| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
RESEARCH AND PRACTICE |
Victor G. Rodwin is with the World Cities Project, a joint venture of the Wagner School, New York University, and the International Longevity Center–USA, New York. Leland G. Neuberg is with the Department of Mathematics and Statistics, Boston University, Boston.
Correspondence: Requests for reprints should be sent to Victor G. Rodwin, MPH, PhD, Wagner School, New York University, The Puck Building, 295 Lafayette St, New York, NY 10012 (e-mail: victor.rodwin{at}nyu.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
Objectives. We investigated the association between average income or deprivation and infant mortality rate across neighborhoods of 4 world cities.
Methods. Using a maximum likelihood negative binomial regression model that controls for births, we analyzed data for 1988–1992 and 1993–1997.
Results. In Manhattan, for both periods, we found an association (.05% significance level) between income and infant mortality. In Tokyo, for both periods, and in Paris and London for period 1, we found none (5% significance level). For period 2, the association just missed statistical significance for Paris, whereas for London it was significant (5% level).
Conclusions. In stark contrast to Tokyo, Paris, and London, the association of income and infant mortality rate was strongly evident in Manhattan.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
We departed from more conventional comparative analyses of national health statistics. We took as our units of analysis large administrative neighborhoods in New York, London, Paris, and Tokyo. These world cities have heterogeneous populations (albeit much less so in Tokyo) and increasing income disparities between rich and poor. Within each city, we investigated the relationship between infant mortality rate and an income-related variable across its neighborhoods.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
These urban cores share a number of convergent characteristics. They are medical capitals with a disproportionate share of hospitals and specialist physicians. They are destinations for large immigrant communities from around the world (with the exception of Tokyo). The foreign-born population of Paris (1999), London (2001), and Manhattan (2000) is 22.7%, 33.7%, and 29.4%, respectively.12 In Paris, of the foreign-born from outside the European Union, most are from northern Africa (40.3%), Asia (21.1%), and sub-Saharan Africa (16.9%). In London, those from outside the European Union are mostly from Africa (30.8%), the Caribbean (25.6%), and Bangladesh (17.3%). In Manhattan, recent immigrants are predominantly from Latin America (including the Caribbean [48%], Asia [27%], and Europe [18%]). However, despite these differences, in all 3 cities, recent immigrants have exacerbated social and spatial inequalities.13–15
To some extent, defining neighborhoods is arbitrary. We relied on 2 criteria: existing designations or administrative boundaries and data availability. We obtained data on live births, infant deaths, and an income-related measure that is as similar as possible across cities (our data set is posted at http://www.ilcusa.org/_lib/pdf/ajph.dataappendix.pdf). For Manhattan, we used the 10 subborough units for which the New York City Housing and Vacancy Survey provides mean household income estimates during periods between the decennial census. For London, we used the 14 boroughs of inner London; for Paris, the 20 arrondissements; and for Tokyo, 11 ku.
Infant mortality—deaths in the first year of life—is an important indicator of social well-being, which reflects multiple social determinants of health. The infant mortality rate is defined as the number of infant deaths for a period divided by the number of live births for that period. Neighborhoods with relatively small numbers of births have less stable neighborhood rates. Because some neighborhoods in the 4 cities we examined have a relatively small number of annual live births, to increase the stability of the rates for these neighborhoods, we chose (as we did in a previous study) to study 5-year rather than 1-year periods.10 By studying 2 such periods (1988–1992 and 1993–1997), we can learn whether the relation between infant mortality rate and an income-related measure changes from 1 period to the next. Data for births and infant deaths are as comparable as one can find in making international comparisons of infant mortality rates.2–4
Research Questions
Figure 1
shows the infant mortality rate distributions in the 4 cities for both periods. Except for Tokyo, median neighborhood rates fell between the 2 periods. For both periods, the Manhattan distribution exhibits greater spread than those of the other cities and is also more skewed in the direction of high rates. However, the figure reveals nothing about the relationship of income and infant mortality rate in the cities. Do lower income neighborhoods of a city tend to have higher rates than neighborhoods in the rest of the city? If so, does the magnitude of the difference change from 1 period to the next?
|
Because deaths is a nonnegative count variable, a Poisson regression model was a possibility. However, to account for greater variation than in a true Poisson process, we used instead a maximum likelihood negative binomial regression model that constrains the predicted number of deaths to a nonnegative number. Number of deaths is the response, number of births is the exposure, and an income-related variable is the explanatory variable.
For income, a similar measure of pretax, average household income, by neighborhood, is available for Manhattan, Paris, and Tokyo, so we began with this variable for these 3 cities. Since household income data are not available in the United Kingdom, for London, following British custom, we began with "deprivation" indices in place of direct income measures. Income figures for Manhattan, Paris, and Tokyo are in dollars, francs, and yen, respectively. For London, we began with the Carstairs Deprivation Index (http://census.ac.uk/cdu/Datasets/1991Census_datasets/Area_Stats/Derived_data/Deprivation_scores/Carstairs_index.htm) for period 1 and the Department of the Environment, Transport and the Regions Deprivation Index (http://www.detr.gov.uk) for period 2.
Using income and 2 different deprivation indices as explanatory variables in the models would make London and the other 3 cities hard to compare. Therefore, we used income and the deprivation indices to define an indicator variable that we used as the explanatory variable in the models. For Manhattan, Paris, and Tokyo, we let income (I) = 0 for a neighborhood if it was in the lower income quartile of neighborhoods. These are the 2 lowest income neighborhoods in Manhattan, the 5 lowest in Paris, and the 3 lowest in Tokyo. For the other neighborhoods in these cities, we let I = 1. For London, we let I = 0 for each of the 4 neighborhoods in the highest deprivation index quartile. For the other neighborhoods of London, we let I = 1. If both deprivation indices in London included the 4 lowest income neighborhoods in their upper deprivation quartile, our combination of income and deprivation indicators selected the lower income quartile of neighborhoods for all 4 cities.
We reported not the estimate of the underlying coefficient of I but the exponential of the estimate (i.e., the estimated incident rate ratio). The incident rate ratio here is the ratio of the value of the infant mortality rate in the rest of the city to its value in the low-income (or high-deprivation) neighborhoods. Our null hypothesis in each city and period was that the true value of the incident rate ratio for I is 1, that is, that there was no difference in infant mortality rates between the low-income (or high-deprivation) neighborhoods and those of the rest of the city. Our alternative hypothesis was that the incident rate is less than 1, that is, the low-income (or high-deprivation) neighborhoods have higher infant mortality rates than those in the rest of the city.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
), all of the incident rate ratios were less than 1, as expected. That is, with births fixed, shifting a neighborhood out of the lower income (or higher deprivation) quartile lowered its number of deaths or infant mortality rate. More income for the neighborhood was associated with a lower infant mortality rate.
|
In London, we estimated that the low-deprivation neighborhoods had a rate 94% of the rate in the high-deprivation neighborhoods in the first period and 82% in the second period. However, only the second period estimate was statistically significant at the 5% level.
In Paris, we found no difference in the low-and high-income neighborhood rates in the first period. We estimated that the high-income neighborhood rate was 87% of the low-income neighborhood rate in the second period, a figure that just missed statistical significance at the 5% level.
Finally, in Tokyo we found no significant difference (at the 5% level) between low-and high-income neighborhood rates in either period.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
We found that the evidence of higher infant mortality rates in low-income neighborhoods, when births are fixed, was especially strong in Manhattan. Central and East Harlem were the Manhattan neighborhoods in the lower income quartile, and they were also the neighborhoods with the highest infant mortality rates. A causal reading of this finding suggests a clear policy strategy. To lower infant mortality rates in Central and East Harlem, New York City should promote economic development in these neighborhoods and thereby raise their average income levels.
The problem, of course, is that even after controlling for births, an association between income level and infant mortality rate does not establish that the former causes the latter. Even if our Manhattan model was perfectly specified, no conclusion that low income causes more infant deaths is warranted because our data were ecological. The infants who died in the low-income Manhattan neighborhoods could have been children of the higher income residents of those neighborhoods. Only if our data were at the individual level and experimental—not observational—would our findings support a causal argument. Nonetheless, our regression results and broader comparative analysis raise at least 3 important questions for debates about the direction of policy in Manhattan:
First, what characteristics of high-infant-mortality, low-income neighborhoods, other than insufficient income, contribute to raising infant mortality? Are such neighborhoods characterized by inadequate provision of family planning, prenatal care, and other health care services leading to low levels of maternal health?
The New York Department of Health and Mental Hygiene (DHMH) recently introduced several new initiatives all of which suggest an affirmative answer to this question. For example, after an internal study that reported an 80% rate of unintended pregnancies in neighborhoods with high infant mortality rates,16 the DHMH established a citywide family planning initiative. In addition, the DHMH established an infant mortality case review committee to examine the causes of each infant death and now is attempting to coordinate citywide and regional perinatal forums to improve access to care for high-risk mothers and newborns. Perhaps most important of all, the DHMH established district public health offices in high-infant-mortality neighborhoods, including 1 for East and Central Harlem, whose responsibilities include leveraging resources and improving coordination of DHMH programs. For example in Central Harlem a strategic action committee designed a range of interventions that address health care as well as social factors related to mothers at risk.
Second, do high-infant-mortality, low-income neighborhoods reflect patterns of racial segregation and other forms of discrimination that might affect both the incomes and access to health care of minority women in Manhattan?
There is abundant evidence that race is an important determinant of infant mortality in the United States.17,18 Ellen’s evidence on this issue has the strength of analyzing individual rather than ecological data, but because the Internal Revenue Service does not disclose household income data at the individual level, the relative role of race and income level as sources of health outcomes cannot be explored.19 Had we included percentage of African Americans in our Manhattan model, the incident rate ratio for our income indicator might well have been 1.
Third, why do inequalities among high-and low-infant-mortality Manhattan neighborhoods remain so flagrant in spite of the well-known decrease in the overall citywide rate?
Based on our findings, one might suspect that Manhattan provides inadequate levels of health care access to mothers in poorer neighborhoods. New York State’s Prenatal Care Assistance Program (PCAP) became part of the Medicaid program after expansion in income eligibility thresholds in 1990. Nevertheless, Joyce’s evaluation of the PCAP’s impact on birth outcomes of Medicaid recipients in New York City suggests that expansion of prenatal care for women at risk is an inadequate strategy with which to achieve significant birth outcomes.20 A broader range of services, including family planning, comprehensive health care coverage, and targeted outreach efforts would probably be more effective than the PCAP. In this sense, the absence of universal health insurance in the United States exacerbates the problem of access in comparison to what Joseph White has called the "international standard."21 For example, Great Britain ensures health care coverage under its National Health Service; France and Japan ensure universal coverage under their national health insurance programs. Perhaps the extent to which a country lacks a commitment to universal coverage strengthens the connections between infant mortality and income.
Most research on the determinants of infant mortality highlights the importance of 3 types of variables in explaining variation in infant mortality rates across geographic units: material conditions (income and deprivation indicators, including race), income inequalities, and levels of available health services for mothers.22–33 The research is less clear on the relative importance of these variables. Moreover, the range of relevant health services is wide. It includes health education, contraception services, pregnancy counseling and abortion services, prenatal care (particularly for high-risk mothers), routine primary health care, and neonatal medical care and follow-up of high-risk mothers after birth.
Because we have neither a measure of income inequality nor a measure of health service use for each of our neighborhoods, our analysis does not shed light on the relative importance of neighborhood average income, health service use, and income inequality on infant mortality. Just as the absence of race in our Manhattan model could account for the importance of income there, the absence of income inequality and health service use in our models could account for the significant effect of income in Manhattan in contrast to Paris and Tokyo.
We know that local governments in Paris, Tokyo, and London operate nationally funded programs to identify high-risk mothers and offer them special services.34–36 In Paris, Tokyo, and London (albeit less so), there are aggressive efforts, at the neighborhood level, to follow all mothers in the course of their pregnancies and after birth. Moreover, in Paris, there are even financial incentives from the central government—the Protection Maternelle et Infantile—for mothers to seek out these services.37,38
Unfortunately, comparable data across all 4 cities on the extent and effectiveness of these services and on the spatial distribution of primary health care and public health professionals, by neighborhood units, are difficult to obtain. Could differences in the distribution of services and personnel responsible for them explain the differences we found for the role of income or deprivation among these cities? Only models that include all 3 variables at once—income level, income inequality, and a measure of accessible services—can begin to answer this question. In the meantime, there is increasing recognition at the New York City DHMH, and within the new Strategic Health Authorities responsible for London, that reducing disparities among neighborhood infant mortality rates will require intense targeting of high-rate neighborhoods and disproportionate resources directed to them.
|
Acknowledgments |
|---|
We thank Michael Gusmano and 3 anonymous reviewers for helpful suggestions, and Robert N. Butler, Françoise Forette, Sally Greengross, Dominique Jolly, Philip R. Lee, Roland Moreau, and Shigeo Morioka for helping us incubate the idea of this project from its most embryonic stage. We also thank Gabriel Montero who assisted with the preparation of our data set posted on the International Longevity Center Web site.
Finally, in each city, we are grateful for assistance from the following colleagues in obtaining and interpreting publicly available data on infant mortality and income: Melanie Besculides, Birgit Bogler, Tamara Dumanovsky, Fabienne Laraque, Kai-Lih Liu, all at the New York City Department of Health and Mental Hygiene (New York); Justine Fitzpatrick, London Health Observatory (London); Marie Claude Bonnefoi, Christine Grouas, Robert Poinsard (Paris Department of Health and Social Affairs), Eric Jougla (INSERM), Jeanne Fagnani (CNRS), and Diane Slama-Lequet (DRESS) (Paris); and Yukiko Kudo, Yoshiko Yamada (International Longevity Center–Japan), and Keiko Honda (International Longevity Center–USA) (Tokyo).
|
Footnotes |
|---|
Human Participant Protection
Local institutional review board approval was not sought, because this study was a secondary analysis of publicly available data that contained no individual identifying information.
Accepted for publication March 6, 2004.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
2. Sepkowitz S. International rankings of infant mortality and the United States’ vital statistics natality data collecting system: failure and success. Int J Epidemiol. 1995;24:583–588.
3. Howell E, Blondel B. International infant mortality rates: bias from reporting differences. Am J Public Health. 1994;84:850–852.
4. Liu K, Moon M, Sulvetta M, Chawla J. International infant mortality rankings: a look behind the numbers. Health Care Financ Rev. 1992;13:105–118.[Medline]
5. Hogue CJ, Hargraves MA. Class, race and infant mortality in the United States. Am J Public Health. 1993;83:9–12.
6. Hall P. The World Cities. 3rd ed. London, UK: Weidenfeld and Nicholson; 1984.
7. Sassen S. The Global City: New York, London, Tokyo. 2nd ed. Princeton, NJ: Princeton University Press; 2001.
8. Scott A. Global City-Regions: Trends, Theory, Policy. New York, NY: Oxford University Press; 2001.
9. Rodwin VG, Gusmano MK. The World Cities Project: rationale and design for comparison of megacity health systems. J Urban Health. 2002;79:445–463.[CrossRef][ISI][Medline]
10. Neuberg L, Rodwin V. Neighborhood matters—infant mortality rates in four cities: London, Manhattan, Paris, and Tokyo. Indicators. 2002–3;2:15–38.
11. Savitch HV. Post-Industrial Cities: Politics and Planning in New York, Paris and London. Princeton, NJ: Princeton University Press; 1988.
12. World Cities Project Fact Sheet. International Longevity Center-US. Available at: www.ilcusa.org/_lib/pdf/wcp_fact.pdf. Accessed September 28, 2004.
13. Sassen S. Economic restructuring as class and spatial polarization. In: Sassen S, ed. The Global City: New York, London, Tokyo. 2nd ed. Princeton, NJ: Princeton University Press; 2001:251–325.
14. Preteceille E. La ségregation sociale dans les grandes villes. Paris, France: Documentation Française, Problèmes Politiques et Sociaux No. 684; 1992.
15. Hamnett C. Social polarization in global cities: theory and evidence. Urban Stud. 1994;3:401–424.[CrossRef]
16. Besculides M, Laraque F. Unintended pregnancies among the urban poor. J Urban Health. 2004;81: 340–348.[CrossRef][ISI][Medline]
17. Racial and ethnic disparities in infant mortality rates: 60 largest US cities, 1995–1998. MMWR Morb Mortal Wkly Rep. 2002;51:329–332.[Medline]
18. Hogue CJ, Hargraves MJ. Class, race and infant mortality in the United States. Am J Public Health. 1993;83:9–12.
19. Ellen I. Is segregation bad for your health? The case of low birth weight. In: Brookings-Wharton Papers on Urban Affairs. Washington, DC: Brookings Institution Press; 2000.
20. Joyce T. Impact of augmented prenatal care on birth outcomes of Medicaid recipients in New York City. J Health Econ. 1999;18:31–67.[CrossRef][ISI][Medline]
21. White J. Competing Solutions: American Health Care Proposals and International Experience. Washington, DC: Brookings Institute; 1995.
22. Séguin L, Xu Q, Potvin L, Zunzunegui MV, Frohlich KL. Effects of low income on infant health. Can Med Assoc J. 2003;168:1533–1538.
23. DiLiberti J. The relationship between social stratification and all-cause mortality among children in the United States: 1968–1992. Pediatrics. 2000;105:e2.
24. Bird ST, Bauman KE. State-level infant, neonatal, and postneonatal mortality: the contribution of selected structural socioeconomic variables. Int J Health Serv. 1998;28:13–27.[ISI][Medline]
25. Pearce N, Davey Smith, G. Is social capital the key to inequalities in health? Am J Public Health. 2003;1: 122–129.
26. Davey Smith G. Income inequality and mortality: why are they related—income inequality goes hand in hand with underinverstment in human resources. BMJ. 1996;312:987–988.
27. Ben-Shlomo Y, White IR, Marmot M. Does the variation in the socioeconomic characteristics of an area affect mortality? BMJ. 1996;312:1013–1014.
28. Kawachi I, Kennedy BP, Wilkinson RG, eds. The Society and Population Health Reader, Vol. I: Income Inequality and Health. New York, NY: The New Press; 1999.
29. Mellor JM, Milyo J. Reexamining the evidence of an ecological association between income inequality and health. J Health Polit Policy Law. 2002;3:487–522.
30. House JS. Relating social inequalities in health and income. J Health Polit Policy Law. 2001;26: 523–532.[CrossRef][ISI][Medline]
31. Kawachi I, Blakely T. When economists and epidemiologists disagree. . . . J Health Polit Policy Law. 2001;26:534–541.
32. Daniels N, Kennedy B, Kawachi I. Justice is good for our health. Boston Rev. 2000;25:4–21.
33. Sohler N, Arno P, Chang C, Fang J, Schecter C. Income inequality and infant mortality in New York City. J Urban Health. 2003;80:650–657.[CrossRef][ISI][Medline]
34. Child Health in 1990: The United States compared to Canada, England and Wales, France, the Netherlands and Norway. Proceedings of a conference: Washington, DC; March 18–19, 1990. Pediatrics. 1990;86 (suppl, part 2):1025–1127. Theme issue.
35. Leppert PC. An analysis of the reasons for Japan’s low infant mortality rate. J Nurse Midwifery. 1993;38: 353–357.[CrossRef][ISI][Medline]
36. Richardson G. A Welcome for Every Child: How France Protects Maternal and Child Healt. A New Frame of Reference for the United States. Arlington, Va: National Center for Education in Maternal and Child Health; 1994.
37. Bergman B. Saving Our Children from Poverty: What the United States Can Learn from France. New York, NY: Russell Sage; 1996.
38. Rodwin VG. The health care system under French national health insurance: lessons for health reform in the United States. Am J Public Health. 2003; 93:31–37.
This article has been cited by other articles:
|
|
 |
|
  D. Weisz, M. K. Gusmano, V. G. Rodwin, and L. G. Neuberg Population health and the health system: a comparative analysis of avoidable mortality in three nations and their world cities Eur J Public Health, April 1, 2008; 18(2): 166 - 172. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. K. Gusmano, V. G. Rodwin, and D. Weisz A new way to compare health systems: avoidable hospital conditions in Manhattan and paris. Health Aff., March 1, 2006; 25(2): 510 - 520. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
