Despite a recent decline, ischemic heart disease (IHD) remains the world's leading cause of death as well as a major contributor to health care expenses. France has a much lower heart disease mortality rate (78.4 per 100,000) than most other nations belonging to the Organization for Economic Cooperation and Development (OECD) and less than half that of the U.S. (163.4)1 – a phenomenon often dubbed the “French paradox.”2
Cross-national comparisons indicate that the U.S. provides higher rates of revascularization procedures -- cardiac catheterization, percutaneous transluminal coronary angioplasty (PTCA) and coronary artery bypass graft surgery (CABG) – than France and other developed nations.3 A study based on micro-level data sources, the most recent state-of-the art international comparison on the use of high-tech interventions, concludes that the U.S. is more aggressive than Canada, Scotland, Sweden, Israel, Australia and Denmark in providing services following heart attacks.4 Such studies, however, may be misleading for four reasons. First, cross-national studies often reflect a misunderstanding of how U.S. data are coded and aggregated.5 Second, most studies do not adjust treatment rates for differences in the burden of IHD. Third, cross-national analyses of IHD prevalence and treatment are also limited because they tend to rely on national averages of crude rates. This masks enormous local variations by age and area of residence. Fourth, although deaths due to IHD disproportionately affect persons over 65 years of age (85% in the U.S., and 87% in France),6 most cross-national comparisons do not focus on older people.7
To address these limitations of cross-national studies, we compare mortality, morbidity and treatment modalities for IHD between two age cohorts (45-65 years and 65+) of residents in two nations (U.S. and France) and the hearts of two world cities (New York, and Paris). Paris and Manhattan – the “urban cores”8 of these cities, have heterogeneous populations and renowned medical centers. In this respect, they are more similar in socio-demographic characteristics and levels of health care resources than their respective countries. Higher rates of difficulties in access are reported in Manhattan than Paris and even older residents insured under the Medicare Program are known to encounter significant barriers to care. In France there are virtually no financial barriers to primary or tertiary services; in the U.S. as much as 16 percent of the population and in Manhattan 28 percent of the population is uninsured.9 Our comparison therefore raises questions about the effects of these two health systems on access to specialized medical services. While we cannot draw causal inferences from an ecological study, analysis of data on health status and medical care practice between Manhattan and Paris goes beyond the level of comparative analysis made possible by OECD data and yields insights for further comparative international health services research.
Spatial Units of Analysis: Rationale for Comparing Manhattan and Paris
Comparison of national health systems can provide insights into the effects of health system characteristics on medical practices, but a focus on national aggregates masks important variations within nations, e.g., between urban and rural areas, large and small cities, or cities and suburbs. In addition to having different systems of health care finance and delivery, between and within each country there are differences in levels of medical resources10-11 and practice patterns, as well as medical culture, vary. This makes it difficult to assess the effects of any one factor on medical practices. To address these limitations we examine city-level data. While this does not overcome entirely the problems noted above, in contrast to nation-states, world cities share more characteristics and problems in common. To the extent that there are differences in sociodemographic and health system characteristics, they provide a unique framework within which to investigate differences in the treatment of heart disease.
New York (population of 8 million) and Paris and its first ring (6 million), are the largest cities among the higher-income nations belonging to the OECD. They also have the highest concentration of older persons in their respective nations. Beyond considerations of scale, New York and Paris share world city status due to their concentration of high-level functions in government, business, health care, media and the arts.12
The value of these city-level comparisons is enhanced when we define carefully our units of analysis. Paris, a city of 2.1 million inhabitants living within 105 square kilometers, is minuscule in comparison to New York City's 826 square kilometers and 8 million people. Nonetheless, it is comparable to Manhattan, the urban core of New York. Manhattan and Paris are similar in terms of population size, density and socio-demographic characteristics.13 They are also similar with respect to the concentration of medical resources. The urban core of each city has a much higher density of physicians than their suburbs. They also have a higher concentration of acute care hospital beds (public and private combined); Manhattan has 2.5 times as many beds as its first ring while Paris has 1.5 times as many.
Table 1 National and City-Level Characteristics
*sources: New York State Department of Health; Ministere de l'Emploi et de la Solidarite - SESI - repertoire ADELI au 1er janvier 98; Society of Thoracic Surgeons; American College of Cardiology; DREES : Annuaire des statistiques sanitaires et sociales 2000
We obtained U.S. data for hospital discharges and rates of PTCA and CABG from the National Hospital Discharge Survey, National Center for Health Statistics. We obtained these data for Manhattan, by area of residence, from the Statewide Planning and Research Cooperative System (SPARCS), a comprehensive inpatient hospital patient data system established in 1979 as a result of cooperation between the health care industry and the New York State Department of Health. We obtained comparable hospital discharge data for France and Paris, by area of residence, from the French Ministry of Health's Hospital Reporting System, Programme pour la Médicalisation des Systèmes d'Information (PMSI). PMSI centralizes discharge data from all French hospitals by diagnosis, procedure, age and geographic origin of patients. All hospitals with more than 100 beds provide data to this system. For each city, we analyze data for specific diagnostic and procedure codes by gender, for age cohorts 45-64 and 65.
We aggregate Manhattan hospital discharge data for 12 community districts and Paris hospital discharge data for the 20 arrondissements . We also examine mortality data, from INSERM (Institut Nationale pour les Statistiques, Etudes et Recherche Medicale) and the New York City Department of Health (NYCDOH), for these codes, cohorts and time periods. To assure an adequate number of deaths, hospital discharges, and procedures for meaningful comparisons, we calculate four-year averages for Manhattan and Paris (1997-2000).
The WHO MONICA Project highlights some of the difficulties associated with comparing death rates across populations.14 For example, project investigators observed that documentation of nonfatal cardiac events were more reliable, in part, because a large proportion of deaths were unclassifiable due to a lack of diagnostic information or any medical history. When categorizing nonfatal episodes, they classified events as definite, possible, or “no myocardial infarction” using only ECG criteria because cardiac enzyme criteria are not universally available and standardized.15
National IHD mortality rates are clearly not uniform within countries in the MONICA study and, as noted earlier, mortality data may not be reliable. Some have expressed concern that the differences in criteria for ascribing death to IHD in France may be the reason for the lower French rates.16,17
Addressing these concerns, we rely on hospital discharge data to test the reliability of the mortality data. These hospital discharge data can validate findings where accepted clinical diagnostic criteria exist, as in the case of AMI. Other national comparisons focus on broad categories of coronary artery or ischemic heart disease, but these include conditions for which diagnoses are less reliable, such as congestive heart failure (CHF).18 To avoid omitting patients that might have been misdiagnosed or miscoded, we adopt a conservative strategy and identify patients using the ICD-9-CM and ICD-10 diagnostic codes for AMI and also all ischemic heart disease (IHD). Examining AMI separately acts as a check on the reliability of broader categories.
Our use of SPARCS and Ministry of Health data on procedure rates for residents of Manhattan and Paris avoids a common misinterpretation of the number of CABG procedures in the U.S. OECD estimates are based on a sample of inpatient records from short stay-hospitals in the U.S. (National Hospital Discharge Survey, National Center for Health Statistics) which are widely disseminated in the "Heart and Stroke Statistical Update" of the American Heart Association. The data for cardiac revascularization (bypass - ICD 9 codes 36.1-36.3) are presented as procedures (553,000 in 1998) and as patients (336,000 in 1998). The “procedure” variable represents the number of coded procedures recorded, but for a given operative procedure on a single patient either a single code (representing an arterial or a venous conduit bypass) or two codes (employed when a combination of arterial and venous conduits have been used) may be counted. The frequent use of two codes, for what is in fact a single operative procedure, exaggerates the number of procedures performed in the U.S.19
Measuring Treatment for IHD
To ascertain differences in invasive treatment for ischemic heart disease we calculate age-adjusted rates of revascularization (PTCA and CABG) per 100,000 population for two cohorts, those 45-65 years and those 65 years and over. We do not examine diagnostic cardiac catheterization or coronary angiography rates since they are performed as outpatient procedures in both countries and data on their volume are unreliable.
Assessing the Relationship between Treatment Rates and Disease Prevalence
To assess the relationship between treatment rates and the prevalence of IHD, we introduce a simple index based on the ratio of procedure rates to AMI hospital discharge rates and AMI mortality rates. Although the true prevalence of ischemic heart disease in any population will never be known, since the illness may be asymptomatic, as a proxy for the prevalence of ischemic heart disease, we examine both mortality and hospital discharge rates for AMI. We do not intend to suggest that every patient diagnosed with an AMI receives one of these procedures, nor do we intend to suggest that this is the only diagnosis for which these procedures are an appropriate intervention. Examining the ratio 7 of procedure rates to AMI hospital discharge rates is merely an attempt to adjust for the prevalence of heart disease in the two cities. While our index represents a preliminary effort limited by available data, failure to consider some measure of disease prevalence when analyzing treatment rates can be misleading.20
Deaths rates from IHD in Manhattan and Paris reflect the well-known differences in mortality from IHD between the U.S. and France. Manhattan residents exhibit higher mortality rates than their Parisian counterparts for AMI and all IHD. This contrast is more pronounced with age. For those 65 years old, rates of mortality due to AMI in Manhattan are 53.4 percent higher; for those 45-64, they are 37.9 percent higher.
Table 2 Age-Adjusted 1 Mortality and Hospital Discharge Comparisons: 1997-2000 Average
*=significant at p<.05 level
U.S. and Manhattan residents exhibit a higher rate of hospitalization for AMI and ischemic heart disease than residents of France and Paris. Once again, the contrast is more pronounced for older persons than for those 45-64. For those 65 years old, hospital discharge rates of AMI in Manhattan are 45.4 percent higher; for those 45-64, they are only 28.5 percent higher (Table 2). Thus, the comparison of hospital admission rates is consistent with the differences in mortality rates.
When we examine age-adjusted rates of revascularization, without attempting to account for national and city-level differences in disease rates, our findings are consistent with previously reported national findings. The age-adjusted rate of revascularization per 100,000 is significantly higher (53.1%) in the U.S. than France for persons 65 years and over and for persons 45-64 years old (44.5%).
In Manhattan and Paris, the difference in the age-adjusted rates of revascularization per 100,000 is consistent with the national figures for persons 65 years and over (21.4% higher in Manhattan than Paris), but the difference between these cities is not as great as the difference between the two nations. The rate of revascularization per 100,000 for persons 45-64 years old, however, is actually higher in Paris than Manhattan (2%), despite significantly higher rates of AMI and IHD among residents of Manhattan (Table 3).
Table 3 Age-Adjusted Procedure Comparison: 1997-2000 Average
*=significant at p<.05 level
When we use our index to adjust for the prevalence of disease in the two nations and cities, the contrast between our findings and those of previous national comparisons are even more striking. The ratio of revascularization procedure rates to AMI discharge rates is only 11.4% higher in the U.S. than France for persons 65 years and over. For persons 45-64 years old, the ratio of revascularization procedure rates to AMI discharge rates is higher in France than the U.S. (20%) (Table 4).
Table 4 Ratio of Age Adjusted Procedure Rate to AMI Admission Rate, 1997-2000 Averages
*=significant at p<.05 level
The ratio of revascularization procedure rates to AMI discharge rates is significantly higher in Paris than in Manhattan for persons 65 years and over (15%), and even greater for persons 45-64 years old (44%) (Table 4).
Our findings regarding the hospital discharge rates by area of patient residence, for AMI, as well as IHD, are consistent with previous cross-national comparisons of heart disease in France and the U.S.; death rates are higher in the U.S. and Manhattan than in France and Paris, respectively. Likewise, based on hospital discharge rates for AMI and all IHD, there is strong evidence that the burden of ischemic heart disease is significantly higher in the U.S. and Manhattan than in France and Paris (Table 3). Moreover, the disparity in disease burden between the U.S. and France and Manhattan and Paris increase with age, and mortality rates from all IHD are more than three times as high in the U.S. and Manhattan than in France and Paris.
A recent comparison of trends in invasive procedure use for patients with heart disease indicates that the U.S. provides higher rates of cardiac catheterization, PTCA, and bypass surgery, within one year after a heart attack, compared with the other high-income OECD nations in the study.21 Similarly, a comparison of the U.S. and Canada in the GUSTO-1 study found that the rates of PTCA and CABG among Canadian patients, following AMI were much lower than any of the regional rates reported for the U.S.22
Our analysis of revascularization rates in the U.S. and France and Manhattan and Paris differ markedly from the literature on cross-national comparisons. We find that U.S. residents 45-64 years of age receive invasive procedures at a significantly lower rate than residents of France when treatment is indexed to a measure of disease prevalence. While U.S. residents age 65 years and over do receive these procedures at a significantly higher rate than residents of France, the difference between the two countries is only about half what it appears to be without applying our index.
Furthermore, Manhattan residents receive invasive procedures at a significantly lower rate than Parisians when treatment is indexed to a measure of disease prevalence. The ratio of revascularization rates to AMI discharge rates is 44 percent higher in Paris than in Manhattan for the 45-64 year cohort and 15 percent higher for persons 65 years of age.
What explains the higher use, based on our ratio, of revascularization procedures in France, among younger adults, and Paris, among all adults? Previous studies suggest that large geographic variations in the rates of these services in the U.S. are attributed to factors other than disease incidence.23 These have included, age24 , race25 , sex26,27 , income, co-morbid conditions, location of care and health insurance status.28,29 Other studies point to an inverse relationship between distance to health care personnel or facilities and utilization of services.30-33
Without patient specific clinical data, it is, of course, not possible to separate the effects of health insurance from other factors that may influence the use of cardiac procedures. It is well established, in the literature, that there are a variety of barriers to gaining access to high-tech treatment modalities for IHD in New York and elsewhere in the U.S. Thus, using population-based data, our finding that ratios of discharge rates for revascularization to AMI are significantly higher among Parisians than Manhattanites, and among persons age 45-64 years in France than in the U.S., may reflect the impact of these barriers on the use of these treatment modalities.
Our findings also demonstrate clearly why it is crucial to account for disease prevalence when comparing medical treatment rates in different areas. Previous comparisons of the treatment of heart disease among OECD countries suggest that the U.S. is an outlier, with rates of revascularization that are much higher than other developed nations. However, when we account for the prevalence of disease, however, the use of these procedures in the U.S. is lower in comparison to France among persons 45-64 years and lower among residents of Manhattan in comparison to residents of Paris for both age cohorts.
* Michael K. Gusmano, Victor G. Rodwin, Daniel Weisz, World Cities Project, International Longevity Center–USA; please address all comments, questions and suggestions to MichaelG@ilcusa.org. This paper has previously been published as W27 in the Working Paper Series of the Scholars in Health Policy Research Program. The preparation of this paper was assisted by a grant of The Robert Wood Johnson Foundation, Princeton, New Jersey.
1. French Ministry of Health, INSERM, 1999; U.S. National Center for Health Statistics, 1999.
2. Renaud, S and R. Gueguen. 1998. “The French Paradox and wine drinking” Novartis Foundation Symposium 216: 208-17.
3. OECD (2002). OECD Health Data 2002: Comparative Analysis of 30 Countries, Paris.
4. The Technological Change in Health Care Research Network. Technological Change around the World: Evidence from Heart Attack Care. Health Affairs 2001; 20(3): 25-42.
5. The documentation in the “data sources” of the OECD CD Rom states that: “The rates presented for coronary artery bypass discharges relate to the number of procedures as opposed to the numbers of patients. A separate procedure code is employed for arterial and for venous conduit bypasses, thus a single coronary bypass patient can be counted as receiving more than one bypass procedure if both types of grafts are performed. For example the bypass procedure discharge rate (presented in this database) for 1992 was 184.8 per 100 000 population, but the discharge rate (for patients) was only 121.9 per 100 000.” Similarly the American Heart Association, in the annual Heart and Stroke Statistical Update based on data from the Health Resources Utilization Branch CDC/NCHS reports rates for both procedures and patients for PTCA and bypass codes with the note “(b)- Some patients have both internal mammary artery grafts and saphenous vein grafts. These numbers represent a combination of code and vessel data. It's impossible from this (mixed) data to determine the average number of vessels per patient.” Clearly in one year CABG patients do not undergo on average 1.5 open heart procedures (184.8/121.9).
6. Lakatta E.G. Age-Associated Cardiovascular Changes in Health: Impact on Cardiovascular Disease in Older Persons. Heart Failure Review 2002; 7(1): 29—49; National Vital Statistics Reports , Vol. 47, No. 19, June 30, 1999; National Institute of Health and Medical Research web site, updated September 7, 2001.
7. Houterman S, HC Boshuizen, WM Verschuren, S Giampaoli, A Nissinen, A Menotti, and D Kromhout. Predicting Cardiovascular Risk in the Elderly in Different European Countries. European Heart Journal 2002; 23(4): 294-300.
8. Rodwin V.G. and M.K. Gusmano. 2002. The World Cities Project: Rationale and Design for Comparison of Megacity Health Systems. Journal of Urban Health 2002; 79(4): 445-463.
9. Sandman D, Schoen C, Des Roches C, Makonnen M. The Commonwealth Fund Survey of Health Care in New York City. New York, NY: The Commonwealth Fund; 1997.
10. Wennberg JE, Freeman, JL, Culp WJ. Are Hospital Services Rationed in New Haven or Over-Utilised in Boston? Lancet 1987; 8543:1185-89.
11. Wennberg JE and Gittelsohn A. Small Area Variations in Health Care Delivery. Science 1973; 182: 1102-1108.
12. Geddes P. Cities in Evolution. London: Williams and Norgate, 1951.
13. Rodwin, V and Gusmano, M. The World Cities Project: Rationale and Design for Comparison of Megacity Health Systems. Journal of Urban Health 2002; 79(4): 445-463.
14. Tunstall-Pedoe H, Kuulasmaa K, Amouyel P et al. Myocardial infarction and coronary deaths in the World Health Organization Monica Project: Registration procedures, event rates and case-fatality rates in 38 populations from 21 countries in four continents. Circulation 1994; 90(1): 583-612.
15. Burke GL, Edlavitch SA, Crow RS. The effects of diagnostic criteria on trends in coronary heart disease morbidity: the Minnesota Heart Survey. Journal of Clinical Epidemiology 1989;42: 17-24.
16. Artaud-Wild, S.M., S.L. Connor, G Sexton, and W.E. Connor. Lipids/Atherosclerosis/Arteries: Differences in Coronary Mortality Can be Explained by Differences in Cholesterol and Saturated Fat Intakes in 40 Countries But Not in France and Finland: A Paradox. Circulation 1993; 88(6): 2771-2779.
17. Nestle, M. 1994. “Diet and alchohol in heart disease risk: the French Paradox” Contemporary Drug Problems Spring: 71-76; Renaud, S and R. Gueguen. The French Paradox and wine drinking. Novartis Foundation Symposium 1998; 216: 208-17.
18. Akosah KO, K Moncher, A Schaper, P Havlik, S Devine. Chronic heart failure in the community: missed diagnosis and missed opportunities. Journal of Cardiac Failure 2001; 7(3): 239-240.
19. OECD Health Data, 2001.
20. The Technological Change in Health Care Research Network. Technological Change Around the World: Evidence from Heart Attack Care. Health Affairs 2002; 20(3): 25-42.
21. The Technological Change in Health Care Research Network. Technological Change Around the World: Evidence from Heart Attack Care. Health Affairs 2002; 20(3): 25-42.
22. Rouleau JL, Moye LA, Pfeffer MA et al . A comparison of management patterns after acute myocardial infarction in Canada and the United States. New England Journal of Medicine 1993; 328:779-784.
23. Pilote L, Califf RM, Sapp S et al. (the GUSTO-1 Investigators. Regional variation across the United States in the management of acute myocardial infarction. N Engl J Med 1995;333:565-572.
24. Pashos Cl, Newhouse JP, McNeil BJ. Temporal changes in the care and outcomes of elderly patients with acute myocardial infarction, 1987 through 1990. JAMA 1993;270:1832:1836.
25. Ayanian JZ, Udvarhelyi IS, Gatsonis CA et al. Racial differences in the use of revascularization procedures after coronary angiography. JAMA. 1993;269:2642-2646.
26. Ayanian JZ, Epstein AM. Differences in the use of procedures between men and women hospitalized for coronary hear disease. N Engl J Med 1991;325:221-225.
27. Leape LL, Hilbourne LH, Bell R et al Underuse of cardiac procedures: do women, ethnic minorities and the uninsured fail to receive needed revascularization? Ann Intern Med 1999;130:183:192.
28. Carlisle DM, Leake BD, Shapiro ME. Racial and ethnic disparities in the use of cardiovascular procedures: associations with type of health insurance. Am J Public Health 1997;87:263-267.
29. Philbin EF, McCullough PA, DiSalvo TG et al. Underuse of invasive procedures among Medicaid patients with acute myocardial infarction. Am J Public Health 2001; 91:1082-1088.
30. Blustein J. High technology cardiac procedures: The impact of service availability on service use in New York State. JAMA 1993;270:344-349.
31. Grumbach K, Anderson GM, Luft HS et al. Regionalization of cardiac surgery in the United States and Canada: Geographic access, choice and outcomes. JAMA 1995;274:1282-1288.
32. Ben-Shlomo Y, Chaturvedi N. Accessing equity in access to health care provision in the UK: Does where you live affect your chances of getting a coronary artery bypass graft? J Epidemiol Community Health 1995;49:200-204.
33. Gregory PM, Malka ES, Kostis JB et al. Impact of Geographic Proximity to Cardiac Revascularization Services on Service Utilization. Medical Care 2000;38:45-57.
Edited and posted on the web on 9th December 2004