Community Health Urbanisation heart disease mortality ... · African Americans in the South suffer greater cardiovascular disease mortality than African Americansin otherregions."8

25ournal of Epidemiology and Community Health 1996;50:252-257

Urbanisation and coronary heart diseasemortality among African Americans in the USSouth

Elizabeth Bamett, David Strogatz, Donna Armstrong, Steve Wing

AbstractStudy objective - Despite significant de-clines since the late 1960s, coronary mor-tality remains the leading cause of deathfor African Americans. African Am-ericans in the US South suffer higher ratesof cardiovascular disease than AfricanAmericans in other regions; yet the mor-tality experiences of rural-dwelling Af-rican Americans, most ofwhom live in theSouth, have not been described in detail.This study examined urban-rural differ-entials in coronary mortality trends amongAfrican Americans for the period 1968-86.Setting - The United States South, com-prising 16 states and the District of Co-lumbia.Study Population - African American menand women aged 35-74 years.Design - Analysis of urban-rural differ-entials in tenmporal trends in coronarymortality for a 19 year study period. Allcounties in the US South were groupedinto five categories: greater metropolitan,lesser metropolitan, adjacent to metro-politan, semirural, and isolated rural.Annual age adjusted mortality rates werecalculated for each urban status group. In1968, observed excesses in coronary mor-tality were 29% for men and 45% forwomen, compared with isolated ruralareas. Metropolitan areas experiencedgreater declines in mortality than ruralareas, so by 1986 the urban-rural differ-entials in coronary mortality were 3% formen and 11% for women.Conclusions - Harsh living conditions inrural areas of the South precluded im-portant coronary risk factors and con-tributed to lower mortality rates comparedwith urban areas during the 1960s. Thedramatic transformation from an agri-culturally based economy to manu-facturing and services employment overthe course of the study period contributedto improved living conditions which pro-moted coronary mortality declines in allareas of the South; however, the most fa-vourable economic and mortality trendsoccurred in metropolitan areas.

(Jr Epidemiol Community Health 1996;50:252-257)

Coronary heart disease remains the leading causeof death among African Americans,' who suffera disproportionate burden of premature deathscompared with European Americans.2 Trends in

coronary mortality in the United States havechanged remarkably over the course of the 20thcentury. Until the mid 1960s, coronary mortalityamong African Americans steadily increased forthe nation as a whole. After the late 1960s, adownward trend began that continued through1986 at the national level.'Favourable national level trends may, how-

ever, obscure less favourable trends in certainregions and localities. The relative importanceof known risk factors in the aetiology of cor-onary heart disease may vary across differentpopulations and localities.3 Studies which dis-aggregate national trends to describe and ana-lyse geographic variation are crucial both foraetiological inquiry and for planning publichealth interventions.

In the 19th and early 20th centuries, totalmortality among African Americans was sig-nificantly higher in urban areas than in ruralareas, due to crowded and inadequate housing,poor sanitation systems, and poverty.4 There isevidence that as the overall standard of livingof African Americans improved through themid-20th century, the urban-rural gap in mor-tality narrowed.4 However, despite the re-duction in geographic inequality for totalmortality, an increase in geographic inequalityfor coronary mortality among African Am-ericans was observed for the period 1962-82.'Urban areas in the United States differ from

rural areas both economically and culturally,&8and various aspects of the social environmentsof both urban and rural areas are expected toinfluence coronary heart disease among AfricanAmericans, but not necessarily in a consistentmanner. Urban areas are characterised bygreater economic development and higherlevels ofcommunity resources, which have beenshown to be related to lower levels of coronaryheart disease mortality among EuropeanAmericans.910 At the same time, family andcommunity organisation within the cultural mi-lieu of rural areas has been shown to haveimportant health benefits, despite lower eco-nomic resources."'-" In addition, the transitionfrom a rural to an urban way of life has beenshown to have negative health effects, leadingto increased coronary heart disease amongEuropean Americans in the US South.'4"

In this study, we examined the associationbetween urban status of counties and trends incoronary heart disease mortality among AfricanAmerican men and women living in the USSouth during the years 1968-86. Few studiesof geographic variation in coronary mortalityamong African Americans have been done.'6'7

Prevention ResearchCenter, West VirginiaUniversity, PO Box9005, Morgantown,WV 26506-9005 USAE Bamett

Department ofEpidemiology, Schoolof Public Health,SUNY-Albany, USAD Strogatz

Department ofEpidemiology, Schoolof Public Health,UNC-Chapel Hill,USAD ArmstrongS Wing

Correspondence to:Dr E Barnett.Accepted for publicationNovember 1995

252

on June 17, 2020 by guest. Protected by copyright.

http://jech.bmj.com

/J E

pidemiol C

omm

unity Health: first published as 10.1136/jech.50.3.252 on 1 June 1996. D

ownloaded from

Urbanisation and coronary heart disease

An important strength of this study was ourexamination of coronary mortality among Af-rican Americans living in remote rural areas,mortality rates for this group have not pre-viously been reported.We focussed on the South for several reasons.

African Americans in the South suffer greatercardiovascular disease mortality than AfricanAmericans in other regions."8 In addition, over

50% of all African Americans and over 90%of rural African Americans lived in the Southduring the study period. Finally, communitiesin the South share a common history and pat-tern of population distribution and economicdevelopment. From the 1960s to the 1980sthere was a dramatic transformation in therural South from agriculturally based localeconomies to manufacturing based eco-

nomies.619-21 The proportion of African Am-ericans living in rural areas also declined duringthis time period. We will discuss the potentialimpact of economic and cultural change inSouthern communities on coronary risk factorsand mortality for African Americans.

MethodsThe study population consisted of AfricanAmerican men and women aged 35-74 years

living in the United States South during theyears 1968-86. The South, as defined by theUS census Bureau includes 16 states and theDistrict of Columbia. Age and gender specificpopulation totals for each county in each studyyear were drawn from Census Bureau in-tercensal population estimates.

Deaths from coronary heart disease were

counted, based on the underlying cause ofdeath listed on the death certificate. Since theInternational Classification of Disease (ICD)22 3

was revised midway through the study period,we attempted to use cause of death codes thatwould provide good comparability between the8th (1968-1978) and 9th (1979-) ICD re-

visions. Based on previous research,2425 thefollowing codes were chosen: for 1968-78,codes 410-413 (ischaemic heart disease) were

used, and for 1978-86, codes 410-414 (isch-aemic heart disease), 402 (hypertensive heartdisease), and 429-2 (cardiovascular disease un-specified) were used. Death certificate datawere obtained from the National Center forHealth Statistics.The classification ofthe urban status ofcoun-

ties was based on a scheme developed by theInformation Sciences Research Institute of theNational Center for Health Statistics.26 Thisscheme assigned each county to one of fiveurbanisation categories based on metropolitanstatus, population size of the county, and itsrelative proximity to, or isolation from, majormetropolitan areas. The assignment of countiesto urban status categories was based on 1980census data, and the relative urban status of

counties was assumed to be constant through-out the study period.The greater metropolitan group consisted of

counties within Standard Metropolitan Stat-istical Areas (SMSAs) with a total population ofone million or more. Lesser metropolitan counties

were counties within SMSAs with total popu-lation less than one million. The adjacent groupwas composed of all non-metropolitan countieswhich were adjacent to metropolitan countiesand had easy access to the central city of themetropolitan area. Semi-rural counties werenon-metropolitan, not adjacent and had anurban population of at least 2500, and isolatedrural counties were not adjacent to metro-politan areas and had an urban population lessthan 2500. An urban status code was assignedto each of the 1391 counties in the South.Data on unemployment, median family in-

come and white collar employment were ob-tained from the US census for each county,and population weighted averages were com-puted to provide summary measures for eachof the five urban status groups. Data on theindustrial sector of African American workerswere not available on standard Census Bureaucomputer tapes for counties in 1960 and 1970.We used published census reports to obtainindustrial sector for African American workers(aged 16 and older) in the South for urbanand rural areas.'729 The rural population wasdefined by residence outside of a place with apopulation of 2500 or more. The urban-ruraldichotomy used by the Census Bureau was notbased on county of residence, but rather onplace of residence within counties, and so wasnot directly comparable with the five levels ofurbanization used in our mortality analyses.

ANALYSISMortality analyses were conducted separatelyfor men and women. Coronary heart diseasedeaths and population counts for five year agegroups from 35 to 74 were summed withinurban groups. Direct age adjusted rates werecalculated for each urban status category forthe years 1968-86. The 1970 US populationwas used as the standard population for ageadjustment. For each study year, the mortalityrate for a particular urban status group rep-resented an average of the coronary mortalityexperience of all the counties in that group.Graphs of coronary mortality trends from1968-86 for each of the urban status groupswere created. Small populations in rural areasled to some temporal instability in the rates;therefore we graphed three-year running av-erage rates for all urban status groups. For1968, the average of the rates for 1968 and1969 was graphed and for 1986 the average ofthe rates for 1985 and 1986 was graphed. Totaldeclines in coronary mortality were calculatedusing the observed age adjusted rates for 1968and 1986.

ResultsThe majority of African Americans aged 35-74(63 9%) lived in metropolitan counties in 1968.By 1986 the proportion of African Americansin the study population who lived in semiruralor isolated rural areas had dropped from 18-3%to 15-0%, while the proportion living in metro-

253

on June 17, 2020 by guest. Protected by copyright.

http://jech.bmj.com

/J E

pidemiol C

omm

unity Health: first published as 10.1136/jech.50.3.252 on 1 June 1996. D

ownloaded from

Barnett, Strogatz, Armstrong, Wing

Table 1 County urbanisation and local economic resources in the United States South

Greater Lesser Adjacent to Semirural Isolated Totalmetropolitan metropolitan metropolitan rural

No of counties 78 251 476 392 194 1391% of African American population aged 35-74 in

each area in 1986 32-0 37-3 15-7 12 9 2-1 100Median family income 1960 ($, US) 5828 4828 3350 3307 2478 4483Increase in median family income 1960-80 ($, US) 17 006 13 590 12 249 11 563 10 248 13 991% employed in white collar jobs 1960 46-6 39-8 27-6 29-7 23-2 37 1Increase in % employed in white collar jobs 1960-80 12-6 12 6 11-5 10.9 9 7 12-1% unemployed 1960 4-8 5-0 5-3 6-1 6-4 5-2Change in % unemployed 1960-80 -0 1 +0-8 + 1-4 +1 0 +2-3 +0 7

politan areas had increased to 69-3% (table 1).Median family incomes in 1960 for all raceswere more than twice as high in greater met-ropolitan areas as an isolated rural areas (table1). Although family incomes increased from1960 to 1980 in all areas, inequalities betweenmetropolitan areas and rural areas were notreduced. African American median family in-come for the South as a whole was 54% ofwhite median family income in 1967,3°and 56% of white median family income in1980.3' Similar associations with county urb-anisation were observed for rates ofwhite collaremployment and unemployment. In 1960,46-6% of the labour force in greater metro-politan areas were employed in white collaroccupations, compared with only 23-2% ofworkers in isolated rural areas. White collaremployment increased in all areas, with thegreatest increases in metropolitan areas. Un-employment rates in 1960 were lowest in met-ropolitan areas, and highest in semirural andisolated rural areas. From 1960 to 1980, greatermetropolitan areas experienced no net increasein the rate of unemployment, while in isolated

1960 0 2 20..:.....

i 7 4

1970 17 36 15 10 99

1980 963 11 1 -

0 20 Z.r t60 0. 00

Figure 1 Leading industries ofAfrican American worn1960-80.

17 6 6 4:: : ::::

1970 18 25 15

1980-8 _2 11-1980 l8_2 6 tl9. 1

Hospitalls, 1htealithservicesEduCatIase(:) vFiO- \ icesTi-a nspe-tation,

co-blin-ilitimitisRetail, wholeszile tr-ade

nIst[rI ct ti a

AgriceLiit re. forestr,vfisher-ie. m aalilpll

ters in the rural US South,

Hospitals,h;iealthIse("r-vicesEducational serVicesTr-ansportattio(nCoal)II11ncLin1 catIos.:Lsblhli Utilitie'sRletaili wholesale tradeMa 1L faCt l ilncC{e St r lItionIMPrlivate h-10LIS(ehO1CdIS,per-sonal seo-vices

rural areas unemployment increased 37%, from6.4% to 8-7%.From 1960 to 1980, the total farm popu-

lation in the South declined 75%, while theAfrican American farm population for the en-tire United States declined 87% (both figuresunadjusted for population growth).? Thetransformation ofagriculturally based Southernrural economies resulted in changes over timein the industries in which African Americansfound employment. In 1960, 40% of all AfricanAmerican workers living in rural areas of theSouth were employed in agriculture, forestry,fisheries, or mining (fig 1). Private householdsand personal services accounted for another20% of rural employed African Americans.Although slavery had ended almost 100 yearspreviously, in 1960 most rural African Am-ericans were still employed in agriculture orprivate households. By 1980 the industrial dis-tribution of the rural African American labourforce had dramatically changed. Only 9% ofrural African Americans worked in agricultureor other extractive industries, and only 6%worked in private households or personal ser-vices. Over 43% of employed African Am-ericans now worked in manufacturing,compared with 22% in 1960. Retail and whole-sale trade and hospitals and health services alsoincreased their share of Southern rural labourmarkets between 1960 and 1980.

In urban areas of the South the leadingindustries of African American workers alsoshifted from 1960 to 1980, although not asdramatically as in rural areas (fig 2). Privatehousehold and personal service workers ac-counted for 31% of employed urban AfricanAmericans in 1960, but accounted for only 8%in 1980. Employment in most other industriesincreased moderately from 1960 to 1980. Ser-vice sector employment predominated in 1980,with only 25% of urban African Americansworking in manufacturing or construction.The relationship between urban status and

levels of coronary mortality was the same forAfrican American women and men (figs 3 and4). At the beginning of the study period thehighest coronary mortality was found in greatermetropolitan areas, followed by lesser met-

ropolitan areas. Counties adjacent to metro-politan areas and semirural areas had similarrates, and the lowest coronary mortality ratesoccurred in isolated rural areas. For AfricanAmerican men living in greater metropolitanareas in 1968, there was a 29% excess in cor-

onary heart disease mortality compared withisolated rural areas, while for African American

1960K

0 20 40 ho 50 100

Figure 2 Leading industries ofAfrican American workers in the urban US South,1960-80.

254

on June 17, 2020 by guest. Protected by copyright.

http://jech.bmj.com

/J E

pidemiol C

omm

unity Health: first published as 10.1136/jech.50.3.252 on 1 June 1996. D

ownloaded from

Urbanisation and coronary heart disease

Year

Figure 3 Coronary heart disease mortality among African American men aged 35-74years in the US South.

-- Greater metropolitan

o^ -U-w Lesser metropolitan

C- Adjacent to metropolitan

,500< Semi-rural~~~Isolated rural

In

.2. 300

a)

< 2001968 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 81

Year

Figure 4 Coronary heart disease mortality among African American women aged 35-7years in the US South.

women there was a 45% excess. Over the cours

of the study period, coronary mortality trendfor the five urban status groups converged foboth women and men, so that in 1986 thexcess in CHD mortality for greater metropolitan areas compared with isolated rural area

was only 3% for men and 11% for women.Total declines in coronary mortality over th4

19-year study period were substantial for botlAfrican American women and men (table 2)Although African Americans in isolated ruraareas experienced the lowest rates of coronarmortality, the greatest declines in mortality oc

curred in greater metropolitan areas of thSouth. There was a clear urban-rural gradienin both total percentage declines and total absolute declines in coronary mortality, with isolated rural areas experiencing the smallesdeclines. In all urban status groups, AfricarAmerican women experienced greater relativ4but smaller absolute declines in coronary mor

tality than African American men.

Table 2 Total declines in coronary mortality from 1968-86 for African Americans aged35-74 years in the United States South

Greater Lesser Adjacent Semi- Isolatedmetropolitan metropolitan metropolitan rural rural

% declineMen 41-3 41-8 39 3 36-1 26-3Women 50 3 49.7 46-0 40-8 34-7

Absolute decline(deaths/100 000)Men 359 362 305 276 178Women 291 281 227 203 138

DiscussionHistorically, coronary heart disease has beenpositively associated with industrialisation, so-cietal wealth, and higher social class.32 TheUnited States and western Europe continue toexperience dramatically higher rates than lessindustrialised and underdeveloped nations.However, the onset of decline in coronary mor-tality was accompanied by a change in thecharacter of coronary heart disease from a dis-ease of the affluent to a disease increasingly ofthe disadvantaged and poor.33 This trans-

6 formation reflects overall improvements instandards of living which led to changes in thematerial conditions of the poor and workingclass. In the first half of the 20th century, manyAfrican Americans lived in conditions whichmore closely approximated those of popu-

_ lations in underdeveloped areas than they re-sembled the material conditions of mostEuropean Americans.434 This was especiallytrue in the rural South, where poverty and alife ofhard physical labour precluded importantrisk factors for coronary heart disease - that is,a sedentary lifestyle and a diet rich in meat andfat. Baseline data from 1960 collected for theCharleston heart study and the Evans County,

3 Georgia heart study showed that rural African> Americans had lower mean cholesterol con-I centrations than urban African Americans, and6 rural African American women were much less

likely to smoke cigarettes than urban AfricanAmerican women.35-38 An analysis of data from

4 the national health and nutrition examinationsurveys I (1971-74) and II (1976-1980) foundthat the prevalences of hypertension amongblack men and women aged 45-54 years in the

e South were higher in metropolitan areas thanIs in non-metropolitan areas during both timeor periods.39e Many rural African American communities

were characterised by close-knit social supportIS networks and positive cohesive cultural en-

vironments which may have mitigated againste the alienation and psychosocial stress which

were more common in urban areas." 12 In ad-dition, because migration patterns both before

1 and during the study period were pre-Y dominantly from rural areas to urban areas,34

African American adults in isolated rural arease had probably spent their entire lives living in theLt same rural community. Many were primarily- employed in agriculture, or may have worked

part time on a family farm. The lower coronaryt mortality rates compared with metropolitann residents observed for this group in 1968 aree consistent with the lower prevalences of cor-

onary risk factors discussed above.In contrast, the African American residents

of metropolitan areas in our study were a moreheterogeneous group. Some of them may havelived their whole lives in cities, but given the

- large migration ofAfrican Americans from ruralto urban areas which began in the 1940s and

- continued through the 1960s, we assume thata significant proportion ofthese urban residentsspent their childhoods in rural areas. The neg-ative health effects of recent urbanisation forworking class European Americans in the South

- have been documented," 14 and we suggest that

255

on June 17, 2020 by guest. Protected by copyright.

http://jech.bmj.com

/J E

pidemiol C

omm

unity Health: first published as 10.1136/jech.50.3.252 on 1 June 1996. D

ownloaded from

Barnett, Strogatz, Armstrong, Wing

similar effects held true for African Americansin the early years of this study. Given thewidespread racism with which all African Am-ericans had to contend, the loss of a communitybased on kinship networks and personally sup-portive relationships may have been particularlydifficult for recent urban migrants. Psycho-social stress is associated both with coronaryheart disease and hypertension, an importantcoronary risk factor.4"43As the overall standard ofliving for all African

Americans improved in the latter half of the20th century, economic disparities between thewealthy and the poor did not disappear, butqualitative changes in the material conditionsof the poor and working class emerged whichcontributed to the urban-rural convergence incoronary mortality observed in this study. Im-provements in food distribution, relativelylower food costs, and the convenience of pro-cessed foods contributed to dietary change inthe rural South. Rural residents with improvedeconomic resources were able to eat meat andother high fat foods regularly, and were morelikely to lack physical fitness as a result ofthe transition from agricultural to low-levelmanufacturing and services employment.4445Urban African Americans in the South mayhave benefited more from the gains of thecivil rights movement in the 1960s than ruralAfrican Americans, through access to betterjobs, housing, health education, medical care,and health promoting activities such as leisuretime physical activity and participation in thearts and recreation.44

METHODOLOGICAL ISSUESA limitation of the population data used in thisstudy was the underenumeration of the AfricanAmerican population by the United States cen-sus in 1970 and 1980. Estimates of the extentof underenumeration by the 1970 census are10 1% for African American males and 5-3%for African American females for all ages com-bined.46 In the 1980 census there was a 7>5%underenumeration of African American malesand a 2-1% undercount of African Americanfemales.47 Within the age range 35-74 the de-gree of underenumeration was not consistent;for example, in 1980 there were large un-dercounts for African American men 45-54(15-1 %) and small overcounts for AfricanAmerican men 65-74 (4 3%).One effect of the census undercount of Af-

rican American populations is a probable over-estimation of mortality rates, especially forAfrican American men, assuming that deathswere not undercounted and age at death wasclassified properly. Bias may be present in thisstudy if the extent of mortality rate over-estimation varied by the urban status of coun-ties or by year. It has been suggested thatpopulation undercounts were greater in metro-politan areas,46 but there are no data availableto address this question. The possibility thatthe observed relationships between level of cor-onary heart disease mortality and urban statusof counties were the result of biased mortalityestimates can not be excluded. However, it

should be noted that approximately one-halfof the deaths in this study occurred amongAfrican Americans aged 65-74, for whom therewere actually small census overcounts in 1980.A second limitation and potential source

of bias in this investigation involves mis-classification of underlying cause of death ondeath certificates. Numerous studies at-tempting to assess the validity and reliability ofcause of death certification have been done.48Many have shown a lack of agreement betweenthe cause of death on the death certificate, andcause of death determined from other datasources: autopsies, medical records, hospitaldischarge records. A study comparing under-lying cause of death from the death certificateto multiple diagnoses in the hospital record for9724 deaths in Vermont found 87% agreementfor acute myocardial infarction (ICDA-8 410)but only 65% agreement for ischaemic heartdisease (ICDA-8 412). For all specific causesof death, relative agreement between deathcertificate coding and hospital discharge re-cords varied dramatically by hospital.49 This isindirect evidence that validity of death cer-tificate coding may vary geographically. How-ever, since a broad grouping ofICD codes wasused to define CHD in our study, classificationshould be better than in studies of single ICDcodes.A recent study showed that, nationwide, ap-

proximately 3% of deaths among African Am-ericans aged 35-74 in 1968 were coded to theICD cause of death category symptoms, signs,and ill-defined conditions.50 This proportion de-clined over time; in 1988, 1-7% of deathsamong men and 1-3% of deaths among womenwere coded to this category. There is evidenceto suggest that many of these deaths may havebeen sudden coronary deaths among adultswith no previous history of heart disease.50 Tothe extent that coronary deaths were mis-diagnosed and miscoded on the death cer-tificate, the mortality rates calculated in thisstudy are underestimates. In addition, if theproportion of all deaths among African Am-ericans in the South coded to ill-defined con-ditions varied by urban status, the potential forbias due to differential misclassification existsin this study. We indirectly evaluated the po-tential for information bias by calculating allcause mortality rates for each of the urbanstatus groups (data not shown) and found thesame urban-rural mortality gradient which weobserved for coronary heart disease mortality.

CONCLUSIONWhile coronary heart disease mortality hasbeen declining nationwide among African Am-ericans since the late 1960s, the rate of declinein coronary mortality slowed after 1975.24 Weobserved a similar deceleration for the South.In a recent study of coronary mortality trendsin North Carolina during the 1980s, there wasno decline for African American women from1980 to 1 988.5' If the current trends continue,a rural excess in coronary mortality for AfricanAmericans in the South will emerge. In orderto preserve the tremendous improvements in

256

on June 17, 2020 by guest. Protected by copyright.

http://jech.bmj.com

/J E

pidemiol C

omm

unity Health: first published as 10.1136/jech.50.3.252 on 1 June 1996. D

ownloaded from

Urbanisation and coronary heart disease

coronary mortality which have occurred overthe past 25 years, the mechanisms by whicheconomic and cultural resources of com-munities help prevent coronary heart diseaseneed to be clarified, with the goal of designinginterventions and public policy initiatives whichwill lessen geographic inequalities and improvethe public health.

The authors would like to thank Dr Janet Croft for providingrisk factor prevalence data from the Evans County and Charles-ton Heart Studies, and an anonymous reviewer for helpfulcomments on an earlier draft of this paper. At the time of thisstudy, Dr Barnett was supported by a graduate fellowship fromthe National Science Foundation. This study was part of alarger project, Community Structure and Cardiovascular Dis-ease Mortality Trends, funded by the National Heart, Lung,and Blood Institute (grant RO1-HL42320, Dr Wmg, principalinvestigator). An earlier version of this study was presented atthe 118th Annual Meeting of the American Public HealthAssociation.

1 Gillum RF. Coronary heart disease in black populations: I.

Mortality and morbidity. Am HeartJ_ 1982;104:839-51.2 Kapantais G, Powell-Griner E. Characteristics offpersons dying

ofdiseases ofthe heart: preliminary datafrom the 1986 nationalmortality followback survey. Advance data from vital andhealth statistics; no 172. Hyattsville, MD: National Centerfor Health Statistics, 1989.

3 Gary R. Madan JH, Kleinman JC. Regional variation inischemic heart disease incidence. Clin Epidemiol 1992;45:149-56.

4 Ewbank DC. History of black mortality and health before1940. Millbank Q 1987;65:100-28.

5 Wing S, Casper M, Davis W, Hayes C, Riggan W, TyrolerHA. Trends in the geographic inequality of cardiovasculardisease mortality in the United States, 1962-1982. SocSci Med 1990;30(3):261-6.

6 Falk WW, Lyson TA. High tech, low tech, no tech. Albany:State University of New York Press, 1988.

7 Farley R. The color line and the quality of life in America. NewYork: Oxford University Press, 1989.

8 Green GP. Finance capital and uneven development. Boulder,Co: Westview Press, 1987.

9 Wmg S, Dargent-Molina P, Casper M, Riggan W, Hayes CG,Tyroler HA. Changing association between communityoccupational structure and ischaemic heart disease mor-tality in the United States. Lancet 1987;ii: 1067-70.

10 Wing S, Barnett E, Casper M, Tyroler HA. Geographic andsocioeconomic variation in the onset ofdecline ofcoronaryheart disease mortality in white women. AmJ Public Health1992;82(2):204-9.

11 Cassel J, Patrick R, Jenkins D. Epidemiological analysis ofthe health implications of culture change: a conceptualmodel. Ann NYAcad Sci 1960;84:938-49.

12 Green V. The black extended family in the United States:some research suggestions. In: Shimkin DB, Shimkin EM,Frate DA. The extended family in black societies. Hague,Netherlands: Mouton DeGruyter, 1978;379-87.

13 House JS, Landis KR, Umberson D. Social relationshipsand health. Science 1988;241:540-5.

14 Tyroler HA, CasselJ. Health consequences ofculture change- II: the effect of urbanization on coronary heart diseasemortality in rural residents. J3 Chronic Dis 1964;17:167-177.

15 Cassel J, Heyden S, Bartel AG, Kaplan BH, Tyroler HA,Comoni JC, Hames CG. Incidence of coronary heartdisease by ethnic group, social class, and sex. Arch IntemMed 1971;128:901-5.

16 Leaverton PE, Feinleib M, Thom T. Coronary heart diseasemortality rates in United States blacks, 1968-1978: in-terstate variation. Am Heart 1984;108(3):732-737.

17 Kleinman JC, DeGruttola VG, Cohen BB, Madans J.Regional and urban-suburban differentials in coronaryheart disease mortality and risk factor prevalence. Jf ChronicDis 1981;34:11-19.

18 Manton KG, Patrick CH, Johnson KW. Health differentialsbetween blacks and whites: recent trends in mortality andmorbidity. Millbank Q 1987;65:129-199.

19 Rosenfeld S. A divided South. Southern Exposure 1983;14:10-17.

20 Rosenfeld SA, Bergmar EM, Rubin S. After the factories:changing employment patterns in the rural South. ResearchTriangle Park, NC: Southern Growth Policies Board,1985.

21 James DR. The transformation of the southern racial state:class and race determinants oflocal state structures. Amer-ican Social Reviews 1988;53:191-208.

22 International classification ofdisease. 8th Rev, adapted for usein the United States. Vol I.DHEW PHS Publication No1696, 1969.

23 International classification of diseases. Manual of the inter-national statistical classification of diseases, injuries, andcauses of death. Based on the recommendations of the9th revision conference, 1975, and adapted by the 29thWorld Health Assembly. Vol I. 1977.

24 Sempos C, Cooper R, KovarMG, McMillen M. Divergenceof the recent trends in coronary mortality for the fourmajor race-sex groups in the United States. Am 7 PublicHealth 1988;78(11): 1422-27.

25 Ingram DD, Gillum RF. Regional and urbanization differ-entials in coronary heart disease mortality in the UnitedStates, 1968-85. J Clin Epidemiol 1989;42(9):857-868.

26 Human Resources Administration. The area resource file:a health professional's planning and research tool. Wash-ington DC: DHHS Publication no (HRA) 81-9.

27 Bureau of the Census. 1960 census of population. Vol ICharacteristics of population, chapter C. Washington DC:US Government Printing Office, 1963.

28 Bureau of the Census. 1970 census of population. Vol ICharacteristics of population, chapter C. Washington DS:US Government Printing Office, 1973.

29 Bureau of the Census. 1980 census of population. Vol ICharacteristics of population, chapter C. Washington DC:US Government Printing Office, 1983.

30 US Bureau of the Census. Statistical abstract of the UnitedStates: 1969. Washington, DC: US Government PrintingOffice, 1969.

31 US Bureau of the Census. Statistical abstract of the UnitedStates: 1984. 104th ed. Washington: US GovernmentPrinting Office.

32 Marmot MG, Adelstein AM, Robinson N, Rose GA. Chang-ing social distribution ofheart disease. BMJ 1978;ii: 1109-12.

33 Wmg S. Social inequalities in the decline of coronary mor-tality. AmJrPublic Health 1988;78:1415-16.

34 Fusfield DR, Bates T. The political economy of the urbanghetto. Carbondale: University of Illinois Press, 1984.

35 Stevens J, Keil JE, Rust PF, Tyroler HA, Davis CE, GazesPC. Body mass index and body girths as predictors ofmortality in black and white women. Arch Intern Med1992;152: 1257-62.

36 Tyroler HA, Knowles MG, Wmg SB, Logue EE, Davis CE,Heiss G, Heyden S, Hames CG. Ischemic heart diseaserisk factors and twenty-year mortality in middle-age EvansCounty black males. Am HeartJr 1984;108:738.

37 Stevens J, Keil JE, Rust PF, Verdugo RR, Davis CE, TyrolerHA, Gazes PC. Body mass index and body girths aspredictors of mortality in black and white men. Am JEpidemiol 1992;135:1137-46.

38 Johnson JL, Heineman EF, Heiss G, Hames CG, TyrolerHA. Cardiovascular disease risk factors and mortalityamong black women and white women aged 40-64 yearsin Evans County, Georgia. Am J Epidemiol 1986;123:209-20.

39 Casper ML. Antihypertensive treatment and US trends instroke mortality, 1962-1980. Ann Arbor, Michigan: UMIDissertation Service; 1991. (Dissertation).

40 James SA, Hartnett SA, Kalsbeek WD. John Henryism andblood pressure differences among black men. J BehavMed 1983;6(3):259-78.

41 Harburg E, Erfurt JC, Chape C, Hauenstein LS, Schull WJ,Schork MA. Socioecological stressor areas and black-white blood pressure: Detroit. J Chronic Dis 1973;26:595-611.

42 Harburg E, Erfurt JC, Hauenstein LS, Chape C, Schull WJ,Schork MA. Suppressed hostility, skin color, and black-white blood pressure: Detroit. Psychosom Med 1973;35(4):276-96.

43 Landsbergis PA, Schurman SJ, Israel BA, Schnall PL, Hu-gentobler MK, Cahill J, Baker D. Job stress and heartdisease: evidence and strategies for prevention. New So-lutions 1993;3:42-58.

44 Cobb JC. Industrialization and southern society. LexingtonKY: The University Press of Kentucky, 1984.

45 Ford AM. Political economies of rural poverty in the South.Cambridge, MA: Ballinger Publishing Co. 1973.

46 Seigal JS. Estimates of coverage of the population by sex,race, and age in the 1970 census. Demography 1974;11:1-23.

47 US Bureau of the Census. Current population reports. SeriesP-23, no 115, Coverage of the national population in the1980 census, by age, sex, and race: preliminary estimates bydemographic analysis. Washington DC: US GovernmentPrinting Office, 1982.

48 Gittlesohn A, Royston P. Annotated bibliography of causeof death validation studies 1958-1980. Vital and HealthStatistics, Series 2, No 89. DHHS Publication No (PHS)82-1363, 1982.

49 Gittlesohn A, Senning J. Studies on the reliability of vitaland health records: I. Comparison of cause of death andhospital record diagnoses. Am J Public Health 1979;69:680-9.

50 Armstrong D, Wimg S, Tyroler HA. US mortality from ill-defined causes, 1968-1988: potential effects on heartdisease mortality trends. IntJ'Epidemiol 1995;24(3):522-7.

51 Bamett E, Casper M. Recent trends in coronary heart diseaseand stroke mortality among adults, North Carolina, 1980-1991. Raleigh, NC: Center for Health and EnvironmentalStatistics, no 98, 1995.

257

on June 17, 2020 by guest. Protected by copyright.

http://jech.bmj.com

/J E

pidemiol C

omm

unity Health: first published as 10.1136/jech.50.3.252 on 1 June 1996. D

ownloaded from