Review Article

20 Years of Research on Socioeconomic Inequality and Children's—Unintentional Injuries Understanding the Cause-Specific Evidence at Hand

Table 5

Individual-level studies for childhood burn, fall, poisoning, and drowning injuries: summary of methodological features and results ( ).

Author & year country (city/region)Outcome/s b, F, P, D Age group/s data source SES measureAnalysis covariatesResults: the level of 95% is used for all confidence intervals (CI)

Cho et al. 2007 South Korea (whole country)D Deaths, stratified by sex and age group10–14 and 15–19 years R: death register, health insurance beneficiary datasetParental income (based on insurance contribution - 3 levels)Cox proportional hazards model NoneDrowning deaths showed no socioeconomic gradient among boys or girls for either age group (eg, for boys 10–19 years RR=1.26, CI 0.82–1.92, p for trend = 0.28, lowest compared to highest income tertile)
Delgado et al. 2002 Peru (Lima)B Hospitalisations (all burn types)0–17 years Q: structured questionnaire with guardiansHousehold income (2 levels), crowding (2 levels), maternal education (2 levels)Logistic regression (case-control study) No water supply, living room in house, own house, patient is not child of household headChildren in low income (OR=2.8; CI 2.0–3.9) and crowded (OR=2.5; CI 1.7–3.6) households have increased risk of burn injuries compared to those in households with higher income and no crowding; children of mothers with at least a high school education have lower risks compared to those with mothers without this education (OR=0.6; CI 0.5–0.9)
Donroe et al. 2009 Peru (Lima)P, B, F Severe enough to require medical consultation0–18 years I: household survey with guardian or with child if aged 12 yearsPoverty (2 levels), parental education (2 levels)Logistic regression Sex, age, other SES, overcrowding, number of children in the homeNo association between SES and individual injury in multivariate model but increased odds of falls for children who are from homes that are both poor and with low parental education (OR=1.30; CI 1.0–1.7). Children in poor households had increased odds of burn injuries (adjusted OR=1.34; CI 1.0–1.8) compared to those in more affluent households
Edwards et al. 2006 United Kingdom (England and Wales)B Deaths from exposure to smoke, fire, and flames0–15 years R: population-based death registerFamily occupational status (8 levels)Death rates (95% CI) NoneChildren from family with the least favourable occupational status had 37.7 (CI 11.6–121.9) times higher death rates than those from the most favourable one
Engström et al. 2002 Sweden (whole country)F Hospitalisations and deaths combined, stratified by age0–4, 5–9, 10–14 and 15–19 years R: linkage of health, death and census recordsParental social class (4 levels)Logistic regression, slope index of inequality, relative index of inequality Parents’ country of birth, single parent home, receipt of welfare benefitsNo association between SES and risk of fall injuries except among 0–4 year olds (RR=1.08; CI 1.0–1.1 for children both of unskilled and skilled workers compared with children of intermediate and high level employees)
Forjuoh et al. 1995 Ghana (Ashanti region)B Injuries with evidence of physical scar0–5 years I: household survey of caretakersMaternal education (2 levels)Logistic regression (case-control study) Presence of pre-existing impairment in child, history of sibling burn, storage of flammable substance in homeMaternal education was not significantly associated with childhood burns (OR=0.76, CI 0.55–1.05 for educated mother compared to a mother without education)
Giashuddin et al. 2009 Bangladesh (randomly selected areas of whole country)D Deaths and nonfatal injuries separately1–4 years I: household surveyAssets Index (quintiles)Concentration indexDrowning morbidity and mortality were 3.8 and 7.0 times higher, respectively, in the least as compared the most deprived quintile. Concentration indices −.21 and −.28, respectively) showed significant inequalities among the groups ( )
Gilbride et al. 2006 Canada (Alberta province)P, B Cases requiring physician consultation0–17 years R: administrative health databaseReceipt of healthcare premium subsidy (as proxy for low SES – 2 levels)Logistic regression Sex, ageCompared to children from families without subsidies, those from low SES families had higher odds of burns (OR=1.35; CI 1.3–1.4) and poisoning (OR=1.60; CI 1.5–1.7)
Laursen & Nielson 2008 Denmark (whole country)P, B, F Injuries occurring at home and seen in emergency department. Falls: from 1 metre0–14 years R: national injury registerParents’ education (3 levels), and income (4 levels)Poisson regression Age, sex, distance from hospital, number of children, age at childbirth, family type, crowding, dwelling typeIncreasing injury with decreasing SES for each cause. Compared to children of parents with a tertiary education, those of parents with a primary school education had higher risks of poisoning (RR=1.9; CI 1.6–2.3), burns (RR=1.6; CI 1.4–1.9) and high falls (RR=1.4; CI 1.2–1.7). Compared to children of parents in the most affluent group, those of parents in the lowest income group had higher risks of poisoning (RR=1.7; CI 1.4–2.1), burns (RR=1.9; CI 1.6–2.3) and high falls (RR=1.2; CI 1.0–1.4)
Roberts 1997 United Kingdom (England and Wales)P, B, F Deaths0–15 years R: death registerSocial class of father (6 levels)Poisson regression NoneMortality differentials were steepest for fire-related deaths (OR=1.89; CI 1.8–2.0), followed by falls (OR=1.46; CI 1.3–1.6) and poisoning (OR=1.36; CI 1.1–1.6)
Scholer 1998 United States (State of Tennessee)B House fires resulting in at least one fatality0–5 years R: linkage of birth certificates, census data & death certificatesMaternal education (4 levels), neighbourhood income (5 levels)Poisson regression (cohort study) Maternal age, race, marital status, residence, number of children, first prenatal care visit, child sex & gestational ageLow maternal education was positively associated with an increased risk of fatal fire events (RR=19.36; CI 2.6–142.4 for 12 years education compared to 16 years). The association between neighbourhood income and injury did not persist in the multivariate analysis

Note B=burns, F=falls, P=Poisoning, D=Drowning; R=register; I= interview, Q=self-administered questionnaire.