RESEARCH AND METHODOLOGIES
Indicatori di condizioni materiali e sociali per aggiustare per deprivazione negli studi di piccola area su ambiente e salute: rassegna e prospettive
Roberto PasettoI; Letizia SampaoloI; Roberta PirastuI,II
IDipartimento di Ambiente e Connessa Prevenzione Primaria, Istituto Superiore di Sanità, Rome, Italy
IIDipartimento di Biologia e Biotecnologie, Università di Roma "Sapienza", Rome, Italy
The present review describes and critically analyzes the main characteristics of deprivation indices (DIs), meant as measures of material and social circumstances at a population level, used to adjust for deprivation in small-area studies of environment and health. A systematic search strategy in the period 1990-2009 was run on PubMed/Medline and Embase databases, and 41 articles were selected. In most of the reviewed studies DIs appear to be pragmatically applied and information is not adequate to evaluate whether the use of DIs is efficient. Suggestions for the use of DIs are given foreseeing that more data on exposure, outcomes and other predictive factors will be acquired, and information will be growingly available to disentangle the complex interplay between exposure, health and deprivation.
Key words: environmental exposure, socioeconomic factors, confounding factors, review.
La presente rassegna descrive e analizza criticamente le principali caratteristiche degli indici di deprivazione (ID), intesi come indicatori di circostanze materiali e sociali a livello di popolazione, utilizzati negli studi di piccola area su ambiente e salute per aggiustare per deprivazione. E' stata eseguita una strategia di ricerca sistematica sulle basi di dati PubMed/Medline ed Embase per il periodo 1990-2009. Dalla bibliografia risultante sono stati selezionati 41 articoli. Nella maggior parte degli articoli gli ID risultano applicati in modo pragmatico e le informazioni fornite non sono sufficienti per valutare se l'aggiustamento risulti efficiente o meno. Alla luce di quanto emerso vengono forniti suggerimenti per l'uso degli ID, prevedendo che nel prossimo futuro saranno maggiormente disponibili sia i dati sull'esposizione, gli outcome e altri fattori predittivi, che le informazioni per comprendere la complessa interrelazione tra esposizione, salute e deprivazione.
Parole chiave: esposizione ambientale, fattori socioeconomici, fattori di confondimento, rassegna.
Deprivation can be defined as "a state of observable and demonstrable disadvantage relative to the local community or the wider society or nation to which an individual, family or groups belong" . It is a multidimensional concept in which two main domains can be distinguished: material and social circumstances. Following Townsend, the first involves "the material apparatus, goods, services, resources, amenities and physical environment and local life", the second "the roles, relationships, functions, customs, rights and responsibilities of membership of society and its subgroups" .
Area-based measures of material and social circumstances, defined as deprivation indices (DIs), are indicators of relative deprivation at a population level. They were developed in Great Britain in the '80s to describe and study inequalities in health [2, 3].
In small-area studies of environment and health, potential confounding from deprivation is present in many cases being its role predictive of several diseases [4, 5]. For example the risk is higher in more disadvantaged categories for respiratory cancers (nose, larynx and lung), and cancers of the mouth, pharynx, oesophagus, and stomach [6, 7]. This is a major problem especially since any risk from typical environmental pollution tends to be small, and may be swamped by deprivation because deprived population may concentrate in polluted areas . To appropriately adjust for material and social circumstances, DIs have been extensively used since the mid '90 [4, 8].
In small-area research, territorial unit at study is identified in various ways depending on data availability which is variable in different countries; data for numerator - events at study - and for denominator - population at study - must be available at the same area level. According to Carstairs, "the specific value of small-area analysis is that it permits the examination of data for population which tend to be more homogeneous in character and in their environmental circumstances than are larger and more widely spread populations" .
In the present paper the main characteristics of DIs in small-area studies of environment and health are reviewed; their application is analysed; suggestions for future studies are made.
Coherently with the paper aim, the authors identified keywords, an initial search strategy, and determined the studies inclusion/exclusion criteria. Ecological studies were considered eligible if small-area designed, and evaluating the association between environmental exposure and mortality or morbidity. Since DIs have been extensively used from the mid '90s in ecological small-area studies, it was agreed to settle a 1995-2007 time range for publication date. No publication language limit was set. Then an informal trial search was fixed on PubMed database The following terms were used: small-area, deprivation, socio-economic, epidemiology, morbidity, mortality, cancer, proximity, environment.
By this search relevant studies were retrieved and selected, and further articles were taken into account by examining bibliography included in the selected studies, or because already known to the authors. Aware that some relevant and well-known articles were missing, the authors considered that potential articles citing DI were also to be retrieved, even if not specifically addressing to it as main investigative topic. Therefore, a systematic search strategy simultaneously taking into account the four basic concepts was set up: a) measure of material and social circumstances (DI); b) study design (small-area); c) geographic location of exposed populations (proximity to the source of exposure); d) potential environmental exposure. Points a) and b) were included in the following main search filter, from which occupational studies were excluded:
(("small-area" OR "small area" OR ecologic OR ecological OR environment OR environmental) AND ((socioeconomic OR "socio economic" OR "social and economic" OR deprivation) AND (index OR indices OR indexes)) NOT (occupational OR occupations OR occupation OR job OR work OR worker OR workers) Limits: Humans.
Each concept was translated from natural English language into controlled terms, when available, and into text words and synonyms, to attain also studies not yet or selectively indexed. This explicit search strategy (Table 1) was successively run on PubMed/ Medline and Embase databases to perform a systematic review of the literature published from January 1990 to November 2009. The online search yielded 152 titles and abstracts; 29 of these were original articles dealing with small-area studies and evaluating the association between environmental risks and health effects, therefore were considered pertinent and their full text was requested. Based on experts' suggestions or on studies already known to the authors and on the analysis of the bibliography included in the selected studies, 12 more articles were collected.
The selected studies were examined to describe the use of DIs in small-area studies of environment and health. Territorial unit at study, data source, component variables and reference years were analysed for each DI. For each investigation the main characteristics, i.e. exposure, study and reference population, health outcome, observation period and risk parameter were set forth.
As a result of the search strategy and selection of the papers a total of 41 articles qualified for inclusion in the review; 11 of these were conducted in Italy (4 among them are in Italian), 26 in Great Britain, two in New Zealand, one in Australia and one in Spain. Table 2 describes the main characteristics of DIs applied; index variables are described to the letter as reported in the reviewed articles.
It came into light that Italian studies use ad hoc DIs [10-13] and DIs developed for various epidemiological settings [14, 15]. Several among them [16-21] adopt the same index  but the variables are not homogeneously reported. In Great Britain studies published before 1997 apply ad hoc indices based on the same variables [22-24] while later only Townsend  and Carstairs  indices are used. Non-ad hoc DIs [27-29] are employed in New Zealand [30, 31], Australian  and Spanish  investigations.
It is known that DIs variables pertain to both material and social domains : employment/unemployment is common, education is frequent among DIs, the ownership of material goods is recurrent and expressed in different ways, e.g. living or not in an own house, surface of dwellings or car ownership. Some DIs include also social domains like social class or single parenthood with children.
As far as the population size, in Great Britain in 23 of the collected studies the territorial unit is the Enumeration District [22-24, 34-53] with an average of 400 inhabitants, in two is the Electoral Ward [54, 55], having an average of approximately 5136 inhabitants, and in the remaining one the postcode sector, where residents are on average 6600 people, is seemingly used . In Italy instead, in seven studies [13, 16-21] municipality is the investigated area and the census tract is considered in four [1012, 57]: about 70% of Italian municipalities include less than 5000 residents, while the mean of census tract population is approximately 200 residents . The two New Zealand investigations acquired look at census area unit where residents range between 3000 and 8000 inhabitants [30, 31]. Lastly, in Australia the area analyzed is the postcode sector - number of residents not specified in the reviewed article  - and in Spain is the census tract, where the mean population under study is about 1000 subjects .
As far as DIs construction, it must be noted that in all studies DIs were built using variables from the National Census. Three are the main methods used to treat variables in DIs calculation (for a review of the topic see references 3 and 58). The first is to calculate standardized z-scores of a priori selected variables. The second is to define the major components in large datasets of variables by factor analysis. The third is to assign weights to the selected variables using the evaluation of their relative importance from a representative sample of population (e.g. through questionnaires submitted to health experts). The first and third can be combined.
In the reviewed articles, most DIs are derived from the sum of standardized z-scores of few selected variables. The initial value of each variable is generally defined on the basis of the proportion of population with the characteristic of interest in each territorial unit (e.g. education can be defined as the proportion of population with elementary degree or less). Then indices can be classified in population quantiles - the sum of populations of territorial units belonging to one class is the same for each class - or in territorial units quantiles - each class has the same number of territorial units.
For the reviewed investigations, the main characteristics in terms of exposure, study population, and reference entity are described in Table 3. Environmental exposure/s under investigation can be variously distinguished: multiple sources of pollution from industries [10, 12, 13, 16, 18, 20, 21, 35-39, 46, 51, 56], air pollution from specific pollutants [30, 31, 33, 48-50, 55], or traffic sources [44, 54], landfills [21, 40, 41, 43, 45, 47, 52, 53], incinerators [11, 23, 42, 57], asbestos/asbestiform fibres [17, 19], TV and radio transmitters [24, 32, 34] and nitrates in drinking water . Study population is defined as exposed either because of the presence of environmental exposure/s in the area/s [11-13, 16, 18-21, 30-33, 35, 44, 48-50, 53, 55] or because of residence within a given distance from the putative pollution source [10, 17, 22-24, 34, 36-43, 45-47, 51, 52, 54, 56] e.g. residents within 7.5 km from cokeworks [38, 39], within 2 km from landfills [40, 41] or within 3 km from a combustion plant .
The exposed population is identified at a single point in time - usually census year [10-12, 16, 2224, 30, 34, 35, 39, 44, 55], at two points in time - mainly decennial census year [20, 36, 37, 41, 51, 56], or over several years [21, 32, 33, 38, 40, 42, 43, 45, 47-50, 52, 53, 57].
In the articles included in the review the comparison entity is represented either by the national or local population or by population resident farther than a given distance: e.g. residents more than 2 km from landfills [40, 41] or more than 3 km from a combustion plant .
Table 3 also shows the studies outcomes and risk parameters. The outcomes under investigation are the following: cause-specific mortality [10, 16-20, 30, 33, 37, 44, 46, 50, 51, 55], cancer incidence/hospital admissions for specific diseases [12, 23, 24, 31, 32, 34, 35, 39, 41, 48, 49, 54, 56, 57], cancer mortality and incidence [11, 13, 22, 36], adverse reproductive effects and/or congenital anomalies [21, 38, 40, 42, 43, 45, 47, 52, 53].
The study results are expressed as standardized mortality ratio (SMR), standardized incidence ratio (SIR), relative risk (RR) or odds ratio (OR). In 22 articles the results are presented with and without adjustment for DI [10-13, 17, 20, 30, 31, 33, 37, 4042, 44-46, 48-50, 52, 53].
Lastly, as far as the reference year for DI calculation, in six papers it is not specified [37, 45-47, 53, 54], and in 16 more studies DI variables are from the Census in the same year of identification of the exposed population [10, 12, 13, 16, 22-24, 30, 31, 34-36, 38, 39, 44, 56], in the remaining papers the DI year precedes the identification of population (e.g. in  DI year is 1991 and population is identified in 1997) or is included in the period of population enumeration (e.g. in  DI year is 1986 and population was represented by residents in 1972-1990).
The use of DIs: critical aspects
DIs can be constructed in various ways and calculated at various area levels. Different area-based DIs show unlike associations with health effects from different causes [59, 60].
It is expected that the use of DIs to adjust for deprivation in small-area environment and health studies should result in better risk estimates. This holds true if the following conditions are present: a) exposure is causally associated with outcome; b) DI is a proxy for factors causally associated with outcome; c) exposure and DI are associated -exposed and unexposed areas are different in respect to DI. The risk estimates of most of the 22 reviewed studies reporting adjusted and unadjusted results for deprivation do not significantly differ. An example of an exception is found in two studies assessing the association between air pollution and lung cancer [33, 55]. In these studies conditions a) and b) are defined [61, 62]. In one of them also condition c) is ascertained as it shows correlation between some air pollutants and deprivation  .
The interplay between environmental exposure, deprivation and health is complex. One element of the complexity is the possible reverse causality between deprivation and health. It can occur when studying health effects of long-term environmental exposures, since the worsening in health status caused by environmental exposures can result in downgrading in socioeconomic status. On the other hand, a Great Britain study shows that areas with similar long standing economic disadvantage do not have similar high mortality rates, being some of them resilient to external disadvantage factors. According to Tunstall, "these areas might be doing 'better than expected' or 'overachieving'. This status implies that there may be protective factors or practices in particular areas, which weaken the usually strong relationships between economic adversity and poor health" . Then again, a Dutch investigation documents that some deprived areas are healthier, and wealthy areas are instead unhealthier than expected on the basis of their socio-economic level . The relationship between deprivation and health may also potentially be confounded by phenomenon of selective migration into deprived neighbourhoods of people already in poor health status . Conversely, other studies highlight the apparent clustering of hazardous and polluting sites in areas inhabited by ethnic minorities. The consequence is that socially disadvantaged people become subject to the additional burden of a more polluted or hazardous environment, thus providing evidence of environmental inequities [66-68]. A recent study exemplifies interplay complexity and environmental inequalities. It shows that populations living closer to waste facilities are also the more deprived and both adjusted and unadjusted mortality excesses are higher among them. The authors suggest the possible occurrence of effect modification by deprivation .
In studies of environment and health where deprivation is thought at play another point to be considered is that mortality and morbidity could be influenced by past and by present deprivation as well . This influence is disease-specific, since under or over control for deprivation could occur if DIs time patterns should not fit the relevant time window of the study. An additional issue to be taken into account is that the relationship between deprivation and health could be different in various areas within a country .
In the light of the above issues when adjusting for deprivation in environment and health studies it is advisable to report both adjusted and unadjusted results, to explicit relevant time windows of the investigation conducted, and to document overtime socioeconomic characteristics of the area at study.
Another critical aspect in the use of DI is related to the population size of the territorial unit as reducing it does not automatically lead to a better estimate of deprivation. In fact, smaller units are more homogeneous but DI becomes unstable due to greater sensitivity to local variation . Population size should also be considered in relation to the contextual effect of deprivation. In fact, since this effect is meant as an overall socioeconomic neighbourhood influence , population size should be carefully selected to better understand the contextual effect. Deprivation at area level could be not only a proxy for individual socioeconomic status but also a measurement of the contextual area effect. In fact, sometimes measures of socioeconomic conditions at area level were found to be associated to the health outcomes at study independently from individual socioeconomic status [70, 71].
As far as the characteristics of the studies included in the present review, the great majority have been conducted in Great Britain and Italy. In the former, DIs applied in studies after 1997 have been Townsend's and Carstairs' based on clearly stated variables. In the latter instead, the variables used in the construction of DI are not homogeneously reported. However, it appears that in the great majority of studies performed in a given area the construction of ad hoc indices considering the local socioeconomic context is not taken into account.
It should also be pointed out that the choice of a Census year seems to derive from a pragmatic evaluation of data availability, without considering the relevant time window related to the study hypothesis; it is clearly shown, for example, in a study indicating that the more appropriate socioeconomic indicator refers to ten years before breast cancer diagnosis .
Another observation is that in the majority of the reviewed studies DIs are classified in quintiles but rarely information is given about the basis on which quintiles are made and the reason why they are chosen. Generally, quantiles appear to be adopted to avoid the potential residual confounding, and to have stable reference rates for each deprivation category; the latter is an issue that should be weighted in case of rare diseases or small reference populations.
In the majority of the studies adjustment for deprivation is carried out by indirect standardization. The quantiles of DIs and the reference rates for the analyzed outcomes should pertain to the same territorial unit, but the studies in the review do not grant appropriate information about this issue.
Systematic bibliographic research: critical aspects
By matching the results of the two bibliographic searches carried out, it was ascertained that the systematically conducted bibliographic search was lacking of some relevant articles that were instead included in the initial search. By examining how both the obtained articles and the missing ones were indexed some consideration came into light. The main difficulty in information retrieval originated from the fact that there was no applicable controlled term to express the concept of small-area study, unless it was referred to analyzing the variation in utilization of health care in small geographic or demographic areas; for the concept of DI the problem was similar; so, the articles dealing with these two concepts could only be achieved if such terms, or their synonyms, were included in the title or abstract. It must be considered that most of the articles identified through PubMed/Medline and Embase come from countries where DIs have been in use for a long time, therefore they are applied in the studies although the term itself is not found in the title and/or abstract; this implies that such articles are not in the hit list with this search approach. It also came out that different researchers assign different meaning, and associate similar but not equal variables to the concept of DI. In addition, in the MeSH Thesaurus a term adhering to the meaning of deprivation as a multidimensional index is not provided. In the systematic search strategy DI was therefore translated with "socioeconomic factors", and the terms included under the related tree structure, as well as a combining of synonyms and related terms for this other concept, were also added.
It is an opinion shared by the majority that for systematic reviews the combining of online searches, bibliographic examination of the articles already owned, and experts opinion is the best approach for a comprehensive information retrieval . This method proves to be even more appropriate when, as it is in this case, the topic of the study foretells a challenging bibliographic research.
The present review describes DIs used to adjust for deprivation in small-area studies of environment and health in the last fifteen years. To the authors' knowledge this is the first attempt to exhaustively examine such a relevant topic in environmental epidemiology. A main remark is that this review makes clear the difficulty in understanding whether adjusting from deprivation using DIs was efficient or not to assess confounding from socioeconomic factors. The various DIs were adopted from studies looking at health resource allocation or health inequalities, where the role of the environment on health was not the main investigative topic. Therefore the application of DIs appears to have been pragmatic without considering the specific aim of the studies. Time, a key feature in epidemiology, does not appear to have been taken into account either in the choice of DIs or in the interpretation of the results.
In adjusting for deprivation the potential for under or over control is possible and it is specific for a given environmental exposure-disease association. Over/under control is less of a problem in discussing results, if the more specific and the stronger is the exposure disease association, and the rarer and higher is the exposure .
The use of DIs is increasing and expanding to many countries. In the years to come, at different area level, improved exposure information and time-series data on exposure, outcomes and other predictive factors will be growingly available. This enhanced data availability, together with better knowledge on the relationship between deprivation and health, will make possible studies aimed at disentangling the complex interplay between environmental exposure, health and deprivation. In this context one of the main areas to be developed will be the application of mixed design studies combining ecological and individual data [5, 73]. To deal with this challenge, future studies should foresee an a priori assessment of every element at play. Even though the a priori knowledge is often incomplete and analytical solutions to tackle with complex interplay are not always available, this approach has the strength to point out the information we ideally have to collect, giving a framework to discuss the final results. A novel tool proposed to perform a formal a priori evaluation is the direct acyclic graphs (DAG) analysis [74-76]. In the BOX suggestions are given for the use of DIs in small-area environment and health studies at the light of their use in the past, the above discussion and the possible scenarios in the next future.
The authors acknowledge the helpful and constructive comments to the article made by Francesco Mitis and Dolores Catelan.
Conflict of interest statement
There are no potential conflict of interest of any financial or personal relationship with other people or organizations that could inappropriately bias conduct and findings of this study.
1. Townsend P. Deprivation. J Soc Policy 1987;16:125-46.
2. Whitehead M. The health divide. Pelican Books: London; 1988.
3. Carstairs V. Socio-economic factors at area level and their relationship with health. In: Elliott P, Wakefield JC, Best NG, Briggs DJ (Ed.). Spatial epidemiology. Methods and applications. New York: Oxford University Press; 2000. p. 51-67.
4. Dolk H, Mertens B, Kleinschmidt I, Walls P, Shaddick G, Elliott P. A standardisation approach to the control of socioeconomic confounding in small area studies of environment and health. J Epidemiol Community Health 1995;49(Suppl 2):S9-S14.
5. Elliott P, Savitz DA. Design issues in small-area studies of environment and health. Environ Health Perpect 2008;116: 1098-104.
6. Faggiano F, Partanen T, Kogevinas M, Boffetta P. Socioeconomic differences in cancer incidence and mortality in social inequalities and cancer. In: Kogevinas M, Pearce N, Susser M, Boffetta P. (Ed.). IARC Scientific Publications 1997;138:65-176.
7. Blakely T, Hales S, Woodward A. Poverty: assessing the distribution of health risks by socioeconomic position at national and local levels. Geneva: World Health Organization; 2004. (WHO Environmental Burden of Disease Series, No. 10).
8. Elliott P, Dolk H. Foreword. J Epidemiol Commun Health 1995;49(Suppl 2):S2.
9. Carstairs V. Small area analysis and health service research. Community Med 1981;3:131-9.
10. Michelozzi P, Fusco D, Forastiere F, Ancona C, Dell'Orco V, Perucci CA. Small area study of mortality among people living near multiple sources of air pollution. Occup Environ Med 1998;55:611-5.
11. Chellini E, Cherubini M, Chetoni L, Costantini AS, Biggeri A, Vannucchi G. Risk of respiratory cancer around a sewage plant in Prato, Italy. Arch Environ Health 2002;57:548-53.
12. Parodi S, Stagnaro E, Casella C, Puppo A, Daminelli E, Fontana V, Valerio F, Vercelli M. Lung cancer in an urban area in Northern Italy near a coke oven plant. Lung cancer 2005;47:155-64.
13. Biggeri A, Lagazio C, Catelan D, Pirastu R, Casson F, Terracini B. Environment and health in Sardinia, Italy. Epidemiol Prev 2006;30(Suppl 1):1-96.
14. Cadum E, Costa G, Biggeri A, Martuzzi M. Deprivation and mortality: a deprivation index suitable for geographical analysis of inequalities. Epidemiol Prev 1999;23:175-87.
15. Caranci N, Biggeri A, Grisotto L, Pacelli B, Spadea T, Costa G. The Italian deprivation index at census tract level: definition, description and association with general mortality. Epidemiol Prev (in press).
16. Martuzzi M, Mitis F, Biggeri A, Terracini B, Bertollini R. Environment and health status of the population in areas with high risk of environmental crisis in Italy. Epidemiol Prev 2002;26(Suppl 6):1-53.
17. Biggeri A, Pasetto R, Belli S, Bruno C, Di Maria G, Mastrantonio M, Trinca S, Uccelli R, Comba P. Mortality from chronic obstructive pulmonary disease and pleural mesothelioma in an area contaminated by natural fiber (fluoroedenite). Scand J Work Environ Health 2004;30:249-52.
18. Minichilli F, Bartolacci S, Buiatti E, Pierini A, Rossi G, Bianchi F. Mortality in the area around Massa-Carrara 10 years after ANIC-Agricoltura and Farmoplant chemical plants were shut down. Epidemiol Prev 2006;30(2):120-8.
19. Marinaccio A, Scarselli A, Binazzi A, Mastrantonio M, Ferrante P, Iavicoli S. Magnitude of asbestos-related lung cancer mortality in Italy. Br J Cancer 2008;99:173-5.
20. Ianni E, Mignozzi K, Mitis F. Geographic epidemiologic descriptive study on the national priority site for remediation "Laguna di Grado e Marano". Epidemiol Prev 2009;33:27-36.
21. Martuzzi M, Mitis F, Bianchi F, Minichilli F, Comba P, Fazzo L. Cancer mortality and congenital anomalies in a region of Italy with intense environmental pressure due to waste. Occup Environ Med 2009;66:725-32.
22. Sans S, Elliott P, Kleinschmidt I, Shaddick G, Pattenden S, Walls P, Grundy C, Dolk H. Cancer incidence and mortality near the Baglan Bay petrochemical works, South Wales. Occup Environ Med 1995;52:217-24.
23. Elliott P, Shaddick G, Kleinschmidt I, Jolley D, Walls P, Beresford J, Grundy C. Cancer incidence near municipal solid waste incinerators in Great Britain. Br J Cancer 1996; 73:702-10.
24. Dolk H, Shaddick G, Walls P, Grundy C, Thakrar B, Kleinschmidt I, Elliott P. Cancer incidence near radio and television transmitters in Great Britain. I. Sutton Coldfield transmitter. Am J Epidemiol 1997;145(1):1-9.
25. Townsend P, Phillimore P, Beattie A. Health and deprivation: inequalities and the north. London: Croom Helm; 1988.
26. Carstairs V, Morris R. Deprivation and health in Scotland. Aberdeen: Aberdeen University Press; 1991.
27. Crampton P, Salmond C, Kirkpatrik R, Scarborough R, Skelly C. Degrees of deprivation: an atlas of socioeconomic difference. Auckland, New Zealand: David Bateman Ltd; 2000.
28. Australian Bureau of Statistics. Construction of an indicator of socio-economic status. Information paper No. 15. ABS; 1986.
29. Domínguez-Berjón MF, Borrell C, Cano-Serral G, Esnaola S, Nolasco A, Pasarín MI, Ramis R, Saurina C, Escolar-Pujolar A. Constructing a deprivation index based on census data in large Spanish cities (the MEDEA project). Gac Sanit 2008;22:179-87.
30. Scoggins A, Kjellstrom T, Fisher G, Connor J, Gimson N. Spatial analysis of annual air pollution exposure and mortality. Sci Total Environ 2004;321:71-85.
31. Sabel CE, Wilson JG, Kingham S, Tisch C, Epton M. Spatial implications of covariate adjustment on patterns of risk: respiratory hospital admissions in Christchurch, New Zealand. Soc Sci Med 2007;65:43-59.
32. McKenzie DR, Yin Y, Morrell S. Childhood incidence of acute lymphoblastic leukaemia and exposure to broadcast radiation in Sydney--a second look. Aust NZJ Public Health 1998;22(3 Suppl):360-7.
33. Barceló MA, Saez M, Saurina C. Spatial variability in mortality inequalities, socioeconomic deprivation, and air pollution in small areas of the Barcelona Metropolitan Region, Spain. Sci Total Environ 2009;407:5501-23.
34. Dolk H, Elliott P, Shaddick G, Walls P, Thakrar B. Cancer incidence near radio and television transmitters in Great Britain. II. All high power transmitters. Am J Epidemiol 1997;145(1):10-7.
35. Parslow RC, McKinney PA, Law GR, Staines A, Williams R, Bodansky HJ. Incidence of childhood diabetes mellitus in Yorkshire, Northern England, is associated with nitrate in drinking water: an ecological analysis. Diabetologia 1997; 40:550-6.
36. Wilkinson P, Thakrar B, Shaddick G, Stevenson S, Pattenden S, Landon M, Grundy C, Elliott P. Cancer incidence and mortality around the Pan Britannica Industries pesticide factory, Waltham Abbey. Occup Environ Med 1997;54:101-7. Erratum in: Occup Environ Med 1997;54:216.
37. Dolk H, Thakrar B, Walls P, Landon M, Grundy C, Sáez Lloret I, Wilkinson P, Elliott P. Mortality among residents near cokeworks in Great Britain. Environ Med 1999;56:34-40.
38. Dolk H, Pattenden S, Vrijheid M, Thakrar B, Armstrong B. Perinatal and infant mortality and low birth weight among residents near cokeworks in Great Britain. Arch Environ Health 2000;55:26-30.
39. Aylin P, Bottle A, Wakefield J, Jarup L, Elliott P. Proximity to coke works and hospital admissions for respiratory and cardiovascular disease in England and Wales. Thorax 2001; 56:228-33.
40. Elliott P, Briggs D, Morris S, de Hoogh C, Hurt C, Jensen TK, Maitland I, Richardson S, Wakefield J, Jarup L. Risk of adverse birth outcomes in populations living near landfill sites. BMJ 2001;323:363-8. Erratum in: BMJ 2001;323:1182.
41. Jarup L, Briggs D, de Hoogh C, Morris S, Hurt C, Lewin A, Maitland I, Richardson S, Wakefield J, Elliott P. Cancer risks in populations living near landfill sites in Great Britain. Br J Cancer 2002;86:1732-6.
42. Cresswell PA, Scott JE, Pattenden S, Vrijheid M. Risk of congenital anomalies near the Byker waste combustion plant. J Public Health Med 2003;25:237-42.
43. Morris SE, Thomson AO, Jarup L, de Hoogh C, Briggs DJ, Elliott P. No excess risk of adverse birth outcomes in populations living near special waste landfill sites in Scotland. Scott Med J 2003;48:105-7.
44. Maheswaran R, Elliott P. Stroke mortality associated with living near main roads in England and Wales: a geographical study. Stroke 2003;34:2776-80.
45. Morgan OW, Vrijheid M, Dolk H. Risk of low birth weight near EUROHAZCON hazardous waste landfill sites in England. Arch Environ Health 2004;59:149-51.
46. Hodgson S, Nieuwenhuijsen MJ, Hansell A, Shepperd S, Flute T, Staples B, Elliott P, Jarup L. Excess risk of kidney disease in a population living near industrial plants. Occup Environ Med 2004;61:717-9.
47. Palmer SR, Dunstan FD, Fielder H, Fone DL, Higgs G, Senior ML. Risk of congenital anomalies after the opening of landfill sites. Environ Health Perspect 2005;113:1362-5.
48. Maheswaran R, Haining RP, Brindley P, Law J, Pearson T, Fryers PR, Wise S, Campbell MJ. Outdoor air pollution and stroke in Sheffield, United Kingdom: a small-area level geographical study. Stroke 2005a;36:239-43.
49. Maheswaran R, Haining RP, Brindley P, Law J, Pearson T, Fryers PR, Wise S, Campbell MJ. Small-area level ecological study. Outdoor air pollution, mortality, and hospital admissions from coronary heart disease in Sheffield, UK: a small-area level ecological study. Eur Heart J 2005b;26(23):2543-9.
50. Maheswaran R, Haining RP, Pearson T, Law J, Brindley P, Best NG. Outdoor NOx and stroke mortality: adjusting for small area level smoking prevalence using a Bayesian approach. Stat Methods Med Res 2006;15:499-516.
51. Hodgson S, Nieuwenhuijsen MJ, Elliott P, Jarup L. Kidney disease mortality and environmental exposure to mercury. Am J Epidemiol 2007;165:72-7.
52. Jarup L, Morris S, Richardson S, Briggs D, Cobley N, de Hoogh C, Gorog K, Elliott P. Down syndrome in births near landfill sites. Prenat Diagn 2007;27:1191-6.
53. Elliott P, Richardson S, Abellan JJ, Thomson A, de Hoog C, Jarup L, Briggs DJ. Geographic density of landfill sites and risk of congenital anomalies in England. Occup Environ Med 2009;66:81-9.
54. Harrison RM, Leung P, Somervaille L, Smith R, Gilman E. Analysis of incidence of childhood cancer in the West Midlands of the United Kingdom in relation to proximity to main roads and petrol stations. Occup Environ Med 1999; 56:774-80.
55. Elliott P, Shaddick G, Wakefield JC, de Hoogh C, Briggs DJ. Long-term associations of outdoor air pollution with mortality in Great Britain. Thorax 2007;62:1088-94.
56. Wilkinson P, Thakrar B, Walls P, Landon M, Falconer S, Grundy C, Elliott P. Lymphohaematopoietic malignancy around all industrial complexes that include major oil refineries in Great Britain. Occup Environ Med 1999;56:577-80.
57. Federico M, Pirani M, Rashid I, Caranci N, Cirilli C. Cancer incidence in people with residential exposure to a municipal waste incinerator: An ecological study in Modena (Italy), 1991-2005. Waste Management 2010;30:1362-70.
58. Bell N, Schuurman N, Hayes MV. Using GIS-based methods of multicriteria analysis to construct socio-economic deprivation indices. Int J Health Geogr 2007;6:17.
59. Davey Smith G, Whitley E, Dorling D, Gunnell D. Area based measures of social and economic circumstances: cause specific mortality patterns depend on the choice of index. J Epidemiol Community Health 2001;55:149-50.
60. Testi A, Ivaldi E. Material versus social deprivation and health: a case study of an urban area. Eur J Health Econ 2009; 10:323-8.
61. World Health Organization. Air Quality Guidelines Global Update 2005. Report on a working group meeting, Bonn, Germany, 18-20 October 2005. Copenhagen: World Health Organization; 2005.
62. Spitz MR, Wu X, Wilkinson A, Wei Q. Cancer of the lung. In: Schottenfeld D and Fraumeni JF Jr (Ed.). Cancer Epidemiology and Prevention, Third Edition. New York: Oxford University Press; 2006. p. 638-58.
63. Tunstall H, Mitchell R, Gibbs J, Platt S, Dorling D. Is economic adversity always a killer? Disadvantaged areas with relatively low mortality rates. J Epidemiol Community Health 2007;61:337-43.
64. van Hooijdonk C, Droomers MI, van Loon JAM, van der Lucht F, Kunst AE. Soc Sci Med 2007;64:1326-42.
65. Norman P, Boyle P, Rees P. Selective migration, health and deprivation: a longitudinal analysis. Soc Sci Med 2005;60: 2755-71.
66. Brown P. Race, class and environmental health: a review and systematization of the literature. Environ Res 1995; 69:5-30.
67. Oakes JM, Anderton DL, Anderson AB. A longitudinal analysis of environmental equity in communities with hazardous waste facilities. Social Sci Res 1996;25:125-48.
68. Morello-Frosch R, Pastor M Jr, Porras C, Sadd J. Environmental justice and regional inequality in southern California: implications for future research. Environ Health Perspect 2002;110(Suppl 2):149-54.
69. Martuzzi M, Mitis F, Forastiere F. Inequalities, inequities, environmental justice and health. Eur J Public Health 2010; 20:21-6.
70. Stafford M, Gimeno D, Marmot MG. Neighbourhood characteristics and trajectories of health functioning: a multilevel prospective analysis. Eur J Public Health 2008;6:604-10.
71. Webster TF, Hoffman K, Weinberg J, Vieira V, Aschengrau A. Community- and individual-level socioeconomic status and breast cancer risk: multilevel modeling on Cape Cod, Massachussetts. Environ Health Perpect 2008;116:1125-9.
72. Higgins JPT, Green S (Ed.). Cochrane Handbook for Systematic Reviews of Interventions Version 5.0.2 (updated September 2009). The Cochrane Collaboration, 2009. Available from www.cochrane-handbook.org.
73. Wakefield J. Ecologic studies revisited. Ann Rev Public Health 2008;29:75-90.
74. Chaix B, Leal C, Evans D. Neighborhood-level confounding in epidemiologic studies. Unavoidable challenges, uncertain solutions. Epidemiology 2010;21:124-7.
75. Pearl J. Causality. New York: Oxford University Press; 2000.
76. Greenland S. Quantifying biases in causal models: classical confounding vs. collider-stratification bias. Epidemiology 2003; 14:300-6.
Address for correspondence:
Dipartimento di Ambiente e Connessa Prevenzione Primaria, Istituto Superiore di Sanità
Viale Regina Elena 299
00161 Rome, Italy
Received on 31 March 2010.
Accepted on 10 May 2010.