Aim To determine whether emergency hospital admission rates (EAR) for common paediatric conditions in Greater London are associated with measures of child well-being and deprivation.
Design Retrospective analysis of hospital episode statistics and secondary analysis of the Index of Multiple Deprivation (IMD) 2007 and Local Index of Child Well-Being (CWI) 2009.
Setting 31 Greater London primary care trusts (PCTs).
Outcome measures EAR in PCTs for breathing difficulty, feverish illness and/or diarrhoea.
Results 24 481 children under 15 years of age were discharged following emergency admission for breathing difficulty, feverish illness and/or diarrhoea during 2007/2008. The EAR for breathing difficulty was associated with the IMD (Spearman's rho 0.59, p<0.001) and IMD indicators of: overcrowding (Spearman's rho 0.62, p<0.001), houses in poor condition (Spearman's rho 0.55, p=0.001), air quality (Spearman's rho 0.53, p=0.002), homelessness (Spearman's rho 0.44, p=0.013), and domains of the CWI: housing (Spearman's rho 0.64, p<0.001), children in need (Spearman's rho 0.62, p<0.001), material (Spearman's rho 0.58, p=0.001) and environment (Spearman's rho 0.53, p=0.002). There were no statistically significant relationships between the EAR of children admitted for feverish illness and diarrhoea or aged under 1 year for any condition, and the IMD, either IMD indicators or CWI domains.
Conclusions Housing and environmental factors are associated with children's demand for hospital admission for breathing difficulty. Some associations are stronger with the CWI than the IMD. The CWI has potential to identify priority PCTs for housing and environment interventions that could have specific public health benefits for respiratory conditions.
Statistics from Altmetric.com
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.
Deprivation and other forms of social and material disadvantage adversely affect children's health; however, the causal mechanisms are complex and poorly understood.1,–,5 It is not known how disadvantage affects rates of admission for common presenting conditions at different ages. The Index of Multiple Deprivation (IMD) measures deprivation at the small area level and is used by local and central government and primary care trusts (PCTs) to direct policies and funding to improve the quality of life in disadvantaged communities. It comprises seven weighted domains: income (22.5%), employment (22.5%), health and disability (13.5%), education, skills and training (13.5%), barriers to housing and services (9.3%), living environment (9.3%) and crime (9.3%).6 Although there is a supplementary Income Deprivation Affecting Children Index (IDACI),6 the IMD is not specific to childhood disadvantage. The Local Index of Child Well-Being (CWI) provides additional information that increases the potential for understanding the effects of specific aspects of disadvantage on children's health in England.7 Identification of associations between the CWI and specific child health conditions could add to understanding of the mechanisms by which deprivation affects health, and contribute to planning acute healthcare services and public health interventions.
What is already known on this topic
▶ Health and healthcare use in childhood are associated with disadvantage.
▶ The mechanisms by which deprivation affects childhood health are uncertain.
▶ The Local Index of Child Well-Being (CWI) is a child focused alternative to the Indices of Multiple Deprivation (IMD).
What this study adds
▶ Emergency admission rates (EAR) for respiratory disease were associated with overcrowding, houses in poor condition, air quality and homelessness indicators (IMD) and housing, children in need, material and environment domains (CWI) but not in children <1 year old.
▶ EAR for feverish illness and diarrhoea were not associated with either IMD indicators or CWI domains.
The CWI comprises seven domains combined with equal weights: material, health, education, crime, housing, environment, and children (at risk of being) in need.7 8 Both the IMD and CWI are constructed at the level of lower super output areas (LSOAs) which have average populations of 1500. The 32 482 LSOAs in England are ranked from 1 (the area with highest well-being) to 32 482 (the area with lowest well-being). The CWI provides a potential method for examining the relationships between specific domains of disadvantage/deprivation, health conditions and service use.The aim was to determine whether hospital emergency admission rates (EAR) for common paediatric conditions in Greater London are associated with measures of the CWI and deprivation in four age groups: <1 year, 1–4 years, 5–9 years and 10–14 years.
The design incorporated a retrospective analysis of hospital episode statistics (HES) and secondary analysis of the most recently published IMD (2007) and CWI (2009).
Children's emergency admissions were identified for diagnoses associated with the three commonest paediatric presentations to hospital: breathing difficulty, feverish illness and/or diarrhoea.9 A group of International Classification of Diseases (ICD)-10 codes was selected to derive these three categories (see online supplementary table 1). HES were obtained for all children resident in one of the 31 PCTs comprising the London Strategic Health Authority discharged during the 2007/2008 financial year following emergency admission to hospitals across England. In-patient spells (comprising both admission and discharge) per 1000 children aged 0–14 years resident in each PCT were used to calculate EAR. The denominator was mid-2007 population estimates for the age bands published by the Office for National Statistics (ONS): <1 year, 1–4 years, 5–9 years and 10–14 years.
The indicators contributing to the IMD were obtained for each LSOA in Greater London and average scores were calculated for each PCT. The percentage of LSOAs in each PCT that were in the fifth quintile (lowest well-being) of the 32 482 LSOAs in England was calculated for the overall CWI measure and each of its constituent domains.7 As the health domain includes an EAR of children aged 0–18 years in each LSOA as an indicator, there is an in-built association with the EAR for children aged 0–14 years. This domain has therefore been excluded from our reporting in order to avoid an obvious bias.
Data were analysed descriptively using SPSS 15, with associations measured using Spearman's rho correlation due to expectedly skewed distributions. Calculation of associations using Kendall's taub correlation and the percentage of LSOAs in each PCT in the lowest decile of the national distribution of the CWI measure and its domains was conducted as a sensitivity analysis. This confirmed the pattern of the results.
There were 24 481 discharges following emergency admission, 56.2% (n=13 750) for breathing difficulty, 26.0% (n=6373) for feverish illness and 17.8% (n=4358) for diarrhoea. The EAR was 17.9 but varied across PCTs with the highest EAR four times the size of the lowest (table 1, figure 1, and supplementary online table 2).
There is a wide range of disadvantage across Greater London PCTs. The percentage of LSOAs in the lowest quintile of the CWI ranged from 1.0 (Kingston) to 94.6 (Tower Hamlets). Some PCTs had high concentrations of either the most or the least deprived populations. In 10 PCTs more than 50% of their constituent LSOAs belonged to the lowest quintile of the CWI, including three with more than 80% of LSOAs in the lowest CWI quintile (Tower Hamlets, City and Hackney, and Islington). In contrast, in 13 PCTs less than 20% of LSOAs belonged to the lowest quintile of the national distribution of the CWI. Within this pattern there were differences in the profile of indicators and domains of PCTs: the distribution of deprivation is complex and a range of patterns of disadvantage emerge at the level of indicators and domains (see online supplementary table 3).
The EAR for children aged 0–14 admitted for breathing difficulty (table 2) was associated with the IMD (Spearman's rho 0.59, p<0.001) and IMD indicators of: overcrowding (Spearman's rho 0.62, p<0.001), houses in poor condition (Spearman's rho 0.55, p=0.001), air quality (Spearman's rho 0.53, p=0.002) and homelessness (Spearman's rho 0.44, p=0.013). There was also association between the EAR of children admitted for breathing difficulty and the percentage of LSOAs in the lowest quintile and housing (Spearman's rho 0.64, p<0.001), children in need (Spearman's rho 0.62, p<0.001), material (Spearman's rho 0.58, p=0.001) and environment (Spearman's rho 0.53, p=0.002) domains of the CWI (table 3).
There were no statistically significant relationships between the EAR of children aged 0–14 years admitted for feverish illness or diarrhoea and the IMD, IMD indicators or domains of the CWI (table 3). Thus the relationship between admissions and the IMD, IMD indicators and CWI domains was explained by respiratory conditions.
The relationship between the EAR for breathing difficulty and both the IMD and CWI was statistically significant except for children aged under 1 year old (table 4).
There are clear associations between the EAR of PCTs for respiratory illness and indicators of housing and of environment in both the IMD and the CWI. The relationship is strongest with overcrowding, followed by houses in poor condition and air quality, and homelessness. The correlation between admission rates and the CWI housing domain is stronger than with IMD housing measures, which suggests that the CWI domain is a more specific indicator of risk associated with housing than the IMD measures.
Selected housing indicators are spread across two domains of the IMD 2007: living environment (which includes housing in poor condition) and barriers to housing and services (including overcrowding and homelessness). The CWI housing domain combines four indicators into two equally-weighted subdomains: access to housing (including overcrowding, shared accommodation and homelessness indicators) and quality of housing derived from a lack of central heating indicator.
These results suggest that the CWI can be used alongside specific IMD housing indicators to plan children's services and target public health interventions. Overcrowding is a particularly important indicator of respiratory risk and the consistently stronger association between emergency admissions for breathing difficulty and the CWI housing domain compared with the IMD housing indicators identifies a cumulative negative effect of a combination of housing factors on children's respiratory health. These results suggest that interventions should reduce overcrowding and improve housing quality as well as access to housing and air quality. More respiratory healthcare interventions for children over 1 year old, such as asthma clinics, are required in PCTs where there is more overcrowding, poorer quality of housing and air quality, and homelessness.
It is important to note that both the IMD and the CWI are aggregate measures. Aggregation can hide variations within areas and it identifies socio-economic and other conditions affecting a group of people by where they live rather than their individual circumstances. Further loss of discrimination is inevitable when these indicators are assessed at the PCT level. In addition, the calculation of EAR depends on the accuracy of HES data and ONS population estimates. There is a risk of errors in diagnosis codes in the HES,10 although this was reduced by selection of conditions predominantly at the ‘block’ ICD level rather than specific diagnostic codes. The population denominator is based on estimation of changes since the 2001 census and it is known that there is potential for student and other migration to affect the accuracy of population estimates in some areas of London.11 The necessity to calculate EAR at the PCT rather than individual level precluded multivariate modelling that could control for confounders.
Some London PCTs have concentrations of either the least or the most deprived populations. The contrasts observed may not be repeated in areas where deprivation is less concentrated, for example in rural areas.12 Therefore the findings may not be replicated outside London. EAR can also be affected by local policies on criteria for admission and management, observation and assessment units that enable extended observation without overnight hospital stay, and/or community children's nursing services. Variation could also be associated with local differences in clinical and coding practice around the point in the paediatric pathway at which a child is deemed to have been ‘admitted’.
The relationships with EAR for breathing difficulty are consistent with evidence that both housing and environment can be factors in respiratory illness. Poor housing and air quality may be part of the mechanism by which socio-economic disadvantage influences children's respiratory health,13 but these are not the only, or isolated, factors. Acute respiratory disease hospitalisation has been associated with inadequate housing conditions, but the complexity of the relationships was illustrated by multivariate analysis which only found association with maternal education, previous history of wheezing, lack of breastfeeding, use of pacifiers, maternal smoking, age under 6 months, and being male.14 Overcrowding has also been identified as one of a number of risk factors for hospitalisations, which include: low birth weight, perinatal problems, chronic illness, death of a sibling under the age of 5, grandmother as day caretaker, and mother's higher educational level.15 These studies were conducted in Brazil, which may limit their applicability to other settings. In addition, the association between housing conditions and respiratory illness is multi-factorial, including exposures associated with housing quality (damp, mould, rodents) and behaviours (passive smoking, pet ownership).16
There is conflicting evidence about the relationship between environmental pollution and respiratory disease. Air pollution is associated with increases in the risk of death, chronic disease in children and exacerbation of illnesses.17 While there is evidence of association with particulates and ozone, it is not clear which pollutants are most responsible.17 Neither carbon monoxide (CO) nor ozone were associated with asthma hospitalisation in either low or high socio-economic groups in a Canadian study.18 However, CO pollution was associated with child hospitalisations for asthma in California and the effect was greater for children of lower socio-economic status.19 Exposure to air pollution has also been associated with new onset asthma.20 Nitrogen dioxide, a traffic-related air pollutant, has been associated with reduced expiratory flows in schoolchildren, although there was not a consistent association with prevalence of respiratory symptoms or allergic sensitisation.21
Clear association has been identified between low social class at the time of birth and increased morbidity as indicated by hospital admissions during the first 10 years of life for conditions that could be analysed, with the exception of neoplasms.22 The only age group for which the social class gradient was consistent across all conditions was the youngest, aged 0–3 years.22 Therefore the disadvantage associated with social class is associated with higher rates of admission in early life. Being born into a lower social class is associated with a higher duration of hospitalisation during the first year of life, along with other maternal risk factors including smoking during pregnancy, primiparous pregnancy and maternal age.2 Although the impact of social class on hospital inpatient admissions is higher during years 3–10 as compared with the first 2 years of life,2 it is notable that in our analysis of Greater London neither the IMD nor the CWI was associated with rates of emergency admission for breathing difficulty, fever or diarrhoea in the first year of life.
There are specific factors that can affect health in the first year of life, including birth weight23 and breast feeding.24 It may also be that there is a lower threshold for parental help-seeking for children under 1 year of age across all deprivation groups. Acute illness can often be non-specific in children under 1 year of age and gastrointestinal symptoms, for example, are common in infancy.25 Healthcare professionals may be more likely to adopt a more cautious approach with children under 1 year of age in decisions about admission to hospital. Potential local factors include confidence in the abilities of general practitioners (GPs) to diagnose and treat infants, the geographical accessibility of hospitals, local cultures of help-seeking behaviour and healthcare use, and differences in admitting practice at hospitals. Primary care and ambulatory care systems may lack confidence to care for such infants without hospital admission. Thus risk avoidance by parents and across the healthcare system could contribute to higher admission rates in children under 1 year of age.
The rate of children's emergency admission has increased in England in recent years.26 27 There has been a particularly marked increase in the rate among children aged less than 1 year.27 Unplanned short stay admissions are of particular note, with their rate increasing by 46.8% from 123 per 1000 child years in 1997 to 181 per 1000 child years in 2006.27 Performance measurement in Accident and Emergency departments may account for part of this increase as children may be admitted to prevent breach of 4 h targets. However, many children may have been treatable out of hospital,27 perhaps by community children's nurses (CCN) through further development of direct referral pathways between GPs and CCN teams, effectively by-passing secondary care. These findings may suggest that parents are increasingly unwilling to manage self-limiting childhood complaints at home with the support of primary care services. Parental confidence in managing such conditions may contribute to the association between admissions and maternal age and education,28 29 but these maternal factors are not visible in our data because they do not contribute directly to either the IMD or CWI.
EAR for diarrhoea and for feverish illness were not associated with either the IMD or the CWI. Poverty is associated with higher risk of mortality from diarrhoeal disease in developing countries,30 but there is less evidence of association with deprivation in developed countries. Access to care has been associated with mortality from diarrhoea in the USA31 and availability of paediatric beds was associated with gastroenteritis admission rates in Ontario.32 It is possible that admissions practice at different hospitals and the implementation of new advice such as the NICE fever guidelines in 2007,33 could also affect admission rates. Lower maternal education and single parenthood were associated with hospitalisation for infectious disease in a Danish study28 and the risk of repeated admissions was also increased in children from low-income families.28 Maternal factors, including lower levels of education, younger age and single parenthood, were also associated with hospitalisation for rotavirus acute gastroenteritis in the USA, along with insurance status, low birth weight and not being breast fed.29 Some consequences of disadvantage have direct physical effects on infants, including low birth weight, which is associated with disadvantage34 and can affect digestive function25 and breast feeding, which is associated with maternal age and education.35 Deprivation amplifies the increased risk of diarrhoeal disease with formula rather than breast feeding, possibly as a result of overcrowding increasing the spread of infection.24 Other effects of maternal age, education and single parenthood are indirect. They could suggest that limitations in resources of education, experience and social support, reduce mothers' ability to prevent and manage diarrhoea and infectious illness. Neither the IMD nor the CWI includes specific identifiers of maternal age, education or social support, which could explain why they were not associated with EAR for either diarrhoea or feverish illness.
Housing and environmental factors including overcrowding, housing quality, air quality and homelessness were associated with children's hospital admission rates for breathing difficulty in Greater London PCTs. These data suggest that reducing overcrowding and improving the quality and accessibility of housing could reduce respiratory morbidity and emergency admissions. However, further research is required to explain substantial variations in admission rates before the age of 1 year. The correlation with breathing difficulty was strongest with the CWI housing domain, which may be a more specific index of child-specific disadvantage associated with higher rates of hospital admission for respiratory conditions. The IMD and CWI can identify communities in which interventions to improve children's housing and reduce exposure to air pollution could have specific public health benefits for respiratory conditions. They also provide a basis for planning healthcare services to meet the increased needs for respiratory care of children over 1 year of age living in areas where there is more overcrowding, poorer quality of housing, air pollution and homelessness.
The data sources were Hospital Episode Statistics (inpatient data) 2007/08, The NHS Information Centre for Health and Social Care; Indices of Multiple Deprivation 2007, Local Index of Child Well-Being 2009, Communities and Local Government; Mid-Year Estimates 2007, Office for National Statistics.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.