Area of residence and alcohol-related mortality risk: a five-year follow-up study

of 9
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Information Report
Category:

Philosophy

Published:

Views: 8 | Pages: 9

Extension: PDF | Download: 0

Share
Description
Area of residence and alcohol-related mortality risk: a five-year follow-up study
Transcript
  Area of residence and alcohol-related mortality risk:a five-year follow-up study add_3103 84..92 Sheelah Connolly, Dermot O’Reilly, Michael Rosato & Chris Cardwell Centre for Public Health, Queen’s University Belfast, ICSB, block B, RVH Site, Grosvenor Road, Belfast, UK  ABSTRACT Aims  To examine differences in alcohol-related mortality risk between areas, while adjusting for the characteristicsof the individuals living within these areas.  Design  A 5-year longitudinal study of individual and area characte-ristics of those dying and not dying from alcohol-related deaths.  Setting  The Northern Ireland Mortality study. Participants  A total of 720 627 people aged 25–74, enumerated in the Northern Ireland 2001 Census, not living incommunal establishments.  Measurements  Five hundred and seventy-eight alcohol-related deaths.  Findings  Therewas an increased risk of alcohol-related mortality among disadvantaged individuals, and divorced, widowed andseparated males.The risk of an alcohol-related death was significantly higher in deprived areas for both males [hazardratio (HR) 3.70; 95% confidence interval (CI) 2.65, 5.18] and females (HR 2.67 (95% CI 1.72, 4.15); however, onceadjustment was made for the characteristics of the individuals living within areas, the excess risk for more deprivedareas disappeared. Both males and females in rural areas had a reduced risk of an alcohol-related death compared totheircounterpartsinurbanareas;thesedifferencesremainedafteradjustmentforthecompositionof thepeoplewithinthese areas.  Conclusions  Alcohol-related mortality is higher in more deprived, compared to more affluent areas;however, this appears to be due to characteristics of individuals within deprived areas, rather than to some indepen-dent effect of area deprivation  per se . Risk of alcohol-related mortality is lower in rural than urban areas, but thecause is unknown. Keywords  Alcohol-related mortality, composition, context, deprivation, longitudinal analysis, Northern Ireland,urban/rural. Correspondenceto: SheelahConnolly,CentreforPublicHealth,Queen’sUniversityBelfast,ICSB,BlockB,RVHSite,GrosvenorRoad,BelfastBT126BJ,UK.E-mail: sheelah.connolly@qub.ac.ukSubmitted 1 December 2009; initial review completed 18 February 2010; final version accepted 14 June 2010 INTRODUCTION The effects of excess alcohol consumption are wide-ranging: in 2004 alcohol accounted for almost 4% of allglobal deaths, and an even higher proportion of theglobal burden of disease and injury [1]; its excess con-sumption impacts on crime and antisocial behaviour,work-placeproductivityandfamilyrelations,andcurrentestimatesputthecostof alcohol-relatedharminEnglandatabout£20billionperannum[2].Whiletherehasbeenmuch discussion about the need to tackle alcohol-relatedharm [3], this requires the identification of those mostlikely to drink to excess and suffer such harm. Only thencan alcohol prevention programmes be targeted atgroups most in need, and towards factors associated withharmful consumption patterns.Previous research has shown that alcohol-relatedmortality risk is associated with a number of individualand household characteristics including age, gender andmarital and socio-economic status [4–7].There has beenrelatively little research on the relationship between indi-vidual and area characteristics associated with alcohol-related mortality in the United Kingdom. One study inGreat Britain found that while alcohol-related mortalitywas higher for men in manual occupations its relativemagnitude depended upon age. For women in paidemployment there were no consistent class gradients [8].Another study, from England and Wales, has shown astrong positive association between alcohol-related mor-tality and area-based deprivation levels [4]. However, asthat study did not control for the characteristics of indi-viduals living in these areas, it is unclear whether thedifferences found across areas were due to area effects(context) or to variations in the characteristics of thepeople living within these areas (composition). Disentan-glingarea-basedeffectsfromindividualeffectsiscrucialif  RESEARCH REPORT  doi:10.1111/j.1360-0443.2010.03103.x© 2010 The Authors, Addiction © 2010 Society for the Study of Addiction  Addiction ,  106 , 84–92  appropriate policy responses for tackling alcohol-relatedharm are to be implemented; for example, if differencesacross areas are explained by the composition of theindividuals living in these areas, then the appropriateresponse would involve targeting at-risk individuals;however, if the differences are due (in some part) to char-acteristicsof thearea,thentheappropriateresponsemayinvolve attempting to change these contextual factors.Interest in the potential effects of area of residence onhealth has increased in recent years [9]. However, theliteraturehasnotyetreachedconsensusabouttheimpor-tance of place effects on health status. While one reviewof a number of studies examining the effects of neigh-bourhood and local area characteristics on healthoutcomes found that contextual effects were generallymodestandmuchsmallerthanindividualorcompositioneffects[10],itwasnotedthatmostof theincludedstudiesfocused upon all-cause mortality, which might be toogeneral to show any independent effects of area charac-teristics [11]. A study in Helsinki noted that while neigh-bourhood characteristics had modest independent effectsonmortalitywhencomparedtothecharacteristicsof theindividuals living within these areas there was some evi-dence to suggest that certain causes of death, includingalcohol-relateddeaths,wererelatedmorestronglytoareacharacteristics than others [12]. Further investigation inFinland found that while differences in alcohol-relatedmortalityratesbetweenareaswithdifferinglevelsof dep-rivation were attenuated by the inclusion of individuallevel and other area-based characteristics, significantdifferences in alcohol-related mortality rates were notexplained fully by the characteristics of the individualsin these areas [11].A variety of neighbourhood characteristics may beassociated with health outcomes, including deprivation,inequalityandsocialfragmentation.Inthisinitialanalysisof area effects on alcohol-related mortality risk in theUnited Kingdom, two area characteristics will be exam-ined—deprivation and urban/rural composition. Thesocio-economic structure of an area may influence anindividual’s risk of an alcohol-related death (over andabove the individuals’ own socio-economic status) in anumber of ways; for example, behavioural norms [11]maydifferinaffluentanddeprivedareas,andimpactuponan individual’s alcohol consumption, regardless of theindividual’sownsocio-economicstatus.Thestressassoci-atedwithlivinginadeprivedarea[9],withtheirgenerallyhigherlevelsof crimeanddegradationmayalsoimpactonalcohol consumption. Alternatively, more deprived areasmay have a greater supply of alcohol, or fewer treatmentoptions for those experiencing alcohol-related problems.Differences in alcohol-related mortality have beenfound between urban and rural areas [13,14]; such dif-ferences may be due to differing alcohol consumptionpatterns or to access or use of treatment for alcohol-related problems. Alcohol consumption may be expectedto be higher in rural areas, where there is often a lack of alternative social outlets, while the stress associated withphysical and social isolation could contribute to excessalcohol consumption. Alternatively, alcohol mortalitycould be lower in rural areas, as potentially strongcommunity ties encourage moderation. Interestingly, aninvestigation in Scotland suggested that families in ruralareas are more likely to support someone with alcoholdependency and will refuse professional help to avoidthe stigma associated with alcohol misuse [15].The aims of the study are twofold: to examine fordifferences in alcohol-related mortality risk betweendeprived and affluent areas, and urban and rural areas;and to determine whether differences are due to thecharacteristics of individuals living in these areas orto specific area effects. METHODS Study population The Northern Ireland Mortality Study (NIMS) is a pro-spective record linkage study of the whole NorthernIreland population enumerated at the 2001 Census. Aspart of the study, the Northern Ireland Statistics andResearch Agency (NISRA) linked the death records of allNorthern Ireland residents who died during the 5 yearsfollowing the census to their census record. This formeda longitudinal study linking 94% of all deaths occurr-ing over this period. Details of the linkage process aredescribed elsewhere [16]. These data were anonymized,held in a safe setting by NISRA and made available to theresearch team for this study.InkeepingwiththedefinitionprovidedbytheOfficeforNational Statistics (ONS) in the United Kingdom [17],alcohol-relatedmortalitywasdefinedasdeathsregisteredwith the following ICD-10 codes: F10—mental andbehavioural disorders due to use of alcohol; G31.2— degeneration of nervous system due to alcohol; G62.1— alcoholic polyneuropathy; I42.6—alcoholic cardiomy-opathy; K29.2—alcoholic gastritis; K70—alcoholic liverdisease; K73—chronic hepatitis, not elsewhere classified;K74—fibrosis and cirrhosis of liver (excluding K74.3-K74.5—biliary cirrhosis); K86.0—alcohol-inducedchronic pancreatitis; X45—accidental poisoning by andexposure to alcohol; X65—intentional self-poisoning byand exposure to alcohol; and Y15—poisoning by andexposure to alcohol, undetermined intent. Also, inkeeping with the ONS definition the causes listed wereincluded only when they appeared as the underlyingcause of death (rather than a contributory cause) on thedeath certificate. Area and alcohol-related mortality risk  85 © 2010 The Authors, Addiction © 2010 Society for the Study of Addiction  Addiction ,  106 , 84–92  Individual and household characteristics Allattributesof cohortmemberswereasdescribedonthecensus record. Marital status was grouped into three cat-egories (married or cohabiting; single; and separated,divorced or widowed). Household composition wascategorizedaseitherasingle-personhouseholdorhouse-holdswithatleasttwopeople.Socio-economicstatuswasassessed using housing tenure (categorized as owner-occupier; private renter; social renter), car availability(access to two or more cars; or at most one car), highestlevel of educational achievement [degree or higher;A-level plus; General Certificate of Secondary Education(GCSE) plus; GCSE; and no qualifications] 1 and NationalStatistics Socio-economic Classification (NSSEC) (profes-sional;intermediate;ownaccountworker;lowersupervi-sory, routine; not working; student). 2 As previous workhasshownthattenureandcaravailabilityarepotentiallybiased indicators of deprivation in the urban–ruralcontext[18],onlyeducationalattainmentandNSSECareincluded as indicators of socio-economic status in theurban–rural analysis. Morbidity was measured usingtwo self-reported health questions included in the 2001Census: presence of a limiting long-term illness (LLTI),eliciting a yes/no response; and general health (GH) inthe year preceding the census, allowing three responses(good, fairly good and not good health).As neither education nor social class were coded forthose aged 75 and over in the 2001 Census, the analysiswas restricted to those aged less than 75 years to ensurethat socio-economic factors could be applied to the wholeof the analysis population. Those living in communalestablishments at the time of the census were alsoexcluded, as area of residence would not be relevant forthis group. Area characteristics Two indicators relating to area of residence were derivedat census super-output area (SOA) level—SOAs are ageography designed for the collection and publication of small area statistics: these were deprivation and urban– rural composition. Deprivation was assessed as theproportions in the SOA living in households in receipt of means-tested benefits [19]. SOAs were ranked accordingto their level of deprivation and split into quintiles con-taining approximately equal proportions of the popula-tion. The urban–rural composition of area of residencewas assessed by assigning the SOA of residence for eachindividual to one of three broad settlement bands[20]—defined as ‘urban’, ‘intermediate’ and ‘rural’.‘Urban’ includes the two largest metropolitan areas inNorthern Ireland—Belfast (comprising approximately580 000 people) and Derry (approximately 90 000).‘Rural’ comprises villages with populations between1000 and 2250 and smaller villages, hamlets andopen countryside. Finally, ‘intermediate’ areas includetowns and settlements between these two groupings(with populations ranging from 2250 to 75 000). Statistical methods The analysis consisted of three sections. First, the distri-butions of alcohol-related deaths for various individualandareacharacteristicswereassessed.Secondly,Coxpro-portional hazards modelling was used to build a modelthat best described the individual and household factorsassociated with alcohol-related mortality. Finally, toassess the impact of area of residence on alcohol-relatedmortality risk while controlling for the characteristics of individualslivingwithintheseareas,atwo-levelrandom-intercept logistic regression model was used, with indi-vidualsasthelowerlevelunitandSOAasthehigher-levelunit[21].Multi-levelmodelswereimplementedinSTATAversion 10 (using the xtmelogit command). The analysisis presented separately for males and females, as initialanalysis showed significant interaction effects betweengender and socio-economic and area-level factors. RESULTS A total of 720 627 people aged 25–74 years and notliving in communal establishments were enumerated inthe Northern Ireland 2001 census, of whom 578 diedfrom an alcohol-related cause in the 5 year follow upperiod (Table 1); 65% of these deaths were to males and66% were to those aged less than 55 years at the census.Whether assessed by housing tenure, car access, educa-tional attainment, NSSEC or area deprivation, a higherproportion of those living in deprived circumstances diedfrom an alcohol-related illness; this was most evident forcar access, where there was almost a 10-fold differencebetween the proportion of people dying from an alcohol-related cause with no car access, compared to those withaccesstotwoormorecars.Alcohol-relatedmortalitywashigher in those with poor self-reported health (0.17% of  1 In the United Kingdom, after 5 years of secondary education students take examinations in a range of subjects at GCSE level;following2yearsof additionalstudy,studentsmaytakeA-LevelexaminationswhicharerequiredforuniversityentranceintheUnitedKingdom. 2 Since 2001, the NSSEC has been used for all official statistics and surveys. It replaced social class based on occupation andsocio-economic groups. The NSSEC is an occupationally based classification but has rules to provide coverage of the whole adultpopulation. The information required to create an NSSEC is occupation and details of employment status (whether an employer,self-employed or employee; whether a supervisor, manager, etc.). 86  Sheelah Connolly  et al . © 2010 The Authors, Addiction © 2010 Society for the Study of Addiction  Addiction ,  106 , 84–92  Table 1  Demographic and socio-economic characteristics of the study population at the time of the census, and the number andpercentage dying from an alcohol-related cause in the following 5 years. Males Femalesn Populationn (%) of alcohol-related deaths n Populationn (%) of alcohol-related deaths Age (years)25–44 183 943 119 (0.06) 198 616 60 (0.03)45–54 76 468 136 (0.18) 75 793 67 (0.09)55–64 57 078 87 (0.15) 57 280 47 (0.08)65–74 33 893 35 (0.10) 37 556 27 (0.07)Marital statusSingle 77 842 112 (0.14) 65 988 23 (0.03)Married 240 974 113 (0.06) 239 205 99 (0.04)Divorced/widowed/separated 32 566 132 (0.41) 64 052 79 (0.12)Household compositionSingle person household 38 676 136 (0.35) 38 250 49 (0.13)Multi-person household 312 706 241 (0.08) 330 995 152 (0.05)Housing tenureOwner-occupier 291 394 179 (0.06) 295 804 107 (0.04)Private renter 20 242 27 (0.13) 21 973 13 (0.06)Social renter 39 746 171 (0.43) 51 468 81 (0.16)Car accessTwo or more 162 881 61 (0.05) 153 713 29 (0.02)One 151 233 122 (0.08) 161 021 84 (0.05)None 37 268 194 (0.52) 54 511 88 (0.16)NSSEC a Professional 113 643 72 (0.06) 110 961 35 (0.03)Intermediate 23 575 21 (0.09) 63 159 18 (0.03)Own account worker and lower supervisory 106 502 91 (0.09) 36 353 13(0.04)Routine 91 279 131 (0.14) 132 682 103 (0.08)Not working and FT students 16 383 62 (0.38) 26 090 32 (0.12)Education a A-level  +  91 762 56 (0.06) 92 421 16 (0.02)GCSE  +  41 843 30 (0.07) 63 865 27 (0.04)GCSE 57 945 55 (0.09) 67 208 31 (0.05)No qualifications 159 832 236 (0.15) 145 751 127 (0.09)General healthGood 237 955 108 (0.05) 233 817 52 (0.02)Fair 73 545 103 (0.14) 88 366 61 (0.07)Not good 39 882 166 (0.42) 47 062 88 (0.19)LLTINo 277 932 158 (0.06) 289 004 88 (0.03)Yes 73 450 219 (0.30) 80 241 113 (0.14)(Area) Quintile of deprivationLeast deprived 81 313 48 (0.06) 83 911 30 (0.04)2 75 794 47 (0.06) 77 878 37 (0.05)3 70 389 66 (0.09) 72 482 26 (0.04)4 69 906 99 (0.14) 73 159 49 (0.06)Most deprived 53 980 117 (0.22) 61 815 59 (0.10)Settlement bandUrban 135 115 184 (0.14) 150 365 116 (0.08)Intermediate 102 194 122 (0.12) 109 550 59 (0.05)Rural 114 073 71 (0.06) 109 330 26 (0.02) a Duetosmallnumbersof deathsinsomeeducationandNationalStatisticsSocio-economicClassification(NSSEC)categories,itwasnecessarytocombinesomecategoriesforthepurposeof thistabletocomplywiththedisclosurepolicyof theNorthernIrelandLongitudinalStudy.FT:fulltime;GCSE:GeneralCertificate of Secondary Education; LLTI: limiting long-term illness. Area and alcohol-related mortality risk  87 © 2010 The Authors, Addiction © 2010 Society for the Study of Addiction  Addiction ,  106 , 84–92  those reporting not good general health died in thefollow-up period, compared to 0.02% of those reportinggood health) and in those living in urban areas.Table 2 shows the individual and household, demo-graphic and socio-economic characteristics associatedwith alcohol-related mortality. For both males andfemales, the risk of dying from an alcohol-related illnesswas greatest for the 45–54-year age group. However, therelationship between marital status and alcohol-relatedmortalityriskdiffersformalesandfemales:whilemarriedmenhadalowerriskthanothermen,singlewomenwereless likely to die from an alcohol-related illness thanmarried women. Socio-economic gradients in alcohol-related mortality risk were observed, with males andfemales living in social housing 65% and 59%, respec-tively, more likely to die from alcohol-related illness thanthose in owner-occupied housing. Car access was associ-ated significantly with alcohol-related mortality—those Table 2  Individual and household factors associated with alcohol-related mortality in people aged 25–74 years: hazard ratios (HR)(95%confidenceintervals)fromCoxproportionalhazardsanalysis.Eachmodelincludesandadjustsforothervariablesinthecolumn. Males (deaths  =  377) Females (deaths  =  201) Age (years)25–44 1.00 1.0045–54 2.46 (1.89, 3.20) 2.29 (1.57, 3.34)55–64 1.85 (1.36, 2.52) 1.64 (1.06, 2.53)65–74 1.17 (0.78, 1.76) 1.01 (0.60, 1.71)Marital statusMarried 1.00 1.00Single 1.81 (1.34, 2.45) 0.61 (0.36, 1.02)Divorced/widowed/separated 3.00 (2.22, 4.06) 1.10 (0.75, 1.62)Household compositionMulti-person household 1.00 1.00Single-person household 0.97 (0.74, 1.27) 1.32 (0.88, 2.00)Housing tenureOwner-occupier 1.00 1.00Private renter 0.96 (0.63, 1.47) 1.13 (0.62, 2.05)Social renter 1.65 (1.26, 2.16) 1.59 (1.11, 2.28)Car accessTwo or more 1.00 1.00One 1.47 (1.06, 2.03) 2.13 (1.36, 3.32)None 4.48 (3.08, 6.52) 4.05 (2.37, 6.93)NSSECProfessional 1.00 1.00Intermediate 0.98 (0.60, 1.62) 0.63 (0.35, 1.12)Own account worker 1.02 (0.68, 1.53) 1.31 (0.60, 2.86)Lower supervisory 0.99 (0.67, 1.46) 0.35 (0.13, 0.90)Routine 0.95 (0.67, 1.33) 0.95 (0.60, 1.50)Not working 1.49 (1.00, 2.22) 1.21 (0.69, 2.13)Full time student 0.91 (0.13, 6.62 ***EducationDegree  +  1.00 1.00A-level  +  1.09 (0.61, 1.94) 1.76 (0.65, 4.77)GCSE  +  0.92 (0.56, 1.50) 2.43 (1.12, 5.26)GCSE 1.26 (0.81, 1.96) 2.23 (1.02, 4.89)No qualifications 0.80 (0.53, 1.19) 1.71 (0.79, 3.71)General healthGood 1.00 1.00Fair 1.80 (1.33, 2.43) 1.94 (1.29, 2.91)Not good 3.04 (2.12, 4.35) 3.49 (2.13, 5.74)LLTINo 1.00 1.00Yes 1.57 (1.16, 2.12) 1.38 (0.92, 2.08) ***There were no alcohol-related deaths among full-time female students. GCSE: General Certificate of Secondary Education; LLTI: limiting long-termillness; NSSEC: National Statistics Socio-economic Classification. 88  Sheelah Connolly  et al . © 2010 The Authors, Addiction © 2010 Society for the Study of Addiction  Addiction ,  106 , 84–92
Recommended
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks