Child injury: using national emergency department ...

2 downloads 0 Views 209KB Size Report
Jul 10, 2013 - Karen Hughes,1 Philip McHale,1 Sacha Wyke,2 Helen Lowey,1 Mark A Bellis1. 1Centre for Public Health,. Liverpool John Moores. University ...
Downloaded from injuryprevention.bmj.com on July 10, 2013 - Published by group.bmj.com

IP Online First, published on July 10, 2013 as 10.1136/injuryprev-2013-040816 Original article

Child injury: using national emergency department monitoring systems to identify temporal and demographic risk factors Karen Hughes,1 Philip McHale,1 Sacha Wyke,2 Helen Lowey,1 Mark A Bellis1 1

Centre for Public Health, Liverpool John Moores University, Liverpool, UK 2 Knowledge and Intelligence Team (North West), Public Health England, Liverpool, UK Correspondence to Dr K Hughes, Centre for Public Health, Liverpool John Moores University, Henry Cotton Building, 15-21 Webster Street, Liverpool L3 2ET, UK; [email protected] Received 6 March 2013 Revised 29 May 2013 Accepted 16 June 2013

ABSTRACT Background Injury is a leading cause of death in children. Emergency department (ED) data offer a potentially rich source of data on child injury. This study uses an emerging national ED data collection system to examine sociodemographics and temporal trends in child injury attendances in England. Methods Cross sectional examination of ED attendances for key injury types made by children aged 0–14 years between April 2010 and March 2011 (road traffic injury (RTI) n=21 670; assault n=9529; deliberate self harm (DSH) n=3066; sports injury n=88 250; burns n=22 222; poisoning n=12 446). Multivariate analyses examined the impact of demographics (age, gender, residential deprivation) and temporal events (day, month, school and public holidays) on risk of attendance for different injury types. Results For most injury types, attendance increased with deprivation. The attendance ratio between children from the poorest and richest deprivation quintiles was greatest for assaults (4.21:1). Conversely, sports injury attendance decreased with deprivation. Males made more attendances than females for all but DSH. Age and temporal profiles varied by injury type. Assault attendances reduced at weekends while burns attendances increased. RTI and sports injury attendances were increased during school term times. Conclusions ED data can provide a major epidemiological resource for examining both temporal and demographic risks of child injury. Emerging systems, such as the one analysed here, can already inform the targeting of prevention, and with improved data coding and use, their utility would be greatly strengthened.

Emergency departments (EDs) and other acute care units offer an important source of data on child injury. Despite their potential, however, ED data can be under utilised in child injury prevention, often due to poor data availability. In some countries, ED injury surveillance systems have been established to enhance and standardise injury data collection across local, regional or national geographies (eg, Canadian Hospitals Injury Reporting and Prevention Program4; Trauma and Injury Intelligence Group, England5; National Electronic Injury Surveillance System, USA6). While such systems have been successfully used for research and evaluation purposes, they can also struggle with sustainability. In England, injury is a leading cause of child death and hospitalisation.7 Through the Hospital Episodes Statistics (HES) service, England has a comprehensive system of recording episodes of inpatient care to National Health Service hospitals, including private patients. Since 2007, this system has also recorded a basic dataset of attendances at ED services, including major EDs (consultant led, open 24 h a day and 365 days a year with full resuscitation facilities), speciality EDs, walk-in centres and minor injury units. Although currently experimental, the HES Accident & Emergency (A&E) dataset provides data on 74% of all ED attendances in England, including 94% of attendances to major EDs.8 Here we use these data to explore the epidemiology of ED attendance for key injury types in 0–14 year olds and thus examine their utility in informing the targeting of prevention.

METHODS INTRODUCTION

To cite: Hughes K, McHale P, Wyke S, et al. Inj Prev Published Online First: [please include Day Month Year] doi:10.1136/ injuryprev-2013-040816

Preventing childhood injury is a global public health priority.1 Worldwide, over 600 000 children aged 0–14 years lose their lives through injury annually,2 and millions more suffer non-fatal injuries that can have long lasting impacts on physical, mental and social well being.1 The World Health Assembly resolution on child injury prevention, adopted in 2011, urged member states to prioritise child injury prevention and implement multisectoral action.3 A fundamental requirement of the resolution is the availability and use of data systems capable of quantifying the burden and epidemiological profile of child injuries. Such data are critical in understanding the types of injuries suffered by children, at risk groups, resources required for prevention, and where and when such resources should be targeted.

Hughes K, et al. Inj Article Prev 2013;0:1–7. doi:10.1136/injuryprev-2013-040816 Copyright author (or their employer) 2013.

We extracted data on all ED attendances by 0–14 year olds from the HES A&E system for the year 1 April 2010 to 31 March 2011 (n=3 274 825). The dataset records a range of variables on patient demographics, reason for attendance, treatment type and times, and disposal mechanism. With a focus on prevention rather than service delivery, we extracted the following variables: date and time of attendance; patient age, sex and area of residence; attendance type (first or follow-up); primary diagnosis; patient group, which classifies attendances into nine categories of road traffic injury (RTI), assault, deliberate self harm (DSH), sports injury, firework injury, brought in dead, other unintentional injury, other than above and not known; and disposal method (eg, admitted, GP referral).8 We excluded cases that were: follow-up attendances (relating to a

Produced by BMJ Publishing Group Ltd under licence.

1

Downloaded from injuryprevention.bmj.com on July 10, 2013 - Published by group.bmj.com

Original article previously treated injury, n=176 160); brought in dead (n=189); missing area of residence (n=7771) or gender (n=12 873); and non-injury attendances (‘other than above’/ ‘not known’ patient groups with non-injury primary diagnoses, n=1 687 545). The HES system automatically maps patients’ residential postcode to a lower super output area (LSOA; standard geographical areas, population mean 15009). Each LSOA is routinely assigned a measure of deprivation using the Index of Multiple Deprivation (IMD; a composite measure incorporating 38 deprivation indicators10). Thus attendances were assigned a national deprivation quintile based on the IMD value of their LSOA.11 The HES A&E dataset is limited in its classification of injury types. Thus we examined primary diagnoses fields to improve injury categorisation where possible. For example, over 12 000 cases with a patient group of other unintentional injury, other than above or not known (henceforth referred to as a combined ‘other’ injury group) had a primary diagnosis of ‘poisoning and overdose’. These were coded to a new ‘poisoning’ category. The small number of attendances with patient group ‘firework injury’ (n=795) were combined with over 21 000 ‘other injuries’ with a primary diagnosis of ‘burn/scald’ to form a single ‘burns’ category. This process resulted in seven attendance groups: RTI, assault, DSH, sports injury, burns, poisoning and other injury. We excluded attendances coded as DSH or sports injuries in 0–4 year olds (accounting for 0.04% and 0.28% of the remaining cases, respectively); thus analyses of these injury types focused on 5–14 year olds. The final sample was 1 385 871. Most attendances (88.7%) were coded as other injuries. In the primary diagnosis field, a third of these cases were coded as having ‘no diagnosis’ or were missing a diagnosis code. The most common diagnosis types were dislocation/fracture (12.2% of other injuries), laceration (10.6%), contusion/abrasion (9.1%), sprain/ligament (8.7%) and soft tissue inflammation (8.4%). While it was not possible to place these attendances into specific injury categories, one study found >60% of such other injury attendances were falls.5 Without further detail here, however, the utility of data on other injuries in prevention is severely limited and hence this category was only included in basic analysis of sociodemographics and discharge mode. Further analysis focused on the six key injury types: RTI (n=21 670), assault (n=9529), DSH (n=3066), sports injury (n=88 250), burns (n=22 222) and poisoning (n=12 446). Data were analysed in Predictive Analytics Software (PASW) V.18 using χ2 and ANOVA. Backward conditional logistic regression was used to calculate adjusted odds of attendance by demography, with non-attendees calculated by detracting attendances from age, sex and deprivation matched population numbers. Generalised linear modelling (GLM) was used to examine independent impacts of calendar effects on attendances.

RESULTS For all injury types except sports injury, attendance increased with deprivation. Thus children from the most deprived residence quintile accounted for 38.7% of assault attendances and those from the least deprived quintile for 9.2% (table 1). Conversely, sports injury attendance decreased with deprivation. Males accounted for more injury attendances than females overall, with gender differences particularly pronounced for assault and sports injury. However, females accounted for more DSH attendances. Around two-thirds of burns and poisoning attendances involved children aged 0–4 years, while 10–14 year 2

olds accounted for >80% of assault and sports injury attendances and >90% of DSH attendances. Relationships between injury attendance and deprivation remained after controlling for age and gender (figures 1, 2). Compared with children living in the most affluent quintile, those from the most deprived quintile had almost threefold odds of RTI attendance (AOR 2.77, p