You are here

  • Home
  • Archive
  • Volume 68, Issue 5
  • Does a geographical context of deprivation affect differences in injury mortality? A multilevel analysis in South Korean adults residing in metropolitan cities

Article Text

Article menu

Download PDFPDF

Injury
Does a geographical context of deprivation affect differences in injury mortality? A multilevel analysis in South Korean adults residing in metropolitan cities
  1. JeSuk Lee1,
  2. Weon-Young Lee1,
  3. MaengSeok Noh2,
  4. Young-Ho Khang3,4
  1. 1Department of Preventive Medicine, College of Medicine, Chung-Ang University, Seoul, Korea
  2. 2Department of Statistics, Pukyong National University, Busan, Korea
  3. 3Department of Preventive Medicine, College of Medicine, University of Ulsan, Seoul, Korea
  4. 4Institute of Health Policy and Management, College of Medicine, Seoul National University, Seoul, Korea
  1. Correspondence to Dr W-Y Lee, Department of Preventive Medicine, College of Medicine, Chung-Ang University, 84 Heukseok-Ro, Dongjak-Gu, Seoul 156-756, Korea; wylee{at}cau.ac.kr

Abstract

Background This study aimed to examine whether the socioeconomic context of urban areas affects differences in adult mortality from injuries in the districts of all seven South Korean metropolitan cities, after adjusting for individual demographic and socioeconomic indicators.

Methods Two different sets of data were used in this study: (1) the National Death Registration data from 2003 to 2008; and (2) the National Census in 2005. Variables for individual characteristics were gender, age, residential area and educational level. A geographic deprivation index was calculated based on the Carstairs Index. Multilevel Poisson regression models were used to analyse the relationship between area deprivation levels and injury mortality.

Results Greater mortality risks of traffic accidents, falls, suicide and all injuries were found in the elderly, the less educated and men, compared with their counterparts. The most deprived districts were at greater risks of death due to traffic accidents (risk ratio (RR)=1.34; 95% CI 1.05 to 1.73), falls (RR=1.63; 95% CI 1.20 to 2.20), suicide (RR=1.09; 95% CI 1.01 to 1.17) and all injuries (RR=1.14; 95% CI 1.07 to 1.22) compared with the least deprived districts, even after individual level socioeconomic variables were controlled for. However, area level deprivation did not show cross level interactions with the individual level education in estimating fatal injury risks.

Conclusions Both contextual and compositional effects of socioeconomic status on injury mortality among urban areas in South Korea should be considered in allocating resources for injury prevention.

https://doi.org/10.1136/jech-2013-203082

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    An unequal distribution of injury according to socioeconomic indicators is ethically unacceptable and an important finding with regard to improving population health. As injury mortalities are neither unavoidable nor irreversible, they should not reflect differences in wealth.1 In addition, the size of the gap between mortality and morbidity rates of the most and least disadvantaged groups could be regarded as an indication of the potential for improvement in a nation's (or a living area's) health and safety.2 Identification of groups who are at risk of poor health can inform sound governance of public health services. Accumulating evidence shows that there is an association between area level or individual level socioeconomic position (SEP) and injury mortality.3 Studies on these associations would be particularly valuable for injury prevention in countries such as South Korea where mortality rates from both suicide and traffic accidents have been recorded as the highest among Organisation for Economic Cooperation and Development (OECD) member countries.4 ,5

    Most large cities in many countries, along with their urbanisation, have poverty stricken neighbourhoods.6 The concentration of poverty in cities has resulted in geographic inequalities in health, of which inequalities in unintentional injury and in violence are major public health problems.7 ,8 Moreover, the majority of injury deaths in industrialised countries occur in cities. For example, approximately 80% of the total deaths due to major external causes (traffic injury, falls and suicide) in South Korea occurred in urban areas.9 In this regard, inequalities in injury mortality between different areas in cities need to be investigated to prioritise injury prevention policy.

    However, there are a few studies which have investigated intra-urban inequalities in mortality rates from injuries in the adult population in cities in the USA,10 Canada11 and western European Union member states.12–14 Studies10 ,11 ,13 on differences in adult fatal injuries in urban areas with adjustment for individual SEP are much rarer. According to these studues,10 ,11 ,13 differences in mortality rates from injuries in urban areas vary by gender, injury type or indicators representing area level SEP. For example, some studies10 ,11 reported that intra-urban inequalities in some types of fatal injury existed between the most and least deprived areas.

    Recently, Kim et al15 reported that both individual level and area level SEP influenced childhood fatal injuries in South Korea. Moreover, intentional and unintentional injuries might be simultaneously examined in evaluating the impact of individual and geographical SEP. Cohen et al16 argued that an integrated injury prevention approach could significantly impact on the underlying factors of intentional and unintentional injuries, allowing practitioners to achieve greater outcome and more efficient use of resources. For example, a local government may develop a more comprehensive urban planning policy for safer neighbourhoods by addressing road design (for traffic accident) and safety measures in tall buildings (for suicide, by jumping and fall injuries) at the same time. In light of this, the aim of our study was to explore whether the socioeconomic context (urban areas) affects the major types (traffic accident, falls and suicide) and all types of adult fatal injuries, independent of individual level SEP, using data from all seven major metropolitan areas in South Korea.

    Methods

    Data sources and study population

    The study used two sets of data from Statistics Korea: (1) the National Death Registration data (NDR) between 2003 and 2008, with the 10th revision of the International Classification of Disease (ICD-10) for information on causes of death; and (2) the National Census in 2005. Both NDR and the 2005 census data are publicly accessible through microdata service systems provided by Statistics Korea.9 Subjects included in the study were aged 35 years and over. Adults aged ≤34 years were excluded from the study because some may not have completed their education, and the numbers of primary school graduates in these younger ages were very small, resulting in unstable numbers for injury deaths. As both NDR and the National Census had information about individual educational level, this was used as the only indicator of individual level SEP.

    We extracted demographic records (gender, age, residential area, educational level) of the deceased aged 35 years and over by injury, which was coded as ‘V’, ‘W’, 'X’ or ‘Y’ (ICD-10) in the NDR between 2003 and 2008. After further elimination of a few individuals who had missing educational data (0.067% of total deaths), the total number of deaths in the final analysis was 36 242, within which three main causes of death by injury or suicide were analysed separately: 20 961 (57.9%) suicides; 2354 (6.5%) traffic accidents; and 1507 (4.2%) falls. In addition, there were other categories: exposure (5.8%), poisoning (4.8%), assault (4.4%) and suffocation (3.6%). In data extracted from the NDR, outcome measures were indexed as traffic accidents (V01-V99), falls (W00-W19), suicide (X60-X84) and all injuries (V01-Y34).

    From the 2005 National Census data,9 we extracted demographic information (gender, age, residential area, education level) of adults aged 35 years and over that had lived in seven South Korean metropolitan cities with 69 urban districts. To measure the deprivation level of the urban districts, the relevant information was collected from the same database. After further elimination of a few individuals who had missing educational data (0.03% of total population), the total population in these cities for the study analysis was 10 583 859.

    Individual level and area level variables

    For individual level SEP, we used educational level of the study subjects (adults aged 35 years and over) categorised into one of three groups: (1) no education or only primary school education; (2) middle and high school education; and (3) university or higher level education. Educational level as an indicator of individual level SEP is stronger than other indicators such as occupation or income: education can be applied to both adult men and women; education changes little after a certain age; and education is largely free from reverse causation in examining the relationship between SEP and health.17 According to Kim and Khang,18 the reliability of self-reported educational level between survey data and mortality data was higher than that of self-reported occupation in the NDR.

    We used metropolitan district (called ‘Gu’ in South Korea) as the spatial unit of analysis, which is the smallest municipal unit responsible for local government and the lowest administrative unit in which official data such as the NDR and National Census are now available. ‘Gu’ has relatively large and heterogeneous populations, ranging from 26 371 to 282 568. For area level SEP measurement, deprivation indexes were assigned to 69 Gus across seven metropolitan cities. In this study, the Carstairs Index19 was used as an index of area level deprivation, which is based on four census indicators: low social class, lack of car ownership, overcrowding and male unemployment. Based on the National Census data in 2005, the following district specific proportions of households were calculated: living without a car, overcrowded living conditions (>1.5 persons/room), an economically active head of household in manual occupation and male (aged 20–64 years) unemployment. We did not add education to the deprivation index as education was used as the individual level SEP in this study and was strongly related to the variable for an economically active head of household in manual occupation. District specific deprivation scores were calculated by averaging the z standardised scores of four proportions. The 69 districts were separated into four quartile groups according to the district specific deprivation scores.

    Statistical analysis

    The data had a hierarchical multilevel structure of 1656 cells at level 1, consisting of individuals in numerators and denominators cross tabulated by age, gender and educational level. These level 1 data were nested within 69 Gus at level 2 and nested within Seoul and non-Seoul metropolitan areas (six metropolitan cities) at level 3. The in-between individual effect at level 1, as a fixed effect, is explained by the Poisson process, and the in-between district effect at level 2 is considered a random effect, assuming normal distribution. As level 3 has seven metropolitan areas and there were slight differences in crude mortality rates during the study period of 2003–2008 between non-Seoul (369.9 per 100 000) and Seoul (308.6 per 100 000) metropolitan areas (table 2), level 3 was taken as a fixed effect for Seoul or non-Seoul area. Therefore, we used a 2 level Poisson mixed effect model, taking 69 districts as random effect. The results of fixed effects are shown as risk ratio (RR) with 95% CIs20 and those of random effects are presented as variance partition coefficients. The models were fitted using proc nlmixed mode in the statistical package SAS (V.9.2).

    Results

    Study population and mortality rates by individual and areal characteristics

    Table 1 presents the number of Gus, the total study population and the distribution of the study population by district in each metropolitan city. Seoul, the capital city of South Korea, had the largest number (n=25) of Gus and Ulsan has the smallest (n=4) among the metropolitan cities. Based on the median population per Gu, Seoul has the highest study population per district (189 166) and Ulsan has the lowest (97 476). There was wide variation in district population within a metropolitan area. The median deprivation index scores of districts ranged from −0.77 (Ulsan) to 1.25 (Busan). The range of deprivation index scores was relatively narrow in Ulsan and wide in Seoul and Daejeon.

    View this table:
    Table 1

    Number of areas (Gu), total study population and summary statistics of the area study population and deprivation index by seven metropolitan cities in South Korea

    Table 2 presents the number of deaths and crude mortality rates by causes of death, individual level characteristics (gender, age and educational level) and macro level variables (area level deprivation and Seoul vs non-Seoul metropolitan area). The ratios of crude mortality rates in men compared with women for traffic accidents, falls, suicide and all injuries were 2.7, 2.1, 2.6 and 2.4, respectively. Regarding age groups, greater mortality rates from all types of injury were observed among those aged 65 years and older compared with younger age groups. Mortality rates from all types of injury increased with the level of individual level education. Mortality rates from all types of injuries were much higher in the most deprived quartile group than in any of the other groups. For metropolitan city, non-Seoul cities showed higher mortality rates than Seoul for all type of injuries except fall related fatal injury.

    View this table:
    Table 2

    Study population, by individual and geographic characteristics, and number of deaths (n) and crude mortality rates (per 100 000) for traffic accidents, falls, suicide and all injuries, 2003–2008

    Multilevel models

    Tables 3 and 4 show fixed and random parameters estimated from multilevel Poisson regression models. For traffic accidents, significant variation was observed in mortality rates across districts (model 1). When individual level characteristics were added (model 2), these individual characteristics explained 51.2% (=(0.222–0.108)/0.22×100) of the area level variation in traffic accident mortality rates. Traffic accident mortality rates in men were greater (RR=3.62, 95% CI 3.30 to 3.98) than those in women, and those aged 45–54 years had a greater traffic accident mortality rate (RR=1.22, 95% CI 1.09 to 1.37) compared with those aged 35–44 years. Those who had lower educational levels showed greater traffic accident mortality rates: those who had less than primary school education had a 5.6 times (95% CI 4.82 to 6.51) greater risk than those with university or higher education. When area level deprivation was added (model 3), the inter-district variation in traffic accident fatal injuries decreased by 11.1% (=(0.108–0.024)/0.108×100), and the most deprived areas had a fatal injury risk 1.34 (95% CI 1.05 to 1.73) times greater than the least deprived areas (model 3). A significant cross level interaction was not found: trends in RRs for traffic accident mortality by individual education level were not different between the richest 75% of districts and the other poorer quartile regions (model 5).

    View this table:
    Table 3

    Parameter estimates from multilevel Poisson regression models for traffic accidents and falls

    View this table:
    Table 4

    Parameter estimates from multilevel Poisson regression models for suicide and all other injuries

    For mortality due to falls, significant variation in mortality across districts was found (model 1). As shown in model 2, individual level variables explained 49.2% (=(0.327–0.166)/0.327×100) of the district level variation in fatal fall injury rates. Mortality from falls was greater in men (RR=3.26, 95% CI 2.91 to 3.65) than in women. Those aged 65 years and older were at 4.1 times (RR=4.10, 95% CI 3.42 to 4.93) greater risk of fall related deaths compared with the 35–44 year age group. Higher mortality rates due to falls were seen with lower educational level. People with primary or no school education had a mortality risk from falls 6.78 times (95% CI 5.47 to 8.41) greater than those with university or higher education. Based on model 3, the inter-district variation decreased by 20.5% (=(0.166–0.132)/0.166×100), and the most deprived areas had a greater mortality risk related to falls of 1.63 (95% CI 1.20 to 2.20) times that of the least deprived area. We did not find a significant cross level interaction in fatal injury by fall, as shown in model 5 in table 3.

    Mortality from suicide showed a significant variation across districts (model 1). The results of model 2 showed that individual level variables explained 95.5% (=(0.199–0.009)/0.199×100) of the area level variation in suicide rates. Mortality rates from suicide were higher in men (RR=3.38, 95% CI 3.28 to 3.49) than in women, and higher among those aged 65 years and older (RR=1.79, 95% CI 1.71 to 1.87) than those aged 35–44 years. Mortality rates from suicide had an inverse association with individual level education. The RR for suicide in persons with no school or primary school education was 4.09 (95% CI 3.89 to 4.30) compared with those with university or higher education. Based on model 3, the most deprived area had a significant association with a fatal injury by suicide (RR=1.09, 95% CI 1.01 to 1.17). A significant cross level interaction in fatal injury by suicide was not observed, as shown in model 5 in table 4. For all injuries, estimated fixed and random effects were similar to the results of the three main causes of injury.

    Discussion

    This study has demonstrated that men and those in the lower educational group were more vulnerable to increased mortality risks from traffic accidents, falls, suicide and all injuries than their counterparts in South Korean metropolitan cities. At the area level, the most deprived urban districts were associated with greater risks of injury mortality than the least deprived districts, even after individual level variables were controlled for.

    For traffic accident mortality, our results are in line with other studies10 ,13 ,21–23 showing a greater risk in men than women. The inverse association of individual level SEP with mortality rates due to traffic accidents has been reported in the USA,10 Canada11 and Spain.13 These inequalities, according to individual level SEP, may be attributable to the relevant characteristics associated with low SEP: lower seat belt use, higher ownership of old cars and driving at higher speed.13 ,24

    Even when individual level SEP was adjusted for, the results of the study showed that traffic accident mortality remained greater in the poorest quartile districts compared with the richest quartiles. Previous studies examining this association showed mixed results. A study of European cities with a cross sectional ecological design12 reported that deprived small areas had greater mortality risks of transport injuries in men but not in women. Meanwhile, the most deprived areas, independent of individual level SEP, had a significantly greater mortality from traffic accidents compared with the least deprived areas in the USA10 and Canada11 but not in Barcelona.13 Several factors may explain the greater fatal injury risks among the deprived areas: poor road design10 ,11 ,25 and less access to emergency services.10

    South Korea has the highest pedestrian death rates caused by traffic accidents, with the highest speed limit regulation (60 km/h) among OECD member countries in most urban areas.4 The allowance of such high speed limits may well contribute to pedestrian fatalities on dangerous urban roads where safety measures for pedestrians, such as traffic calming and police enforced traffic control, are not readily available due to budget, especially in poorest districts.26

    RR values for fall mortality rates in this study were higher in men, older ages and in those with a lower educational level. In the elderly age group, several studies27–29 reported an inverse association between individual level SEP and fall mortality rate, as found in our study. Of two relevant studies taking into account both individual and area level SEP,11 ,13 a study in Barcelona13 reported that there was a greater fall mortality only in women of all ages with no schooling. A Canadian study11 found that the lowest SEP (occupation, income) individuals were at greater fall mortality risk than the highest SEP individuals, but only in men. The magnitude and pattern of individual socioeconomic inequalities in fall mortality varied by region and demographic characteristics.

    This study showed that fall mortality rates in the most deprived areas were greater than those in the least deprived areas, after adjusting for individual level SEP. Likewise, a study in Barcelona13 reported the association with neighbourhood unemployment and fall mortality rate, after adjusting for individual level SEP. In contrast, studies in European cities12 and Canada11 reported that there were no associations between area level deprivation and mortality from fall injuries. In South Korea, senior citizens with a lower household income were reportedly more susceptible to fall related fatalities due to the increased risks from their residential conditions, with hazards such as steep street slopes, poorly designed stairs and slippery bathroom tiles.30 Other individual and geographic characteristics, including individual comorbidities, access to safety aids and access to appropriate facilities, might have contributed to the differences in fall mortality rates.

    The risk of suicide mortality was greater in the low educational group after adjusting for sex, age and area level SEP. This finding is consistent with that of other studies using various individual SEP indicators, such as education and occupation in South Korea.31–33 Studies from other countries3 showed mixed results. The reason for the robust individual socioeconomic inequalities in suicides in South Korea may result from a large increase in social polarisation since the 1997 Asian economic crisis.32 ,33

    Suicide mortality rates were greater in the most deprived areas than in the least deprived areas when individual variables were adjusted for. Previous South Korean studies33 ,34 reported a close association between suicide and area level SEP. Cubbin et al10 and Gotsens et al12 reported the association between area deprivation and suicide. However, Borrell et al13 claimed no association between suicide rates and area SEP levels, after adjusting for individual level SEP. There seem to be particular reasons for the consistent results of the inverse relationship between area level SEP and suicide rates across some South Korean studies,33 ,34 including this study, even though mixed results were found in other studies.10 ,12 ,13 One reason could be that in South Korean metropolitan cities, the most deprived urban areas have much less public welfare expenditure than the least deprived areas, which might have caused higher suicide rates, with a lack of social stress buffering function resulting from insufficient social support (financial aids and social welfare services).35 ,36 In this study, the effect size of area level SEP with regard to all types of injury mortality was much smaller than that of individual level SEP. This result supports Pickett and Pearl's findings,37 drawn from reviewing studies of neighbourhood socioeconomic context on health outcome, concluding that the contextual effect is generally modest and smaller than the compositional effect. On the other hand, Reijneveld38 reported that in small areas of Dutch cities, the effect size of the socioeconomic context of neighbourhoods on health outcome remained very high in some cities, even after adjusting for individual educational level. In this regard, to measure more elaborately the contextual effects of neighbourhood on injury mortality, further studies need to consider other study designs, such as targeting smaller areas than the districts used in this study when such data are available.

    There were several limitations in this study. We used the 2005 census population as the denominator due to the availability of data. This does not exactly represent district specific population and area characteristics in other years. This analysis essentially represents a snapshot of the effects of neighbourhoods on injury mortality. In addition, a methodological weakness was that only educational attainment was used as an individual SEP indicator. However, it was inevitable as education was available in both the census and death certificate data.18 Lastly, areas at the Gu level (districts) were too large to estimate the association between socioeconomic context of neighbourhoods and health outcomes. Nonetheless, it should be noted that Gu is the administrative level of minimal autonomous local government in South Korea and is primarily responsible for protecting residents from external causes of death. Thus the analysis based on the Gu area would provide more valuable information in terms of policy planning and implementation for injury prevention in South Korea.

    In conclusion, the most deprived urban areas in South Korea suffered from mortalities from traffic accidents, falls, suicide and all injuries, even after compositional effects were taken into account. From the standpoint of an integrated approach of injury prevention suggested by Cohen et al,16 more comprehensive interventions by local government needs to be targeted for all types of fatal injury. Moreover, the results of this study imply that the most deprived urban districts should be given a higher priority for support for injury prevention by both local district and metropolitan governments.

    What is already known

    • There have been relatively few studies, especially in non-Western countries, examining the impact of socioeconomic status in an urban area on adult fatal injuries, after adjusting for the compositional effects of individual level characteristics.

    What this paper adds

    • This study adds to the literature dealing with intra-urban area inequalities by deprivation levels in fatal injuries. The results of this study showed that the most deprived urban areas in South Korea suffered from mortalities from traffic accidents, falls, suicide and all injuries, even after compositional effects were taken into account.

    • The results of this study imply that the most deprived urban districts should be given a higher priority for support for injury prevention by both local district and metropolitan governments.

    References

    1. ↵
      1. Laflamme L
      . Social inequalities in injury risks: knowledge, accumulated and strategies for the future. Stockholm: National Institute of Public Health, 1998.
    2. ↵
      1. Blane D
      . Social determinants of health—socioeconomic status, social class, and ethnicity. Am J Public Health 1995;85:903–5.
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      1. Laflamme L,
      2. Burrows S,
      3. Hasselberg M
      . Socioeconmic differences in injury risks: a review of findings and discussion of potential countermeasures. Copenhagen: WHO Regional Office for Europe, 2009.
    4. ↵
      Group ITSDaA. Road safety annual report 2013. Paris: Orgainzation for Economic Cooperation and Development (OECD) / International Traffic Forum (ITF), 2013.
    5. ↵
      OECD. OECD Health Data: Health status.
    6. ↵
      1. Davis M
      . Planet of slums. New York: Verso, 2006.
    7. ↵
      1. Pasarin M,
      2. Borrell C,
      3. Plasencia A
      . Two patterns of social inequalities in mortality in Barcelona, Spain? Gac Sanit 1999;13:431–40.
      OpenUrlCrossRefPubMed
    8. ↵
      World Health Orgainization, UN-HABITAT. Hidden cities: unmasking and overcoming health inequalities in urban settings. Geneva: WHO/UN-HABITAT, 2010.
    9. ↵
      Microdatasystem (MDSS) (database on the Internet), 2011 (cited Feb 17 2013). http://mdss.kostat.go.kr/mdssext/.
    10. ↵
      1. Cubbin C,
      2. LeClere FB,
      3. Smith GS
      . Socioeconomic status and injury mortality: individual and neighbourhood determinants. J Epidemiol Community Health 2000;54:517–24.
      OpenUrlAbstract/FREE Full Text
    11. ↵
      1. Burrows S,
      2. Auger N,
      3. Gamache P,
      4. et al
      . Individual and area socioeconomic inequalities in cause-specific unintentional injury mortality: 11-year follow-up study of 2.7 million Canadians. Accid Anal Prev 2012;45:99–106.
      OpenUrlCrossRefPubMed
    12. ↵
      1. Gotsens M,
      2. Mari-Dell'olmo M,
      3. Perez K,
      4. et al
      . Socioeconomic inequalities in injury mortality in small areas of 15 European cities. Health Place 2013;17:165–72.
      OpenUrl
    13. ↵
      1. Borrell C,
      2. Rodriguez M,
      3. Ferrando J,
      4. et al
      . Role of individual and contextual effects in injury mortality: new evidence from small area analysis. Inj Prev 2002;8:297–302.
      OpenUrlAbstract/FREE Full Text
    14. ↵
      1. Gotsens M,
      2. Mari-Dell'Olmo M,
      3. Martinez-Beneito MA,
      4. et al
      . Socio-economic inequalities in mortality due to injuries in small areas of ten cities in Spain (MEDEA Project). Accid Anal Prev 2011;43:1802–10.
      OpenUrlCrossRefPubMed
    15. ↵
      1. Kim MH,
      2. Subramanian SV,
      3. Kawachi I,
      4. et al
      . Association between childhood fatal injuries and socioeconomic position at individual and area levels: a multilevel study. J Epidemiol Community Health 2007;61:135–40.
      OpenUrlAbstract/FREE Full Text
    16. ↵
      1. Cohen L,
      2. Miller T,
      3. Sheppard MA,
      4. et al
      . Bridging the gap: bringing together intentional and unintentional injury prevention efforts to improve health and well being. J Safety Res 2003;34:473–83.
      OpenUrlCrossRefPubMedWeb of Science
    17. ↵
      1. Galobardes B,
      2. Shaw M,
      3. Lawlor DA,
      4. et al
      . Indicators of socioeconomic position. In: Oakes JM, JS K, eds. Methods in social medicine. San Francisco: John Wiley & Sons, 2006:47–85.
    18. ↵
      1. Kim H,
      2. Khang Y
      . Reliability of education and occupation class: a comparison of health survey and death certificate data. J Prev Med Public Health 2005;38:443–8.
      OpenUrlPubMed
    19. ↵
      1. Morris R,
      2. Carstairs V
      . Which deprivation? A comparison of selected deprivation indexes. J Public Health Med 1991;13:318–26.
      OpenUrlAbstract/FREE Full Text
    20. ↵
      1. Fletcher JM,
      2. Morris RD,
      3. Francis DJ
      . Methodological issues in the classification of attention-related disorders. J Learn Disabil 1991;24:72–7.
      OpenUrlAbstract/FREE Full Text
    21. ↵
      1. Turrell G,
      2. Mathers C
      . Socioeconomic inequalities in all-cause and specific-cause mortality in Australia: 1985–1987 and 1995–1997. Int J Epidemiol 2001;30:231–9.
      OpenUrlAbstract/FREE Full Text
    22. ↵
      1. Braver ER
      . Race, Hispanic origin, and socioeconomic status in relation to motor vehicle occupant death rates and risk factors among adults. Accid Anal Prev 2003;35:295–309.
      OpenUrlCrossRefPubMedWeb of Science
    23. ↵
      1. Steenland K,
      2. Halperin W,
      3. Hu S,
      4. et al
      . Deaths due to injuries among employed adults: the effects of socioeconomic class. Epidemiology 2003;14:74–9.
      OpenUrlCrossRefPubMedWeb of Science
    24. ↵
      1. Shinar D,
      2. Schechtman E,
      3. Compton R
      . Self-reports of safe driving behaviors in relationship to sex, age, education and income in the US adult driving population. Accid Anal Prev 2001;33:111–16.
      OpenUrlCrossRefPubMedWeb of Science
    25. ↵
      1. Graham D,
      2. Glaister S,
      3. Anderson R
      . The effects of area deprivation on the incidence of child and adult pedestrian casualties in England. Accid Anal Prev 2005;37:125–35.
      OpenUrlCrossRefPubMedWeb of Science
    26. ↵
      1. Park K,
      2. Hwang SS,
      3. Lee JS,
      4. et al
      . Individual and areal risk factors for road traffic injury deaths: nationwide study in South Korea. Asia Pac J Public Health 2010;22:320–31.
      OpenUrlAbstract/FREE Full Text
    27. ↵
      1. Huisman M,
      2. Kunst AE,
      3. Andersen O,
      4. et al
      . Socioeconomic inequalities in mortality among elderly people in 11 European populations. J Epidemiol Community Health 2004;58:468–75.
      OpenUrlAbstract/FREE Full Text
    28. ↵
      1. Reyes-Ortiz CA,
      2. Al Snih S,
      3. Loera J,
      4. et al
      . Risk factors for falling in older Mexican Americans. Ethn Dis 2004;14:417–22.
      OpenUrlPubMedWeb of Science
    29. ↵
      WHO. WHO global report on falls prevention in older age. Geneva: World Health Organization, 2007.
    30. ↵
      1. Kim S-R,
      2. Ha Y-C,
      3. Kim J-R,
      4. et al
      . Incidence of hip fractures in Jeju Island, South Korea: a prospective study (2002–2006). Clin Orthop Surg 2010;2:64–8.
      OpenUrlPubMed
    31. ↵
      1. Kim MD,
      2. Hong SC,
      3. Lee SY,
      4. et al
      . Suicide risk in relation to social class: a national register-based study of adult suicides in Korea, 1999–2001. Int J Soc Psychiatry 2006;52:138–51.
      OpenUrlAbstract/FREE Full Text
    32. ↵
      1. Lee WY,
      2. Khang YH,
      3. Noh M,
      4. et al
      . Trends in educational differentials in suicide mortality between 1993–2006 in Korea. Yonsei Med J 2009;50:482–92.
      OpenUrlCrossRefPubMedWeb of Science
    33. ↵
      1. Kim MH,
      2. Jung-Choi K,
      3. Jun HJ,
      4. et al
      . Socioeconomic inequalities in suicidal ideation, parasuicides, and completed suicides in South Korea. Soc Sci Med 2010;70:1254–61.
      OpenUrlCrossRefPubMed
    34. ↵
      1. Khang YH,
      2. Yun SC,
      3. Hwang IA,
      4. et al
      . Changes in mortality inequality in relation to the South Korean economic crisis: use of area-based socioeconomic position. J Prev Med Pub Health 2005;38:359–65.
      OpenUrl
    35. ↵
      1. Zimmerman SL
      . States’ spending for public welfare and their suicide rates, 1960 to 1995: what is the problem? J Nerv Ment Dis 2002;190:349–60.
      OpenUrlCrossRefPubMedWeb of Science
    36. ↵
      1. Yur'yev A,
      2. Varnik A,
      3. Varnik P,
      4. et al
      . Role of social welfare in European suicide prevention. Int J Soc Welf 2012;21:26–33.
      OpenUrlCrossRef
    37. ↵
      1. Pickett KE,
      2. Pearl M
      . Multilevel analyses of neighbourhood socioeconomic context and health outcomes: a critical review. J Epidemiol Community Health 2001;55:111–22.
      OpenUrlAbstract/FREE Full Text
    38. ↵
      1. Reijneveld SA
      . Neighbourhood socioeconomic context and self reported health and smoking: a secondary analysis of data on seven cities. J Epidemiol Community Health 2002;56:935–42.
      OpenUrlAbstract/FREE Full Text

    Footnotes

    • Correction notice The order of authorship has been modified since published Online First.

    • Contributors JL performed the analyses and drafted the manuscript. W-YL suggested the main idea for this paper and advised on all aspects, including the analytical frame and interpretation of the results. MN advised on the statistical technique. Y-HK advised on the analytical frame of the work and contributed to the revision of the draft.

    • Competing interests None.

    • Provenance and peer review Not commissioned; externally peer reviewed.