrome among Medicare beneficiaries as well as the intensity of treatment. Using data from the NHANES III and NHANES 1999–2002, we estimated that nearly half of all beneficiaries meet the clinical criteria for metabolic syndrome.
The NHANES III data indicate that 43 percent of those with metabolic syndrome received no treatment for diabetes, hypertension, or hyperlipidemia during 1988–94. The most recent estimates from NHANES (1999–2002) reveal that the share of patients under medical treatment has increased 11.5 percentage points (p < .05), so that only 32 percent of Medicare beneficiaries with metabolic syndrome were not receiving some form of treatment during those years.
Another notable finding is the sharp rise in the share of beneficiaries receiving treatment for two or three of the conditions (diabetes, hypertension, or hyperlipidemia) associated with metabolic syndrome. The share of patients treated for two conditions more than doubled between the two most recent NHANES survey periods (p < .05). Moreover, the share of clinically indicated metabolic syndrome patients under treatment for all three conditions increased from virtually zero to nearly 6 percent (p < .05) between the two survey periods.
The largest change concerns the medical management of patients with low HDL cholesterol, with the percentage of metabolic syndrome–indicated patients treated rising from 9.7 percent to 27.5 percent between the two NHANES surveys (p < .05). The share of patients receiving an antidiabetic or antihypertensive medication increased as well (p < .10).
The trends depicted in Exhibit 5 must be interpreted with caution because the proportion of beneficiaries meeting the clinical criteria for metabolic syndrome will depend to some degree on treatment intensity. For example, an obese beneficiary with fasting blood glucose greater or equal to 110 mg/dL who is successfully treated for high blood pressure and does not have any of the other conditions defining metabolic syndrome would not be counted as having metabolic syndrome according to our methods. Our prevalence estimates in the top row of the exhibit capture beneficiaries who are untreated as well as those who are unresponsive to treatment, perhaps because of noncompliance with medication regimens. Although measuring the “true” prevalence of metabolic syndrome is inherently difficult, data reported elsewhere indicate that the trend in its treatment is the same regardless of whether one defines clinical prevalence based on both treatment patterns and laboratory values or on laboratory values alone (as we have done): The share of prevalent cases under treatment is increasing.28
Discussion
Our analysis provides new information on the factors driving the growth in Medicare beneficiaries’ health care costs over time. Virtually all of the growth in spending from 1987 to 2002 can be traced to the twenty-percentage-point increase in the share of Medicare patients receiving medical treatment for five or more conditions during a year. The factors responsible for these trends are likely to continue, leading to further increases in spending and intensifying the financial pressures on Medicare.29
Our findings raise important issues concerning the clinical management of Medicare beneficiaries with multiple chronic diseases. More than half of all beneficiaries report receiving medical treatment for five or more conditions during a year. The results highlight the need to develop models and standards of care designed to address the clinical problems facing patients with multiple complex comorbidities. Medicare, the Department of Veterans Affairs (VA), and other payers have increased their efforts to more effectively manage chronic disease, attempting to increase the share of patients receiving all clinically recommended services and improving the value (clinical benefits per dollar spent) of care provided. However, Medicare continues to operate under a fee-for-service model, which complicates the adoption of chronic care treatment models.30 Efforts to control the growth in Medicare spending using lifestyle modification strategies and care coordination will require changes in the way Medicare pays for services and interacts with providers. Another implication concerns the critical need to track the future benefits associated with the more-aggressive preventive treatment of the conditions associated with metabolic syndrome.
Recent data show that more-aggressive use of antihypertensives has been associated with a reduction in the share of adults meeting the clinical standards for hypertension, particularly among the obese.31 At least in the short run, this change in the pattern of clinical care has led to increased spending.
A key issue is whether reductions in these risk factors will translate into reductions in overall cardiovascular disease mortality.32 To the extent that they do, spending may continue to rise, as increases in longevity for people with chronic conditions prolong the period over which they incur high costs year in and year out. Historically, this has not been the case, but mortality reductions resulting from improvements in chronic care may improve health (that is, the “compression of morbidity”) and increase spending simultaneously.33
Ken Thorpe acknowledges support from the Pharmaceutical Research and Manufacturers of America (PhRMA), which assumed no role in the design of the study, its analysis, or its conclusions. The authors appreciate thoughtful comments from two anonymous reviewers and the help of Peter Joski in programming and research assistance. All errors in the analysis are the sole responsibility of the authors.
NOTES
1. Data are derived from the Congressional Budget Office, “The Long-Term Budget Outlook,” December 2005, http://www.cbo.gov/ftpdocs/69xx/doc6982/12-15-LongTermOutlook.pdf (accessed 10 August 2006). This assumes that per capita Medicare spending rises 2.5 percentage points higher than per capita GDP. Over the past thirty-four years, per capita Medicare spending has increased 2.9 percentage points faster than per capita GDP.
2. The paper uses the shorthand “spending among Medicare beneficiaries” with the recognition that we are tracking total health care spending linked to Medicare beneficiaries regardless of the source of payment (out of pocket, Medicaid, supplemental coverage). We also examined trends in spending flowing only through the Medicare program as well; that analysis produced results similar to those presented in this paper.
3. The annual National Health Expenditure Accounts (NHEA) reports created by the actuaries at the Centers for Medicare and Medicaid Services represent the best example of this type of study by provider. Their most recent estimates can be found at C. Smith et al., “National Health Spending in 2004: Recent Slowdown Led by Prescription Drug Spending,” Health Affairs 25, no. 1 (2006): 186–196. The CBO also decomposes the growth in spending into changes in enrollment and changes in spending per enrollee. The CBO estimates that over the past thirty-four years, growth in spending per beneficiary has accounted for approximately 82 percent of the overall rise in federal Medicare spending. CBO, “The Long-Term Budget Outlook,” Box 1-3 (p. 6). However, other studies conducted by the CMS track total health care spending among Medicare beneficiaries regardless of the source of payment. Estimates that use the Medicare Current Beneficiary Survey (MCBS) can track total spending. Moreover, estimates of spending from the CMS by age (for instance, those age sixty-five and older) do track total spending. See, for example, CMS, “Age Estimates in the National Health Accounts: Definitions, Sources, and Methods,” http://www.cms.hhs.gov/NationalHealthExpendData/downloads/age-methodology.pdf (accessed 7 May 2006).
4. The estimate that the Medicare program directly finances 50 percent of total Medicare spending is based on a study by AARP. See C. Caplan, “What Share of Beneficiaries’ Total Health Care Costs Does Medicare Pay?” September 2002, http://assets.aarp.org/rgcenter/health/dd78_costs.pdf (accessed 13 July 2006).
5. This collection of studies can be found as part of a 2005 Health Affairs Web-Exclusive collection, “Health and Costs of the Future Elderly,” available online at http://content.healthaffairs.org/cgi/content/full/hlthaff.w5.r1/DC2 (accessed 13 July 2006).
6. M.E. Chernew et al., “Disability and Health Care Spending among Medicare Beneficiaries,” Health Affairs 24 (2005): W5-R42–W5-R52 (published online 26 September 2006; 10.1377/hlthaff.W5 .R42).
7. J. Lubitz et al., “Health, Life Expectancy, and Health Care Spending among the Elderly,” New England Journal of Medicine 349, no. 11 (2003): 1048–1055.
8. Chernew et al., “Disability and Health Care Spending.”
9. K.E. Thorpe et al., “The Rising Prevalence of Treated Disease: Effects on Private Health Insurance Spending,” Health Affairs 24 (2005): w317–w325 (published online 27 June 2006; 10.1377/ hlthaff.w5.317.
10. Each of the factors listed would result in a rise in treated prevalence. For instance, as mortality rates decline, Medicare enrollment per year would rise. Although this would increase spending annual spending, work by Lubitz and colleagues show that total lifetime spending (after age age) is the same among those with and without a limitation on an activity of daily living. Lubitz et al., “Health, Life Expectancy, and Health Care Spending.” Also see G.F. Joyce et al., “The Lifetime Burden of Chronic Disease among the Elderly,” Health Affairs 24 (2005): W5-R18–W5-R29 (published online 26 September 2006; 10.1377/hlthaff.W5.R18). Several studies have examined the role of technology and innovation as a major factor accounting for rising spending. For a recent example of this literature, see A.A. Okunade and V.N.R. Murthy, “Technology as a ‘Major Driver’ of Health Care Costs: A Cointegration Analysis of the Newhouse Conjecture,” Journal of Health Economics 21, no. 1 (2002): 147–159.
11. E.S. Ford, W. Giles, and W. Dietz, “Prevalence of the Metabolic Syndrome among U.S. Adults: Findings from the Third National Health and Nutrition Examination Survey,” Journal of the American Medical Association 287, no. 3 (2002): 356–359.
12. C.M. Alexander et al., “NCEP-Defined Metabolic Syndrome, Diabetes, and Prevalence of Coronary Heart Disease among NHANES III Participants Age Fifty Years and Older,” Diabetes 52, no. 5 (2003): 1210–1214.
13. Since MEPS and NHANES do not survey institutionalized patients (such as nursing home patients), their spending is excluded from the analysis. Yet because spending on nursing home care among those age sixty-five and older rose at virtually the same average rate as overall health care spending (7.9 percent) relative to personal health care spending in that age group, the omission is not likely to affect the results. Data are from the CMS data set, as in Note 3. Detailed descriptions of the sample design and data collection methods for each survey are available on both the MEPS and NHANES Web sites, http://www.meps.ahrq.gov and http://www.cdc.gov/nchs/nhanes.htm.
14. Our results are not sensitive to this choice, although it does result in lower spending among Medicare beneficiaries than in other estimates. Total reported spending, however, is lower than found in the NHEA and the MCBS (see Note 3). NMES and MEPS do not include spending on institutionalized patients, which results in lower total spending per year in our analysis.
15. An overview of the CCS approach and the codes may be found at Healthcare Cost and Utilization Project, “Clinical Classifications Software (CCS) for ICD-9-CM Fact Sheet,” 22 November 2005, http://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccsfactsheet.jsp (accessed 13 July 2006).
16. Thorpe et al., “The Rising Prevalence.”
17. S.H. Zuvekas and J.W. Cohen, “A Guide to Comparing Health Care Expenditures in the 1996 MEPS to the 1987 NMES,” Inquiry 39, no. 1 (2002): 76–86.
18. The values may be found at National Aeronautics and Space Administration, “Gross Domestic Product Deflator Inflation Calculator,” 21 January 2005, http://www1.jsc.nasa.gov/bu2/inflateGDP.html (accessed 13 July 2006).
19. We also conducted the analysis limiting Medicare beneficiaries to those age sixty-five and older. The results presented in this paper are not sensitive to the decision to include all Medicare beneficiaries.
20. Thorpe et al., “The Rising Prevalence.” Our decomposition of total spending increases into components attributable to increases in treated prevalence (TPV) and costs per case (CPC) is based on the following formula:
(CPC2002 – CPC1987) × TPV1987+
(TPV2002 – TPV1987) × CPC1987+
(CPC2002 – CPC1987) × (TPV2002 – TPV1987)
We attributed the third term to changes in cost per case and treated prevalence based on the relative magnitudes of the first two terms. So, for example, we calculated the impact of the change in costs per case on the change in per capita costs by adding the first term to the product of (1) the third term and (2) the ratio of the first term to the sum of the first and second terms. The impact of the change in treated prevalence is defined analogously. The estimates can also be presented as ranges, where the lower bound of the increase in per capita costs attributable to the change in treated prevalence is the second term in the expression above and the upper bound is the sum of the second and third terms. The proportion of the increase attributable to the change in costs per case ranges from the first term (lower bound) to the sum of the first and third terms (upper bound). The estimates presented in the exhibit are the midpoints of these ranges.
21. K.E. Thorpe, C.S. Florence, and P. Joski, “Which Medical Conditions Account for the Rise in Health Care Spending?” Health Affairs 23 (2004): w437–w445 (published online 25 August 2004; 10.1377/hlthaff.w4.437).
22. R.J. Anderson et al., “The Prevalence of Comorbid Depression in Adults with Diabetes: A Meta-Analysis,” Diabetes Care 24, no. 6 (2001): 1069–1078.
23. Medical treatment means the use of any medical service, such as a physician visit, drug, hospitalization, or other ambulatory care visit associated with the medical condition. For each service, beneficiaries were asked about the medical condition leading to the use of service.
24. E. Ferrannini et al., “Hyperinsulinaemia: The Key Feature of a Cardiovascular and Metabolic Syndrome,” Diabetologia 34, no. 6 (1991): 416–422.
25. NMES and MEPS rely on respondents’ reporting their height and weight to those conducting the survey. In contrast, NHANES relies on actual clinical measures (height, weight, blood pressure) collected by physicians and other health care providers.
26. We also followed the methods outlined in K.E. Thorpe et al., “The Impact of Obesity on Rising Medical Spending,” Health Affairs 23 (2004): w480–w486 (published online 20 October 2004; 10.1377/hlthaff.w4.480), to measure the share of the rise in spending that is linked to a rise in obesity. The two-part regression model included the same covariates in this work, but it also interacted obesity with disability status. This method indicates that the rise in obesity among Medicare beneficiaries accounted for approximately 20 percent of the rise in spending during the period. These results are similar, although slightly higher, than the unadjusted accounting tabulation reported in the current paper.
27. The authors thank an anonymous reviewer of this paper for making this point.
28. E.W. Gregg et al., “Secular Trends in Cardiovascular Disease Risk Factors According to Body Mass Index in U.S. Adults,” Journal of the American Medical Association 293, no. 15 (2005): 1868–1874.
29. Understanding the contribution of each of these factors to increases in treated prevalence is the subject of ongoing work. See D.H. Howard, S. Busch, and K.E. Thorpe, “Understanding Recent Trends in Disease Treatment Rates: Are We Getting Sicker or Getting Screened?” (Working paper, Emory University, 2006).
30. R. Berenson and J. Horvath, “Confronting the Barriers to Chronic Care Management in Medicare,” Health Affairs 22 (2003): w37–w53 (published online 22 January 2003; 10.1377/hlthaff .w3.37).
31. See, for example, Gregg et al., “Secular Trends.”
32. There is some evidence that the rise in spending on antihypertensive drugs has reduced the number of premature deaths from cardiovascular disease. This line of research estimates a health benefit to cost ratio of approximately 11:1. This literature indicates the higher spending may be cost-effective. See, for example, G. Long et al., “The Impact of Antihypertensive Drugs on the Number and Risk of Death, Stroke and Myocardial Infarction in the United States,” NBER Working Paper no. 12096 (Cambridge, Mass.: National Bureau of Economic Research, March 2006).
33. P. Zweifel, S. Felder, and M. Meiers, “Ageing of Population and Health Care Expenditures: A Red Herring?” Health Economics 8, no. 6 (1999): 485–496; and J.F. Fries, “Measuring and Monitoring Success in Compressing Morbidity,” Annals of Internal Medicine 139, no. 5, Part 2 (2003): 455–459.
Ken Thorpe (kthorpe{at}sph.emory.edu) is the Robert W. Woodruff Professor and chair of the Department of Health Policy and Management, Rollins School of Public Health, Emory University, in Atlanta, Georgia. David Howard is an assistant professor in that department.
Read a related article and response from the authors.
DOI: 10.1377/hlthaff.25.w378
©2006 Project HOPE–The People-to-People Health
Foundation, Inc.
|
