pi is 0002934311002956

o

CLINICAL RESEARCH STUDY

The Impact of the Aging Population on Coronary HeartDisease in the United StatesMichelle C. Odden, PhD,a Pamela G. Coxson, PhD,a Andrew Moran, MD, MPH,b,c James M. Lightwood, PhD,d

Lee Goldman, MD, MPH,b Kirsten Bibbins-Domingo, PhD, MDa,e

aDepartment of Medicine, University of California, San Francisco; bCollege of Physicians and Surgeons, Columbia University, NewYork, NY; cDivision of General Medicine, Columbia University Medical Center, New York, NY; dDepartment of Pharmacy, University
f California, San Francisco; eDepartment of Epidemiology and Biostatistics, University of California, San Francisco.
E-mail address

0002-9343/$ -see fdoi:10.1016/j.amjm

ABSTRACT

BACKGROUND: The demographic shift toward an older population in the United States will result in ahigher burden of coronary heart disease, but the increase has not been quantified in detail. We sought toestimate the impact of the aging US population on coronary heart disease.METHODS: We used the Coronary Heart Disease Policy Model, a Markov model of the US populationbetween 35 and 84 years of age, and US Census projections to model the age structure of the populationbetween 2010 and 2040.RESULTS: Assuming no substantive changes in risks factors or treatments, incident coronary heart diseaseis projected to increase by approximately 26%, from 981,000 in 2010 to 1,234,000 in 2040, and prevalentcoronary heart disease by 47%, from 11.7 million to 17.3 million. Mortality will be affected strongly bythe aging population; annual coronary heart disease deaths are projected to increase by 56% over the next30 years, from 392,000 to 610,000. Coronary heart disease-related health care costs are projected to riseby 41% from $126.2 billion in 2010 to $177.5 billion in 2040 in the United States. It may be possible tooffset the increase in disease burden through achievement of Healthy People 2010/2020 objectives orinterventions that substantially reduce obesity, blood pressure, or cholesterol levels in the population.CONCLUSIONS: Without considerable changes in risk factors or treatments, the aging of the US populationwill result in a sizeable increase in coronary heart disease incidence, prevalence, mortality, and costs.Health care stakeholders need to plan for the future age-related health care demands of coronary heartdisease.© 2011 Elsevier Inc. All rights reserved. • The American Journal of Medicine (2011) 124, 827-833

KEYWORDS: Aging; Coronary heart disease; Forecasting; Markov chains

nmdc

Funding: Supported in part by the Swanson Family Fund, Tempe,Ariz, and a grant-in-aid from the American Heart Association WesternStates Affiliate, Burlingame, Calif. (09GRNT2060096). Dr. Odden is sup-ported by a Ruth L. Kirschstein National Research Service Award(T32HP19025). Dr. Moran is supported by an NIH/NHLBI MentoredCareer Development Award (K08HL089675) and an Empire Clinical Re-search Program award from New York State.

Conflict of Interest: None.Authorship: All authors were involved in the design of the study,

interpretation of data, and the writing of the manuscript. All authors haveapproved the final version.

Requests for reprints should be addressed to Michelle C. Odden, PhD,Department of Medicine, University of California, San Francisco, Box1211, San Francisco, CA 94143-1211.

: [email protected]

ront matter © 2011 Elsevier Inc. All rights reserved.ed.2011.04.010

The US population is aging. The combination of the ele-vated birth rate during the “baby boom” and increasing lifeexpectancy will result in a doubling of the population aged65 years and older from year 2010 to 2040,1 with theumber of persons aged 65 and older increasing from 40illion in 2010 to 81 million in 2040.1,2 Coronary heart

isease is strongly associated with age and is the leadingause of death in the US.3 Because coronary heart disease

disproportionately affects the elderly, this demographicshift will result in a considerable increase in burden ofdisease in the US population. The first of the baby boomerswill enter the Medicare population in 2011, so up-to-dateprojections of coronary heart disease burden are of specialimportance for public health planning. Quantitative esti-
mates of the magnitude and distribution of disease are
mailto:[email protected]

rtap

dw

nmrdstctdT

spdct

hvmS

C

f the

828 The American Journal of Medicine, Vol 124, No 9, September 2011

essential for appropriate preparedness strategies at both theprovider and system levels, and in order to optimally targetpreventive efforts.

We aimed to provide a detailed projection of the impactof the aging US population on coronary heart disease bur-den, based on a computer modelthat is representative of the USpopulation. The Coronary HeartDisease Policy Model is a simula-tion model that can account for therelationship of risk factors andchanging demographics on the in-cidence, prevalence, and cost ofcoronary heart disease.4 Since itsinception, the model has beenused to explain the decline in cor-onary heart disease mortality inthe United States5; to estimate thegains in life expectancy achiev-able by various risk factor inter-ventions6; to assess the cost-effec-tiveness of interventions to reducecholesterol,7,8 blood pressure,9,10

and smoking11; and to analyze thepublic health impact of improve-ments in medical care.12,13 Mostecently, the model has assessedhe impact of current adolescent obesity on future rates ofdult coronary heart disease,14 public policies to reduceassive tobacco exposure on coronary heart disease rates,15

clinical guidelines for the use of statins,16 and the impact ofietary salt reductions.17 In 1987, the model’s projectionsere published through 2010.4 In this analysis, we provide

updated estimates of the impact of the aging US populationon the incidence, prevalence, mortality, and cost of coronaryheart disease from 2010 to 2040. In addition, we estimatethe extent to which achievement of national goals for riskfactor improvement could offset the age-related increase incoronary heart disease burden.

METHODS

The Coronary Heart Disease Policy ModelThe Coronary Heart Disease Policy Model is a validatedstate-transition (Markov) model of the incidence, preva-lence, mortality, and cost of coronary heart disease in USresidents 35 to 84 years of age.4 The model has 3 compo-ents. First, the demographic-epidemiologic submodel esti-ates the incidence of coronary heart disease (cardiac ar-

est, myocardial infarction, angina, or coronary heartisease death) and death from other causes, based on age,ex, systolic blood pressure, smoking, high-density lipopro-ein (HDL) cholesterol, and low-density lipoprotein (LDL)holesterol, diabetes, and body mass index (BMI). Second,he bridge submodel characterizes the initial coronary heartisease event and related events in the subsequent 30 days.

CLINICAL SIGNIF

● Absolute coronadence is projectprevalence by 4and costs by 41%of the US popul2040.

● Achievement of2020 major objcontrol could offprojected increa

● A 10% reductionadults also coulate the impact o

hird, the disease history submodel predicts the number of

ubsequent coronary heart disease events, revascularizationrocedures, and deaths among subjects with coronary heartisease, stratified according to age, sex, and history oforonary heart disease events. Modifiable components ofhe model include: population distributions, risk factor lev-

els, risk factor coefficients, eventrates, case fatality rates, costs, andquality of life adjustments.4 Amore detailed description of themodel is given in the Appendix(available online).

Data SourcesIn the present study, we used USCensus data and 2008 projectionsto estimate the demographic shiftin the age structure of the popula-tion.2 Other data sources for themodel include the National Centerfor Health Statistics mortalitydata,18 the Framingham HeartStudy for the association of riskfactors with coronary heart dis-ease,19, 20 and Olmsted Countydata for the incidence of myocar-dial infarction and cardiac ar-rest.21 The prevalence of coronary

eart disease was estimated from the National Health Inter-iew Survey,22 and the risk factor distributions were esti-ated from the National Health and Nutrition Examinationurvey.23 Data from the National Hospital Discharge Sur-

vey were used to calculate rates of myocardial infarction,hospitalization due to cardiac arrest, revascularization pro-cedures, and associated case fatalities.24 We estimated cor-onary heart disease deaths, prehospital deaths caused bycardiac arrest, and non-coronary heart disease deaths basedon data from the US Vital Statistics.18 Total health carecosts and 30-day survival rates were based on data fromMedicare.25 We estimated coronary heart disease costsbased on California data,26 deflated by using cost-to-chargeratios,27 and the ratio of the US national average costs to the

alifornia average.28 We then inflated to 2010 dollars byusing the Bureau of Labor Statistics Consumer Price Indexfor Medical Care Costs.29

SimulationsWe ran the main simulation from years 2010 to 2040 toreflect the changing age demographics of the US populationover these years. Age- and sex-specific risk factor distribu-tions, event rates, case fatality rates, and costs per eventwere held constant across the simulation period in order toisolate the effect of the aging population. We estimated theabsolute and relative annual excess (defined as increaseabove 2010 estimates) in coronary heart disease incidence(stable or unstable angina, myocardial infarction, cardiac

CE

art disease inci-increase by 26%,ortality by 56%,

ause of the agingbetween 2010 to

thy People 2010/s for risk factore majority of the

ody mass index instantially attenu-aging population.

ICAN

ry heed to7%, m

becation

Healectiveset thse.

in bd sub

arrest, and death), prevalence, mortality, and costs across 30

tltttwaFtctt

P

M

C

829Odden et al Heart Disease in the Aging US Population

years. We present the annual number of events by 10-yearage categories.

We next explored how changes in risk factors and treat-ments might alter the projected increase in coronary heartdisease incidence. We modeled the impact of achievingadherence to the major 2010/2020 Healthy People guide-lines for risk factors in 2010, including a 10% reduction inthe prevalence of hypertension, a 40% improvement in thecontrol of hypertension, a 10% reduction in the prevalenceof hypercholesterolemia, a 10% reduction in the mean levelof cholesterol, a 10% reduction in the prevalence of obesity,a reduction of the prevalence of smoking to 12%, a 10%reduction in environmental exposure to smoke, and a 10%reduction in the rate of developing incident diabetes.30 Forhe cholesterol interventions, we modified LDL cholesterolevels instead of total cholesterol in the model. We exploredhe impact of lowering LDL cholesterol and blood pressureo levels used in prevention guidelines.31 We also modeledhe impact a linear 10% increase in BMI over 30 years,hich is consistent with recently published trends,32 as well

s a 10% decrease in BMI over the simulation period.inally, we modeled the impact of a hypothetical interven-

ion that would reduce the in-hospital myocardial infarctionase fatality rate by 25% to 50%, based on improvements inhe survival of patients hospitalized for myocardial infarc-ion in the past 30 years.33,34

RESULTSIn 2010, an estimated 156 million adults 35 to 84 years ofage will be living in the US. This population is expected toincrease by 28% to 200 million in 2040. The 65- to 84-year-old population is expected to increase by 89% percentover this period because of the aging of baby boomers andtheir longer life expectancy. The increase in coronary heartdisease will reflect this population growth: incidence, prev-alence, mortality, and cost are all expected to increase

Table 1 Projected Coronary Heart Disease Incidence, Prevalen

Year

Population 2010

Incidence Total 981,00035-64 y 658,00065-84 y 323,000

revalence Total 11,744,00035-64 y 5,662,00065-84 y 6,082,000

ortality Total 392,00035-64 y 106,00065-84 y 286,000

ost (in Billions) Total $126.235-64 y $69.765-84 y $56.5

monotonically over the next 30 years.

The number of incident coronary heart disease cases isprojected to increase in parallel with the overall populationgrowth by approximately 26% from 981,000 in 2010 to1,234,000 in 2040 (Table 1). The number of incident casesis projected to remain fairly constant in persons under 65years of age because of the relatively smaller populationgrowth in this subgroup. The largest increase in incidentcoronary heart disease events will be expected in persons 65years of age and older, and the number of cases in 75- to84-year-olds will double in the next 30 years. Without achange in the risk factor distributions, the incidence rate isprojected to remain fairly constant, at 680 per 100,000person-years in 2010 as compared with 710 per 100,000person-years in 2040.

The number of prevalent coronary heart disease casesalso is expected to increase as a result of the change inpopulation demographics: the number of persons with prev-alent coronary heart disease is expected to grow by 47%from 11.7 million to 17.3 million in 2040 (Table 1). As withincidence, the greatest growth in prevalence is expected tobe in adults over 65 years of age (Figure 1). Coronary heartdisease mortality will be impacted most strongly by the shiftin age demographics, and coronary heart disease deaths areexpected to increase by over 50% over the next 30 years(Figure 2). The number of deaths in adults 65 to 84 years ofage is projected to increase by two thirds because of thelarge increase in this segment of the population. The greaterincrease in coronary heart disease deaths compared withincidence or prevalence is because death disproportionatelyoccurs at older ages; the increase in the number of olderadults will affect mortality rates more strongly than it willaffect incidence or prevalence rates.

US coronary heart disease-related health care costs inadults 35 to 84 years of age are projected to rise by 41%over the next 30 years, from $126.2 billion in 2010 to$177.5 billion in 2040, based on year 2010 dollars. The

rtality, and Cost in US Adults 35 to 84 Years of Age

2020 2030 2040

1,128,000 1,212,000 1,234,000666,000 637,000 669,000462,000 575,000 565,000

14,382,000 16,699,000 17,322,0006,949,000 7,107,000 7,634,0007,433,000 9,592,000 9,688,000

473,000 586,000 610,000122,000 118,000 125,000351,000 468,000 485,000

$149.5 $170.7 $177.5$79.5 $80.5 $86.4$70.0 $90.2 $91.1

ce, Mo

annual cost of coronary heart disease per working-aged

behvpc

sa


adult (20-64 years of age) is projected to rise by 24%, from$680 in 2010 to $840 in 2040. The proportion of costsassociated with acute coronary heart disease care is pro-jected to increase slightly over the next 30 years comparedwith costs because of chronic disease follow-up, from 48%in 2010 to 50% in 2040.

Effects of Potential Changes in Risk Factorsor TreatmentImmediate achievement of Healthy People 2010/2020 ob-jectives for blood pressure, lipids, obesity, smoking, anddiabetes could offset the increase in disease attributable tothe aging population by over 70% in 2040 (Figure 3). Inaddition, a trend of a reduction in BMI by 10% over the next30 years could offset the projected increase in incidence byapproximately 80%. In contrast, a trend of increase in the

Figure 1 Absolute coronary heart2010 to 2040.

Figure 2 Absolute number of coronary heart disease deaths
by decade of age, from 2010 to 2040.
average BMI of US adults by 10% over the next 30 yearswould add to the burden of coronary heart disease, andcould more than double the increase in incident cases by2040 (Figure 3). If all Americans 35 to 84 years of age hada systolic blood pressure no higher than 140 mm Hg, theprojected incidence of coronary heart disease would be1,130,000 new cases in 2040, or a 15% increase over 2010.Similarly, if all Americans had an LDL cholesterol level of�130 mg/dL, the projected incidence would be 1,151,000in 2040, or a 17% increase over 2010. A 25% to 50%reduction in the in-hospital case fatality for both first andrecurrent myocardial infarction in 2040 as compared with2010 was projected to reduce mortality in 2040 by only 4%to 8%.

DISCUSSIONUnprecedented growth in the 65 years of age and olderpopulation in the United States is expected over the comingdecades. This growth will result in a substantial increase incoronary heart disease incidence, prevalence, mortality, andcost. A quantitative estimate of the magnitude of this in-crease is essential information for stakeholders in the UShealth care system, including patients, providers, trainingorganizations, hospitals, health plans, governments, and in-dustry.35,36 Despite concern over the impact of the agingaby boomers on health care, few detailed quantitativestimates of the effect of the aging population on coronaryeart disease burden have been published. Our study pro-ides estimates that will inform researchers, providers, andolicy makers as they plan to meet future age-related healthare demands.

Our findings expand upon previous research that hashown strong growth in coronary heart disease prevalencend mortality over the next several decades37; Foot et al37

estimated a 128.5% increase in coronary heart disease mor-

prevalence by decade of age, from
disease
tality and a 93% increase in coronary heart disease preva-

a � 10


lence from 2000 to 2050. Our projections are slightly moreconservative, and we provide detailed epidemiologic esti-mates that account for the distribution of age, gender, andrisk factors in the US population. As noted in previousstudies, the aging population will impact the need for morefocused efforts on primary, secondary, and tertiary preven-tion, because optimal treatment can lead to longer andhigher quality of life in coronary heart disease patients.However, these preventive efforts also will increase theneed for providers and services. The increased burden ofdisease will require more resources in hospitals, physicians,medical groups, training institutions, and public health de-partments.37 Physicians and nonphysician clinicians are al-ready in short supply. For example, the Council on GraduateMedical Education (COGME) has determined that the de-mand for physicians is expected to grow more than thesupply over the next 10 years.35,38 Although some of thisburden will require more specialty providers to conductprocedure-based treatments, the majority of the burden islikely to fall on primary care physicians and geriatricianswho manage the chronic care of patients living with coro-nary heart disease.

Risk factors and medical advances to prevent and treatcoronary heart disease will certainly change in the next 30years, and these changes are difficult to predict. Primaryprevention gains would need to be substantial to offset theimpact of the aging population. The achievement of opti-mistic, but feasible, improvements in risk factor control,consistent with Healthy People 2010/2020 objectives, couldoffset more than 70% of the increase in coronary heart

Figure 3 The projected impact of con coronary heart disease incidence,People 2010/2020 objectives includelow-density lipoprotein cholesterol, oindex increase and decrease refers to

disease incidence related to aging of the population. How-

ever, achievement of these goals would require these im-provements across several domains, including a reversal ofthe trend of weight gain and increasing diabetes in theUnited States. If the trend of increasing BMI continues atthe same rate observed over the past few decades, thegrowing burden of coronary heart disease would be evenworse. A previous study based on the model showed asubstantial impact of current adolescent obesity on the fu-ture burden of coronary heart disease; Bibbins-Domingo etal14 estimated an excess of at least 100,000 cases of coro-nary heart disease in 2035.14 Conversely, improvements inBMI could offset the increase in coronary heart disease; a10% reduction in BMI over the next 30 years would offsetapproximately 80% of the impact of the aging populationbecause of its beneficial effects on blood pressure, choles-terol, and diabetes. These simulations suggest that publichealth interventions to lower BMI through improved dietand physical activity should be a high priority. By compar-ison, a 25% to 50% reduction in in-hospital myocardialinfarction case fatality would have only a modest effect oncoronary heart disease, largely because the current survivalrate is approximately 90% for a hospitalized first myocar-dial infarction.24 Taken together, these analyses suggest thatsubstantial improvements from risk factor profiles would benecessary to offset the projected increase in coronary heartdisease.

This increased burden of disease will likely tax the al-ready struggling US health care system. Medical expendi-tures have increased at a high rate for many years, and thefuture impact of population aging on health care costs is

in coronary heart disease risk factors010 to 2040. Adherence to Healthy

iate improvements in blood pressure,, smoking, and diabetes. Body mass% trend from 2010 to 2040.

hangesfrom 2

immedbesity

frequently cited as a source of concern in the literature and

da

etA

eapregcsuttelirw

csnsaeoli

bn


media.36,39,40 Our cost projections related to coronary heartisease events are similar to those reported by Martini etl,36 who estimated a 44% increase in health care costs

because of heart and vascular conditions from 2000 to 2050.Notably, heart and vascular conditions account for moremedical care costs than any other category of disease.36

Population growth amongst working adults could help tooffset these costs slightly, through increased tax revenueand other financial contributions. Coronary heart diseasecosts per working adult aged 20 to 64 years are projected toincrease by 24% over the next 30 years. Economic growthwill only augment the increase in health care costs; previousstudies have demonstrated that economic expansion deter-mines the ceiling for health care expenditures, and demandfor services will continue to grow as patients desire andexpect longer and healthier lives.35

The model is tested and updated regularly to reflectchanges in risk factor distributions, population estimates,risk factor associations, event rates, case fatality rates, andcosts, although these statistics are only estimates of whatmay occur in the future. Projections from any forecastingmodel should be viewed with caution, because there willalways be unpredicted changes that can impact results. Theprevious projections from the Coronary Heart Disease Pol-icy Model published in 1987 underestimated the prevalenceof future coronary heart disease but overestimated incidenceand mortality,4 although the majority of the variation can bexplained by a change in disease definition and by reduc-ions in risk factors and advances in treatment (seeppendix online).Our projections have limitations that should be consid-

red when interpreting the findings. These estimates do notccount for changes in the racial/ethnic structure of theopulation; although this could impact our estimates, theelationship between age and risk of coronary heart diseasevents does not likely differ greatly across racial/ethnicroups. The timing and sustainability of the risk factorhanges that were modeled were optimistic and could varyubstantially in practice. Furthermore, there will likely benexpected changes in risk factor profiles and health careechnology that could affect our results. Improved therapieso treat coronary heart disease may decrease secondaryvents and mortality, but also increase the cost and preva-ence of coronary heart disease. Finally, the model does notnclude incident events for adults over 85 years of age, whoepresent a growing proportion of the population of adultsith coronary heart disease.In conclusion, the demographic shift towards an in-

reased number of older adults in the US will have aubstantial impact on coronary heart disease burden over theext 30 years. Stakeholders in the US health care systemhould prepare for this growth by ensuring that there aredequate resources to care for the high number of adultsxpected to be living with coronary heart disease, becauseptimal treatment can lead to longer and higher quality ofife in coronary heart disease patients. Future improvements
n primary and secondary prevention may attenuate this
urden of disease, although substantial changes would beecessary to offset the effect of the aging population.

References1. Centers for Disease Control and Prevention. Trends in aging—United

States and worldwide. MMWR Morb Mortal Wkly Rep. 2003;52:101-106.

2. US Census Bureau. U.S. interim projections by age, sex, race, andHispanic origin 2008. vol 2009, 2008.

3. Heron M, Hoyert DL, Murphy SL, et al. Deaths: final data for 2006.Natl Vital Stat Rep. 2009;57:1-134.

4. Weinstein MC, Coxson PG, Williams LW, et al. Forecasting coronaryheart disease incidence, mortality, and cost: the Coronary Heart Dis-ease Policy Model. Am J Public Health. 1987;77:1417-1426.

5. Goldman L. Cost-effectiveness perspectives in coronary heart disease.Am Heart J. 1990;119(3 part 2):733-739.

6. Tsevat J, Weinstein MC, Williams LW, et al. Expected gains in lifeexpectancy from various coronary heart disease risk factor modifica-tions. Circulation 1991;83:1194-1201.

7. Goldman L, Weinstein MC, Williams LW. Relative impact of targetedversus populationwide cholesterol interventions on the incidence ofcoronary heart disease. Projections of the Coronary Heart DiseasePolicy Model. Circulation 1989;80:254-260.

8. Prosser LA, Stinnett AA, Goldman PA, et al. Cost-effectiveness ofcholesterol-lowering therapies according to selected patient character-istics. Ann Intern Med. 2000;132:769-779.

9. Edelson JT, Weinstein MC, Tosteson AN, et al. Long-term cost-effectiveness of various initial monotherapies for mild to moderatehypertension. JAMA. 1990;263:407-413.

10. Phillips KA, Shlipak MG, Coxson P, et al. Health and economicbenefits of increased beta-blocker use following myocardial infarction.JAMA. 2000;284:2748-2754.

11. Tosteson AN, Weinstein MC, Williams LW, et al. Long-term impactof smoking cessation on the incidence of coronary heart disease. Am JPublic Health. 1990;80:1481-1486.

12. Goldman L, Phillips KA, Coxson P, et al. The effect of risk factorreductions between 1981 and 1990 on coronary heart disease inci-dence, prevalence, mortality and cost. J Am Coll Cardiol. 2001;38:1012-1017.

13. Tice JA, Ross E, Coxson PG, et al. Cost-effectiveness of vitamintherapy to lower plasma homocysteine levels for the prevention ofcoronary heart disease: effect of grain fortification and beyond. JAMA.2001;286:936-943.

14. Bibbins-Domingo K, Coxson P, Pletcher MJ, et al. Adolescent over-weight and future adult coronary heart disease. N Engl J Med. 2007;357:2371-2379.

15. Lightwood JM, Coxson PG, Bibbins-Domingo K, et al. Coronary heartdisease attributable to passive smoking: CHD Policy Model. Am JPrev Med. 2009;36:13-20.

16. Pletcher MJ, Lazar L, Bibbins-Domingo K, et al. Comparing impactand cost-effectiveness of primary prevention strategies for lipid-low-ering. Ann Intern Med. 2009;150:243-254.

17. Bibbins-Domingo K, Chertow GM, Coxson PG, et al. Projected effectof dietary salt reductions on future cardiovascular disease. N EnglJ Med. 2010;362:590-599.

18. Deaths for 358 selected causes, by 5-year age groups, race and sex:United States, 1999-2000. Atlanta, GA: National Center for HealthStatistics; 2007.

19. Framingham Heart Study CD-ROM. Washington, DC: Department ofHealth and Human Services; 2005.

20. Wilson PW, D’Agostino RB, Levy D, et al. Prediction of coronaryheart disease using risk factor categories. Circulation. 1998;97:1837-1847.

21. Roger VL, Jacobsen SJ, Weston SA, et al. Trends in the incidence andsurvival of patients with hospitalized myocardial infarction, Olmsted
County, Minnesota, 1979 to 1994. Ann Intern Med. 2002;136:341-348.

2

2

2

3

3


22. National Health Interview Survey datasets. Washtington, DC: Na-tional Center for Health Statistics; 2000.

23. National Health and Nutrition Examination Survey. Atlanta, GA:Centers for Disease Control and Prevention; 2008.

24. National Hospital Discharge Survey, 1996-2005. Atlanta, GA: Centersfor Disease Control and Prevention, National Center for Health Sta-tistics; 2006.

25. Centers for Medicare and Medicaid Services. Personal health carespending by type of spending, age group, and source of paymentdistribution, calendar year 1999. Woodland, MD: Centers for Medi-care and Medicaid Services; 2000.

6. State of California, Office of Statewide Health Planning and Devel-opment. Patient discharge data file documentation, January 1-Decem-ber 31, 2000, public version A-24. Sacramento, CA: Office of State-wide Health Planning and Development; 2001.

27. Office of Statewide Health Planning and Development. Hospital fi-nancial data for cost to charge ratio, CA inpatient discharge datahospital annual finanacial data, pivot profiles, 1999-2000. Sacra-mento, CA: California Office of Statewide Health Planning and De-velopment; 2008.

8. US Census Bureau. Statistical abstract of the United States. Averagecost to community hospitals per patient, by state (Table 204). Wash-ington, DC: Goverment Printing Office; 1998.

9. Bureau of Labor Statistics. Consumer price index. Washington, DC:Bureau of Labor Statistics; 2010.

0. US Department of Health and Human Services. Healthy People 2020.Washington, DC; 2010. Available at http://www.healthypeople.gov/2020/default.aspx. Accessed May 6, 2011.

1. Pearson TA, Blair SN, Daniels SR, et al. AHA Guidelines for PrimaryPrevention of Cardiovascular Disease and Stroke: 2002 Update: Con-
sensus Panel Guide to Comprehensive Risk Reduction for Adult Pa-
tients Without Coronary or Other Atherosclerotic Vascular Diseases.American Heart Association Science Advisory and CoordinatingCommittee. Circulation. 2002;106:388-391.

32. Ogden CL, Fryar CD, Carroll MD, Flegal KM. Mean body weight,height, and body mass index, United States 1960-2002. Adv Data.2004;27:1-17.

33. Floyd KC, Yarzebski J, Spencer FA, et al. A 30-year perspective(1975-2005) into the changing landscape of patients hospitalized withinitial acute myocardial infarction: Worcester Heart Attack Study. CircCardiovasc Qual Outcomes. 2009;2:88-95.

34. McGovern PG, Jacobs Jr DR, Shahar E, et al. Trends in acute coronaryheart disease mortality, morbidity, and medical care from 1985through 1997: the Minnesota heart survey. Circulation. 2001;104:19-24.

35. Cooper RA. Weighing the evidence for expanding physician supply.Ann Intern Med. 2004;141:705-714.

36. Martini EM, Garrett N, Lindquist T, Isham GJ. The boomers arecoming: a total cost of care model of the impact of population aging onhealth care costs in the United States by Major Practice Category.Health Serv Res. 2007;42(1 part 1):201-218.

37. Foot DK, Lewis RP, Pearson TA, Beller GA. Demographics andcardiology, 1950-2050. J Am Coll Cardiol. 2000;35:1067-1081.

38. Council on Graduate Medical Education. Reassessing physician work-force policy guidelines for the U.S. 2000-2020. Washington, DC: U.S.Department of Health and Human Services; 2003.

39. Halvorson GC. Credibility and creativity: a conversation with KaiserPermanente’s George C. Halvorson. Health Aff (Millwood). 2004;23:133-142.

40. Rice DP, Fineman N. Economic implications of increased longevity in
the United States. Annu Rev Public Health. 2004;25:457-473.
http://www.healthypeople.gov/2020/default.aspx

http://www.healthypeople.gov/2020/default.aspx

1�

lTahr

dhbdafenlnFonur

cltS

o

nnpmItac“tw

3

833.e1 The American Journal of Medicine, Vol 124, No 9, September 2011

APPENDIX:The Coronary Heart Disease Policy Model is a computer-simulation, state-transition (Markov) model of coronaryheart disease incidence, prevalence, mortality, and costs inthe US population over 35 years of age. In the version of themodel used for this analysis, the US population age 35 to 85years, without a history of coronary heart disease, wasapportioned into 4860 risk cells defined by 6 modifiable riskfactors: systolic blood pressure (SBP) (�130, 130 to 139.9,and �140 mm Hg), low-density lipoprotein (LDL) choles-terol (�2.6, 2.6 to 3.3,and �3.4 �mol/L; �100, 100 to29.9, and �130 mg/dL), high-density lipoprotein (HDL;1.0, 1.0 to 1.5, and �1.6 �mol/L; �40, 40 to 59.9, and

�60 mg/dL), smoking status (active smoker, nonsmokerwith exposure to environmental tobacco smoke, or non-smoker without environmental exposure), body mass index(BMI; �25, 25 to 29.9, or �30 kg/m2), and diabetes mel-itus (yes or no), as well as by sex and 10-year age range.he population with prevalent coronary heart disease waspportioned into 1300 cells according to their age, sex,istory of myocardial infarction (MI), arrest, angina, and/orevascularization.

Coronary heart disease incidence and noncoronary heartisease deaths in the population without previous coronaryeart disease were determined by logistic risk functionsased on Framingham longitudinal data.1 Transitions in theisease history component of the model were based onge-range specific event and case fatality rates estimatedrom national data, and literature-based relative risks ofvents among disease history subgroups (eg, previous MI vsone). Noncoronary heart disease death rates in the popu-ation with coronary heart disease reflect the relative risk ofoncoronary heart disease death for this population in theramingham data. In the absence of evidence of a trend, allf these rates were assumed to remain constant. Absoluteumbers of events vary with temporal changes in the pop-lation, the age range distribution of the population, and inesponse to user-defined interventions.

All population distributions, risk factor levels, coeffi-ients, event rates, case fatality rates, costs, and quality ofife adjustments can be modified for forecasting simula-ions. We ran the model on Fortran 95 (Lahey Computerystems, Incline Village, Nev).

Population EstimatesPopulation estimates for the adult US population 35 years ofage and older in 2000 were obtained from the US Census2000 data by age and sex. We used projections (2010 to2040) of the 35-year-old population, based on the 2008national projections.2

Coronary Heart Disease PrevalenceWe estimated the background prevalence of coronary heartdisease in 2000 from the National Health Interview Survey.3

We estimated the background prevalence of previous revas-
cularization procedures from revascularizations before 2000
and estimated survival after revascularization from the Na-tional Hospital Discharge Survey (NHDS)4 and otherstudies.5,6

Coronary Heart Disease DeathsWe obtained data on coronary heart disease deaths from the2000 Vital Statistics Mortality Data.7 We estimated coro-nary heart disease deaths on the basis of International Clas-sification of Diseases, 10th revision, (ICD10) codes I20 toI25, I46, and 2/3 of I49, I50, and I51.8,9 We consideredther deaths to be noncoronary heart disease deaths.

Arrest (Sudden Death) with ResuscitationThe number who survive arrest to hospital discharge wasestimated from NHDS4 for 1990 to 1999. Because theumbers are very small in any given year, we averaged theational estimates over the 10-year period. Estimates ofrehospital arrest fatalities were based on Vital Statisticsortality data for selected causes by place of death.10 For

CD10 codes I20 to I25, all emergency room deaths andhose dead on arrival were assumed to be deaths fromrrests. All nursing home deaths were considered to behronic coronary heart disease deaths. In-residence andother place” deaths were estimated to have had resuscita-ion attempted based on reported resuscitation rates foritnessed11 or unwitnessed12 cardiac arrest.

Proportion of Arrests with no History ofCoronary Heart DiseaseThe coronary heart disease history of arrest patients isharder to ascertain than for MI because there is no nationalregistry, because the numbers are smaller, and becausefewer studies are available. We estimated the age rangespecific proportions of arrest with and without a history ofcoronary heart disease by a least squares fit to data fromseveral sources.13,14

Myocardial InfarctionWe estimated MI incidence as the average annual number ofdischarges coded as 410 in the NHDS 2000 data set. Weeliminated records of MI in which hospital stay was fewerthan 3 days and no acute revascularization was done in thesame hospitalization as probable “rule out MI” cases. Wereduced remaining counts by the double count fraction re-ported by Westfall and McGloin15 and applied an additional% deduction for miscoding, as reported by Petersen et al.16

Myocardial Infarction Case Fatality RatesWe obtained mean number of MI deaths per adjusted totalMI from the NHDS, 1996 to 2000, for the older age ranges(65 to 84 years) and used the National Registry of Myocar-dial Infarction17 for in-hospital case fatality rates for theyounger age ranges (35 to 44 years). Studies of youngpatients with MI estimate an in-hospital mortality rate of 1%to 6%, compared with a rate of 8% to 22% for older
patients.18

obnThur

aafp

aeedcal

afffoowp

d

A

ltae

[yhit0rttrytdi

e

833.e2Odden et al Heart Disease in the Aging US Population

We estimated in-hospital and 30-day case fatality ratesfrom hospital discharge records from the State of CaliforniaOffice of Statewide Health Planning and Development forthe year 2000.19 The in-hospital case fatality rate was basedn unique person records (duplicate entries were eliminatedy matching social security numbers). We omitted a smallumber of records that did not have social security numbers.he overall rate ratio of 30-day case fatality rate to in-ospital case fatality rate was 1.28953 (12.07/9.36). Wesed this ratio to adjust national in-hospital case fatalityates to 30-day mortality rates.

On the basis of a study by Rieves et al,20 we incorporatedmortality odds ratio of 1.6 for patients with previous MI

nd 1.17 for patients with previous angina. Subset caseatality rates were calculated to reflect these odds ratios andreserve the overall estimated case fatality rate for MI.

Revascularization RatesWe estimated the number of percutaneous coronary inter-vention (PCI) and coronary artery bypass grafting (CABG)procedures from the NHDS for 2000. We adjusted therevascularization rate to reflect a repeated revascularizationrate for PCI and for CABG in the first year. We estimateda trend in the ratio of PCI to CABG for 2000 to 2004. Weassumed that PCI would be included as part of the treatmentfor MI in the same proportion observed in the NHDS dataset for 2000, with emergency CABG complicating 2% ofthese procedures. We included reductions in mortality andreinfarction rates for patients treated with PCI.21,22

Risk Functions for Incident Coronary HeartDisease and Noncoronary Heart Disease DeathWe determined incident coronary heart disease cases (MI,cardiac arrest, or angina) and coronary heart disease andnoncoronary heart disease deaths in each risk factor cell forthe at-risk US population by using risk functions (r) thatincorporated age- and sex-specific parameters (�) and riskfactor-specific �-coefficients (�k; k � 1, 2, 3, . . ., 6), whichare constant over the time span of a simulation, and cell-specific risk factor means (mk; k � 1, 2, 3, . . ., 6), which areltered by user-defined intervention. We determined �-co-fficients for noncoronary heart disease death from the samexamination sets of the Framingham cohort and offspringata that we used for the coronary heart disease �-coeffi-ients, but with diastolic blood pressure, smoking, and di-betes as the only statistically significant covariates in theogistic regression analysis.

We estimated overall incidence of coronary heart diseasend noncoronary heart disease death by age range and sexor 2000 by adjusting the Framingham incidence estimatesor 1986 to take into account the trends in risk factor meansrom 1986 to 2000. We estimated the corresponding valuesf the intercepts by iterative fitting of the risk function to theverall incidence. We estimated incidence of cardiac arrestithout previous coronary heart disease by using the pro-
ortion of cardiac arrest without previous coronary heart f
isease in Olmsted County13 and incidence of MI by usingthe proportion of MI without previous coronary heart dis-ease in several published studies that analyzed data from theNational Registry of Myocardial Infarction 2,23 the Cardio-vascular Cooperative Project,24 and the Worcester Heart

ttack study.17,25

We assumed all risks and rates to be constant over time,in the absence of evidence of a trend. We incorporatedtrends as they became apparent, such as that for the use ofrevascularization between 2000 and the present, but did notproject them into the future. The coronary heart disease andnoncoronary heart disease death risk functions are appliedto every state in every year of a simulation to accommodatethe competing risk for these 2 outcomes naturally over time.

Incident Coronary Heart Disease EventAllocationWe assumed that risk factors would affect the incidence ofMI, cardiac arrest, and angina in proportion to overall inci-dence, except we assumed smoking had a higher relativerisk for infarction and cardiac arrest26 and a proportionatelyower coefficient for angina. We assumed that environmen-al tobacco exposure carried a relative risk of 1.26 for MInd cardiac arrest, compared with nonexposed nonsmok-rs,27 but did not influence angina.

Risk Factor Prevalence and Correlationsbetween Risk FactorsWe estimated the prevalence of each risk factor level andcorrelations among risk factors (and thus the apportionmentof the US population without coronary heart disease into the3240 risk cells) from National Health and Nutrition Exam-ination Study (NHANES), 2003 to 2006.28

Transitions between Risk Factor LevelsTransitions from one risk factor level to another were in-cluded to preserve the NHANES proportions of the popu-lation with each risk factor level. For example, the propor-tion of 35- to 44-year-old men with low (�2.6 �mol/L�100 mg/dL]) LDL cholesterol is 0.215. For 45- to 54-ear-old men, the proportion is 0.133. The shift towardigher LDL cholesterol levels is most likely caused byncreasing LDL levels as people age. In higher age ranges,his trend reverses, so that by age 75 to 84, the proportion is.314. The change in the upper age ranges is most likelyelated to a more complex array of factors, including the facthat people with higher risk are more likely to die. Annualransfer rates between risk factor levels were calculated toeduce the low risk population from 0.215 to 0.133 over 10ears, without regard to the reason for the change, butaking into account the effect of the model’s coronary heartisease incidence and noncoronary heart disease death ratesn the base case.

Because the Coronary Heart Disease Policy Model cat-gorizes risk factors into strata, the transitions between risk
actor levels are applied uniformly across all risk factor

tM

ay

fcrw

t

ts

tdmdv

ow1v


subgroups. For example, if the trend in a particular agegroup is toward higher systolic blood pressure, the transi-tion moves the same proportion of people in the low systolicblood pressure cell to the higher category, without regard toother risk factors. These transitions preserve the NHANESage range specific marginal distributions over time, butcannot represent differential transitions within the risk fac-tor subcategories.

CostsWe estimated total health care costs from the perspective ofthe health care system by using national data.29 We esti-mated the coronary heart disease cost component by usingCalifornia data,19 deflated by using cost-to-charge ratios30

and the ratio of the US national average costs to the Cali-fornia average,31 and then inflated to 2010 dollars by usinghe Bureau of Labor Statistics Consumer Price Index for

edical Care Costs.32 We based health-related quality oflife weights on observational data33 and discounted costsnd quality adjusted life years (QALYs) at a rate of 3% perear in the base case analysis.

ImmigrationPeople enter the Coronary Heart Disease Policy Model atage 35 and exit the model at age 85 or death. The modeldoes not account for persons who immigrate into or emi-grate out of the US population after 35 years of age. Be-cause, on average, the immigration rate exceeds the emi-gration rate, we would expect that this would bias the modelto underestimate the population size and number of annualcoronary heart disease events and mortality. If the immi-grant population is healthier compared with the general USpopulation, this could bias estimates of the incident andprevalence rates away from the null.

Quality Control and ValidationThe Coronary Heart Disease Policy Model was calibrated toreproduce national data on risk factor distributions, totalcoronary heart disease deaths, acute MI, witnessed suddencardiac death, and revascularization procedures in the baseyear. Validation of projections into the future is an ongoingeffort in which the model’s results under a broad range ofscenarios are compared with data from studies, clinicaltrials, and surveys, obtained from public sources or bypersonal communication. Validation required reasonableagreement in outcomes when the conditions that producedthe data were incorporated. For example, simulations ofpersons in the US population age 45 to 64 years that im-posed the before and after LDL cholesterol and HDL cho-lesterol levels recorded in the West of Scotland CoronaryPrevention Study (WOSCOPS)34 resulted in similar resultsor the cumulative percentage of the cohort to have died oforonary heart disease or have had a first MI, and for theatio of events in participants who were and were not treatedith statins.For validation of cost and cost-effectiveness aspects of

he model, the model was used to duplicate a cost-effec-

iveness analysis of secondary prevention based on the 4Study.35 With discounting set to 5%, no quality of life

adjustments, and coronary heart disease and drug costs only(excludes noncoronary heart disease health care costs), themodel produced a cost-effectiveness ratio of $13,100/lifeyear for statins overall in the 35- to 74-year-old secondaryprevention population, as compared with age and sex-spe-cific analyses from 4S that estimated ratios ranging from$4700/QALY (70-year-old men) to $18,800/QALY (35-year-old women).35

Comparison with 1987 ProjectionsIn order to estimate the model’s performance in previousstudies aimed at predicting the future burden of disease, wecompared the projections from a 1987 publication to ob-served statistics from the year 2005.36 The previous projec-ions underestimated the prevalence of future coronary heartisease but overestimated incidence and mortality36; theajority of the variation can be explained by a change in

isease definition, and reductions in risk factors, and ad-ances in treatment.

In 1987, Weinstein et al37 projected that the prevalencef coronary heart disease would be 8,385,000 in 2005,hich is 34% lower than the estimated prevalence of2,649,000, according to the 2005 National Health Inter-iew Survey (NHIS).37 This underprojection is largely ex-

plained by a revision of the survey in 1997, and a redefini-tion of prevalent coronary heart disease. We calculated theaverage age- and sex-specific prevalence in years 1992 to1996 and 1997 to 2001 and found that the definition changewas associated with an increase in reported prevalence rateof approximately 33%. This change in definition appears toexplain the majority of the discrepancy; if the 1987 projec-tions are adjusted to reflect the definition change, there isonly a 12% difference between the 1987 model projectionsand 2005 NHIS estimates. We propose that the remaining12% higher projected prevalence likely is related to im-provements in coronary heart disease risk factors.

In 1987, the Coronary Heart Disease Policy Model pro-jected 888,000 incident coronary heart disease events in2005, whereas the 2006 Heart and Stroke Statistics Updateestimated approximately 700,000 incident coronary heartdisease events, based on data from the Atherosclerosis Riskin Communities Study (ARIC). This difference is consistentwith published temporal trends of coronary heart diseaseincidence in the US; about a 9% per decade decline inage-adjusted incidence rate, thought largely to be related toimproved primary prevention.38 If we apply the observedtemporal trend of a 9% per decade decline in the age-adjusted incidence rate to the estimates from 1987, weobserve only a 6% difference between our estimates andthose from the 2006 Heart and Stroke Statistics Update. Wehypothesize that this difference could be related to the“healthy cohort” bias, in which participants in studies areoften healthier and have lower rates of disease compared
with the general population.39

cdaoOdmt

1

1

1

1

833.e4Odden et al Heart Disease in the Aging US Population

In 1987, the Coronary Heart Disease Policy Model pro-jected that coronary heart disease mortality would be608,000 in 2005, whereas data from the US vital statisticsreported only 321,000 deaths caused by coronary heartdisease in the same year. This trend is consistent withpublished reports of a decline in coronary heart diseasemortality.40,41 Ford et al40 described a 27% decline in theoronary heart disease deaths from 1980 to 2000; if theecline in mortality because of improvements in risk factorsnd treatments is applied to our 1987 projections, they arenly 4% higher compared with the vital statistics report.thers have estimated that this decline in coronary heartisease mortality is approximately half caused by improve-ents in risk factors and half related to improvements in

reatments of patients with coronary heart disease.40,41 Inaddition, there was a decrease in deaths attributable todisease of the heart because of the change in coding incardiac arrest, which was included under ICD-9 coding, butnot ICD10.42

References1. Framingham Heart Study. Washington, DC: Department of Health and

Human Services; 2005.2. US Census Bureau. U.S. interim projections by age, sex, race, and

Hispanic origin 2008. Washington, DC: US Census Bureau; 2008.3. Centers for Disease Control and Prevention. National Health Interview

Survey. Hyattsville, MD: Centers for Disease Control and Prevention;2008.

4. Centers for Disease Control and Prevention. National Hospital Dis-charge Survey. Hyattsville, MD: Centers for Disease Control andPrevention; 2006.

5. van Domburg RT, Foley DP, Breeman A, van Herwerden LA, SerruysPW. Coronary artery bypass graft surgery and percutaneous translu-minal coronary angioplasty. Twenty-year clinical outcome. EurHeart J. 2002;23:543-549.

6. Yusuf S, Zucker D, Passamani E, et al. Effect of coronary arterybypass graft surgery on survival: overview of 10-year results fromrandomised trials by the Coronary Artery Bypass Graft Surgery Tri-alists Collaboration. Lancet. 1994;344:563-570.

7. Centers for Disease Control and Prevention. Year 2000 vital statisticsmortality data: population death by cause. Hyattsville, MD: Centers forDisease Control and Prevention; 2002.

8. Centers for Disease Control and Prevention. Deaths for 358 selectedcauses, by 5-year age groups, race and sex: United States, 1999-2000.Hyattsville, MD: Centers for Disease Control and Prevention; 2002.

9. Consensus recommendations for the management of chronic heartfailure. On behalf of the membership of the advisory council toimprove outcomes nationwide in heart failure. Am J Cardiol. 1999;83:1A-38A.

0. Centers for Disease Control and Prevention. GMWK307: deaths from39 selected causes by place of death, status of decedent when deathoccurred in hospital or medical center, and age: United States, 1999-2004. Hyattsville, MD: Centers for Disease Control and Prevention;2006.

1. de Vreede-Swagemakers JJM, Gorgels APM, Dubois-Arbouw WI, etal. Out-of-hospital cardiac arrest in the 1990s: a population-basedstudy in the Maastricht area on incidence, characteristics and survival.J Am Coll Cardiol. 1997;30:1500-1505.

2. Kuisma M, Jaara K. Unwitnessed out-of-hospital cardiac arrest: isresuscitation worthwhile? Ann Emerg Med. 1997;30:69-75.

3. Goraya TY, Jacobsen SJ, Kottke TE, Frye RL, Weston SA, Roger VL.Coronary heart disease death and sudden cardiac death: a 20-year
population-based study. Am J Epidemiol. 2003;157:763-770.
14. Albert CM, Chae CU, Grodstein F, et al. Prospective study of suddencardiac death among women in the United States. Circulation. 2003;107:2096-2101.

15. Westfall JM, McGloin J. Impact of double counting and transfer biason estimated rates and outcomes of acute myocardial infarction. MedCare. 2001;39:459-468.

16. Petersen L, Wright S, Normand S-L, Daley J. Positive predictive valueof the diagnosis of acute myocardial infarction in an administrativedatabase. J Gen Intern Med. 1999;14:555-558.

17. Vaccarino V, Krumholz HM, Yarzebski J, Gore JM, Goldberg RJ. Sexdifferences in 2-year mortality after hospital discharge for myocardialinfarction. Ann Intern Med. 2001;134:173-181.

18. Choudhury L, Marsh J. Myocardial infarction in young patients. Am JMed. 1999;107:254-261.

19. California Patient Discharge Data, January 1-December 31 2000, Pub-lic Version A-24. Sacramento: California Office of Statewide HealthPlanning and Development; 2001.

20. Rieves D, Wright G, Gupta G, Shacter E. Clinical trial (GUSTO-1 andINJECT) evidence of earlier death for men than women after acutemyocardial infarction. Am J Cardiol. 2000;85:147-153.

21. Lagerqvist Bo, Husted S, Kontny F, et al. A long-term perspective onthe protective effects of an early invasive strategy in unstable coronaryartery disease: two-year follow-up of the FRISC-II invasive study.J Am Coll Cardiol. 2002;40:1902-1914.

22. Zijlstra F, Hoorntje JCA, de Boer M-J, et al. Long-term benefit ofprimary angioplasty as compared with thrombolytic therapy for acutemyocardial infarction. N Engl J Med. 1999;341:1413-1419.

23. Rogers WJ, Canto JG, Barron HV, et al. Treatment and outcome ofmyocardial infarction in hospitals with and without invasive capabil-ity. J Am Coll Cardiol. 2000;35:371-379.

24. Krumholz HM, Chen J, Chen Y-T, Wang Y, Radford MJ. Predictingone-year mortality among elderly survivors of hospitalization for anacute myocardial infarction: results from the Cooperative Cardiovas-cular Project. J Am Coll Cardiol. 2001;38:453-459.

25. Goldberg RJ, McCormick D, Gurwitz JH, Yarzebski J, Lessard D,Gore JM. Age-related trends in short- and long-term survival afteracute myocardial infarction: a 20-year population-based perspective(1975-1995). Am J Cardiol. 1998;82:1311-1317.

26. Parish S, Collins R, Peto R, et al. Cigarette smoking, tar yields, andnon-fatal myocardial infarction: 14000 cases and 32000 controls in theUnited Kingdom. BMJ. 1995;311:471-477.

27. Law MR, Morris JK, Wald NJ. Environmental tobacco smoke expo-sure and ischaemic heart disease: an evaluation of the evidence. BMJ.1997;315:973-980.

28. Centers for Disease Control and Prevention. National Health andNutrition Examination Survey. Hyattsville, MD: Centers for DiseaseControl and Prevention; 2008.

29. Centers for Medicare and Medicaid Services. Personal health carespending by type of spending, age group, and source of paymentdistribution, calendar year 1999. Woodland, MD: Centers for Medi-care and Medicaid Services; 2000.

30. Office of Statewide Health Planning and Development. Hospital fi-nancial data for cost to charge ratio, CA inpatient discharge datahospital annual finanacial data, pivot profiles, 1999-2000. Sacra-mento, CA: California Office of Statewide Health Planning and De-velopment; 2008.

31. US Census Bureau. Statistical abstract of the United States. Averagecost to community hospitals per patient, by state (Table 204). Wash-ington, DC: Goverment Printing Office; 1998.

32. Bureau of Labor Statistics. Consumer price index. Washington, DC:Bureau of Labor Statistics; 2010.

33. Fryback DG, Dasbach EJ, Klein R, et al. The Beaver Dam HealthOutcomes study: initial catalog of health-state quality factors. MedDecis Making. 1993;13:89-102.

34. Shepherd J, Cobbe SM, Ford I, et al. Prevention of coronary heartdisease with pravastatin in men with hypercholesterolemia. N Engl
J Med. 1995;333:1301-1308.

3

3

3

3

4

4

4


35. Johannesson M, Jonsson B, Kjekshus J, Olsson AG, Pedersen TR,Wedel H. Cost effectiveness of simvastatin treatment to lower choles-terol levels in patients with coronary heart disease. ScandinavianSimvastatin Survival Study Group. N Engl J Med. 1997;336:332-336.

6. Weinstein MC, Coxson PG, Williams LW, Pass TM, Stason WB,Goldman L. Forecasting coronary heart disease incidence, mortality,and cost: the Coronary Heart Disease Policy Model. Am J PublicHealth. 1987;77:1417-1426.

7. National Health Interview Survey datasets. Washington, DC: NationalCenter for Health Statistics; 2005.

8. Arciero TJ, Jacobsen SJ, Reeder GS, et al. Temporal trends in the
incidence of coronary disease. Am J Med. 2004;117:228-233.
9. DiMartino LD, Hammill BG, Curtis LH, et al. External validity of thecardiovascular health study: a comparison with the Medicare popula-tion. Med Care. 2009;47:916-23.

0. Ford ES, Ajani UA, Croft JB, et al. Explaining the decrease in U.S.deaths from coronary disease, 1980-2000. N Engl J Med. 2007;356:2388-2398.

1. Hunink MG, Goldman L, Tosteson AN, et al. The recent decline inmortality from coronary heart disease, 1980-1990. The effect of sec-ular trends in risk factors and treatment. JAMA. 1997;277:535-542.

2. Centers for Disease Control and Prevention. Comparability of cause ofdeath between ICD-9 and ICD-10: preliminary estimates. Natl Vital
Stat Rep. 2001;49:1-32.

pi is 0002934311002956

Documents