Community Opioid Dispensing after Injury (CODI): Cohort characteristics and opioid dispensing patterns

BACKGROUND

Despite a focus of opioid-related research internationally, there is limited understanding of long-term opioid use in adults following injury. We analysed data from the 'Community Opioid Dispensing after Injury' data linkage study.

AIMS

This paper aims to describe the baseline characteristics of the injured cohort and report opioid dispensing patterns following injury-related hospitalisations.

METHODS

Retrospective cohort study of adults hospitalised after injury (ICD-10AM: S00-S99, T00-T75) in Queensland, Australia between 1 January 2014 and 31 December 2015, prior to implementation of opioid stewardship programs. Data were person-linked between hospitalisation, community opioid dispensing and mortality collections. Data were extracted for 90-days prior to the index hospital admission, to establish opiate naivety, to 720 days after discharge. Median daily oral morphine equivalents (i.e., dose) were averaged for each 30-day interval. Cumulative duration of dispensing and dose were compared by demographic and clinical characteristics, stratified by drug dependency status.

RESULTS

Of the 129,684 injured adults, 61.3 % had no opioids dispensed in the 2-year follow-up period. Adults having any opioids dispensed in the community (38.7 %) were more likely older, female, to have fracture injuries and injuries with a higher severity, compared to those with no opioids dispensed. Longer durations and higher doses of opioids were seen for those with pre-injury opioid use, more hospital readmissions and repeat surgeries, as well as those who died in the 2-year follow-up period. Median dispensing duration was 24-days with a median daily end dose of 13 oral morphine equivalents. If dispensing occurred prior to the injury, duration increased 10-fold and oral morphine equivalents doubled. Adults with a documented dependency prior to, or after, the injury had significantly longer durations of use and higher doses than the rest of the cohort receiving opioids. Approximately 7 % of the total cohort continued to be dispensed opioids at 2-years post injury.

CONCLUSION

This is a novel population-level profile of opioid dispensing patterns following injury-related hospitalisation, described for the time period prior to the implementation of opioid stewardship programs and regulatory changes in Queensland. Detailed understanding of this pre-implementation period is critical for evaluating the impact of these changes moving forward.

A population-based epidemiological and health economic analysis of fracture-related infection

Aims

The aim of this study was to perform the first population-based description of the epidemiological and health economic burden of fracture-related infection (FRI).

Methods

This is a retrospective cohort study of operatively managed orthopaedic trauma patients from 1 January 2007 to 31 December 2016, performed in Queensland, Australia. Record linkage was used to develop a person-centric, population-based dataset incorporating routinely collected administrative, clinical, and health economic information. The FRI group consisted of patients with International Classification of Disease 10th Revision diagnosis codes for deep infection associated with an implanted device within two years following surgery, while all others were deemed not infected. Demographic and clinical variables, as well as healthcare utilization costs, were compared.

Results

There were 111,402 patients operatively managed for orthopaedic trauma, with 2,775 of these (2.5%) complicated by FRI. The development of FRI had a statistically significant association with older age, male sex, residing in rural/remote areas, Aboriginal or Torres Strait Islander background, lower socioeconomic status, road traffic accident, work-related injuries, open fractures, anatomical region (lower limb, spine, pelvis), high injury severity, requiring soft-tissue coverage, and medical comorbidities (univariate analysis). Patients with FRI had an eight-times longer median inpatient length of stay (24 days vs 3 days), and a 2.8-times higher mean estimated inpatient hospitalization cost (AU$56,565 vs AU$19,773) compared with uninfected patients. The total estimated inpatient cost of the FRI cohort to the healthcare system was AU$156.9 million over the ten-year period.

Conclusion

The results of this study advocate for improvements in trauma care and infection management, address social determinants of health, and highlight the upside potential to improve prevention and treatment strategies.

External causes of emergency department presentations: A missing piece to understanding unintentional childhood injury in Australia

OBJECTIVE

To identify external causes of unintentional childhood injury presenting to Australian EDs.

METHODS

Six major paediatric hospitals in four Australian states supplied de-identified ED data for 2011-2017 on age, sex, attendance time/date, presenting problem, injury diagnosis, triage category and mode of separation. Three hospitals supplied data on external cause and intent of injury. A machine classifier tool was used to supplement the missing external cause coding in the remaining hospitals to enable the compilation of a standardised dataset for childhood injury causes analysis.

RESULTS

A total of 486 762 ED presentations for unintentional injury in children aged 0-14 years were analysed. The leading specified cause of ED presentations was low fall (35.0%) followed by struck/collision with an object (13.8%) with little sex difference observed. Males aged 10-14 years had higher rates of motorcycle, pedal cycle and fire/flame-related injury and lower rates of horse-related injury and drug/medicinal substance poisoning compared with females. The leading specified external cause resulting in hospitalisation was low fall (32.2%) followed by struck/collision with an object (11.1%). The injuries with the highest proportion of children being hospitalised were drownings (64.4%), pedestrian (53.4%), motorcycle (52.7%) and horse-related injuries (50.0%).

CONCLUSIONS

This is the first large-scale study since the 1980s to explore external causes of unintentional childhood injury presenting to Australian paediatric EDs. It demonstrates a hybrid human-machine learning approach to create a standardised database to overcome data deficiencies. The results supplement existing knowledge of hospitalised paediatric injury to better understand the causes of childhood injury by age and sex, which require health service utilisation.

Navigating child product safety: Perspectives from experts on international challenges and priorities in regulation and research

OBJECTIVE

To elicit and summarise collective expert opinion on contemporary child product safety risks, challenges and priorities.

METHODS

An online survey targeted international experts from a cross-section of product safety fields.

RESULTS

Fifty-five experts participated, representing 1,137 years of product safety experience, from a broad range of fields including industry risk management, product assessment and testing, policy and regulation, research, paediatric medicine, advocacy and product liability. Participants identified the leading product safety hazards across all age brackets as falls, drowning and chemical hazards, with variance in specific age brackets, particularly the threat to breathing hazards for infants. The leading products of concern to experts were electrical connection/distribution products, primarily button batteries and lithium-ion batteries, infant furnishing products and household furniture. Product safety priorities and challenges were identified under five themes: regulatory, surveillance, industry, consumer and product-specific.

CONCLUSIONS

The gains in knowledge, insight and understanding from experts on contemporary child product safety risks and issues should inform policy and future research.

IMPLICATIONS FOR PUBLIC HEALTH

There are significant consequences of unsafe consumer products on population health, and the results are timely as we face new product safety issues emerging from e-commerce, the digital transition and innovative product technologies.

Factors associated with higher alcohol concentrations in emergency department presentations: PACE study

INTRODUCTION

The health impact from alcohol is of recognised concern, from acute intoxication as well as increased risk of chronic health issues over time. Identifying factors associated with higher alcohol consumption when presenting to the emergency department (ED) will inform public health policy and enable more targeted health care and appropriate referrals.

METHODS

Secondary testing of blood samples collected during routine clinical care of 1160 ED patients presenting to the Royal Brisbane and Women's Hospital in Queensland, Australia, for 10 days between 22 January and 1 February 2021. Alcohol was measured by blood ethanol (intake in recent hours) and phosphatidylethanol (PEth; intake over 2-4 weeks). Zero-inflated negative binomial regression was used to identify demographic and clinical factors associated with higher alcohol concentrations.

RESULTS

Males were found to have 83% higher blood ethanol and 32% higher PEth concentrations than females (adjusted rate ratio [ARR] 1.83, 95% confidence interval [CI] 1.37-2.45 and ARR 1.32, 95% CI 1.04-1.68, respectively). Blood ethanol concentrations were 3.4 times higher for those 18-44 years, compared to those aged 65+ (ARR 3.40, 95% CI 2.40-4.82) whereas PEth concentrations were found to be the highest in those aged 45-64 years, being 70% higher than those aged 65+ (ARR 1.70, 95% CI 1.19-2.44). Patients brought in involuntarily had eight-times higher blood ethanol concentrations than those who self-attended.

DISCUSSION AND CONCLUSIONS

This study used two alcohol markers to identify factors associated with higher alcohol concentrations in emergency presentations. The findings demonstrate how these biomarkers can provide informative data for public health responses and monitoring of alcohol use trends.

Identifying foci for safety messages targeting child injury from driving quad bikes: A critical beliefs analysis of parental beliefs in Australia

INTRODUCTION

In Australia, between 2017-2021, 16% of quad bike fatalities involved children. Trauma statistics highlight that public awareness of the risks associated with children driving quads is required. Consistent with the Step approach to Message Design and Testing (SatMDT) and, in particular, Steps 1 and 2, this study sought to identify critical beliefs influencing parental intentions to allow their children to drive a quad bike and develop message content. The critical beliefs analysis was based on eliciting the Theory of Planned Behavior's (TPB) behavioral, normative, and control beliefs.

METHODS

An online survey was distributed via parenting blogs, social media posts, and snowballing of the researchers' network list. Parents who participated (N = 71; 53F, 18 M), were aged between 25-57 years (M = 40.96, SD = 6.98), had at least one child aged between 3 to 16 years, and currently resided in Australia.

RESULTS

The critical beliefs analysis identified four critical beliefs that significantly predicted parental intentions to allow their child to drive a quad bike. These beliefs included a behavioral belief (the perceived advantage that allowing their child to drive a quad bike would enable tasks to be completed), two normative beliefs (the perception that one's parents and partner would likely approve of allowing their child to drive a quad bike), and one control belief (a perceived barrier to allowing one's child to drive a quad that was associated with being aware of an increasing cultural concern around the safety of quad bikes).

CONCLUSIONS

Findings contribute to insights regarding parental beliefs underpinning their intention to allow their child to drive a quad bike, an area previously lacking in research evidence.

PRACTICAL APPLICATIONS

With child-use posing a high-risk activity for children, this study provides an important contribution that may help to inform future safety messaging targeting children's use of quad bikes.

The epidemiology of sports and leisure-related injury hospitalisations in Queensland: A five-year review

OBJECTIVE

To describe the epidemiology of sports and leisure-related injury hospitalisations in Queensland DESIGN, SETTING, AND PATIENTS: Retrospective analysis of routinely collected hospital admissions data from all Queensland hospitals (public and private) between 2012 and 2016 for injury-related admissions where the activity engaged in when injured was coded as sports or leisure activity.

MAIN OUTCOME MEASURES

Number of hospitalisations; rate of hospitalisation per 100,000 population and demographic, injury, treatment, and outcome details of hospitalised injury patients.

RESULTS

Between 01 January 2012 and 31 December 2016, 76,982 people were hospitalised for a sports or leisure-related injury in Queensland. More people were hospitalised in public hospitals than private. Rates were highest for those under 14 years (601.5/100,000 population) and were higher in males (130.6/100,000 population) than females (28.9/100,000 population). A total of 18,734 injuries (24.3%; 79.5/100,000 population) were sustained while playing team ball sports, with rugby codes (rugby union, rugby league and rugby unspecified) representing the single largest source of injuries with 6,592. The extremities were the most likely body location of injury (46,644; 198/100,000 population), and the most common injury type was a fracture (35,018; 148.6/100,000 population).

CONCLUSIONS

The findings highlight the significant burden of sport and leisure-related injury hospitalisations in Queensland. This information is important for injury prevention and trauma system planning.

Electric personal MObility DEvices Surveillance (E-MODES) study: Injury presentations to emergency departments in Brisbane, Queensland

OBJECTIVE

The rapid increase in e-mobility globally, including in Australia, has seen a concurrent increase in e-mobility-related injuries. Monitoring and understanding the patterns of injuries is essential to preserving community safety and making responsive and effective policy decisions regarding their safe use.

METHODS

This study reports on the first phase of the E-MODES study, a proactive injury surveillance initiative to examine the incidence and nature of injuries, their circumstances, contributing factors, and treatment, presenting to three hospital emergency departments (EDs) in Brisbane, the first Australian city to trial shared e-scooters.

RESULTS

During the 31-month study period, there were 1048 ED presentations related to e-mobility, the vast majority (90.8%) involving e-scooters, with males accounting for 64.3% of cases, the most common age group being 25-34 years, and weekends being the most common period of the week for presentations. The most common injury was fractures (37%), and the upper extremities and head/face were the most commonly injured body regions.

CONCLUSION

Contributing risk factors of alcohol use, not wearing a helmet, and speeding, were prevalent, though poorly recorded and only alcohol use proportions varied by age and gender, with males being more likely than females to have alcohol use reported. Recommendations to support e-mobility-related injury surveillance and safety outcomes include improved data standardisation and sharing.

Development of a standardized minimum dataset for including low-severity trauma patients in trauma registry collections in Australia and Aotearoa New Zealand

BACKGROUND

Trauma continues to place a burden on individuals, communities and health care systems around the world. To help reduce this burden and improve care, trauma registries in Australia and Aotearoa New Zealand collect standardized data on patients admitted with Injury Severity Scores greater than 12. There is currently no agreed minimum data set for trauma patients with Injury Severity Score less than 13, representing an opportunity to provide more data for quality improvement and injury prevention.

METHODS

A binational, expert, advisory group assessed the value of potential fields for a minimum dataset for low severity trauma. Existing trauma registries in Australia and Aotearoa New Zealand were assessed to ensure compatibility.

RESULTS

Thirty-five data fields met criteria for inclusion in the low-severity minimum dataset. The fields comprised a subset of the Australia New Zealand Major Trauma Registry and were included in existing low-severity registries.

CONCLUSION

A minimum data set for low severity has been defined for use in Australia and Aotearoa New Zealand. In addition to high severity trauma data this will provide a standard for data collection that will contribute to quality improvement and injury prevention.

Bandaids are not the fix: Examining the patterns of injury-related emergency department presentations in Australian children

OBJECTIVE

To provide an epidemiological understanding of the types of injuries treated in ED in Australian children, describe the impact of these injuries in volume and severity, and assess the patterns by demographic and temporal factors.

METHODS

ED data from six major paediatric hospitals in four Australian states over the period 2011-2017 were analysed to identify childhood injury patterns by nature of injury and body region, as well as sex, age group and temporal factors.

RESULTS

A total of 486 762 ED presentations for injury in children aged 0-14 years were analysed. The most common injuries for all age groups were fractures of the upper extremities. Leading injury diagnosis groups varied by age groups and sex. Overall, children aged 1-2 years had the highest number of ED presentations for injury, and from birth more males than females presented to ED with injuries with the highest absolute sex difference observed for 10- to 14-year-olds. Seventeen percent of children who presented to ED were admitted to hospital with the leading type of hospitalised injury being fractures. Little monthly variation in ED presentations was observed, except for higher presentations for drowning in summer months, and for most injury types, ED presentations were higher during weekends and daytime.

CONCLUSIONS

This is the first large-scale quantification of paediatric injury-related ED presentation patterns in Australia since the conclusion of the National Injury Surveillance and Prevention Program about 30 years ago. It provides valuable information to inform paediatric ED resourcing decisions as well as important evidence for injury prevention practitioners.

Understanding compensable and non-compensable patient profiles, pathways and physical outcomes for transport and work-related injuries in Queensland, Australia through data linkage

INTRODUCTION

In many jurisdictions, people experiencing an injury often pursue compensation to support their treatment and recovery expenses. Healthcare costs form a significant portion of payments made by compensation schemes. Compensation scheme regulators need accurate and comprehensive data on injury severity, treatment pathways and outcomes to enable scheme modelling, monitoring and forecasting. Regulators routinely rely on data provided by insurers which have limited healthcare information. Health data provide richer information and linking health data with compensation data enables the comparison of profiles, patterns, trends and outcomes of injured patients who claim and injured parties who are eligible but do not claim.

METHODS AND ANALYSIS

This is a retrospective population-level epidemiological data linkage study of people who have sought ambulatory, emergency or hospital treatment and/or made a compensation claim in Queensland after suffering a transport or work-related injury, over the period 1 January 2011 to 31 December 2021. It will use person-linked data from nine statewide data sources: (1) Queensland Ambulance Service, (2) Emergency Department, (3) Queensland Hospital Admitted Patients, (4) Retrieval Services, (5) Hospital Costs, (6) Workers' Compensation, (7) Compulsory Third Party Compensation, (8) National Injury Insurance Scheme and (9) Queensland Deaths Registry. Descriptive, parametric and non-parametric statistical methods and geospatial analysis techniques will be used to answer the core research questions regarding the patient's health service use profile, costs, treatment pathways and outcomes within 2 years postincident as well as to examine the concordance and accuracy of information across health and compensation databases.

ETHICS AND DISSEMINATION

Ethics approval was obtained from the Royal Brisbane and Women's Hospital Human Research Ethics Committee, and governance approval was obtained via the Public Health Act 2005, Queensland. The findings of this study will be used to inform key stakeholders across the clinical, research and compensation regulation area, and results will be disseminated through peer-reviewed journals, conference presentations and reports/seminars with key stakeholders.

Prevalence of alcohol consumption in emergency presentations: Novel approach using two biomarkers, ethanol and phosphatidylethanol

INTRODUCTION

The aim was to determine the prevalence of alcohol-related presentations to an emergency department (ED) in a major Australian hospital, through a novel surveillance approach using two biomarkers, blood ethanol and phosphatidylethanol (PEth).

METHODS

Observational study using secondary testing of blood samples collected during routine clinical care of ED patients presenting to the Royal Brisbane and Women's Hospital in Queensland, Australia, between 22 January and 2 February 2021. Data were collected from 1160 patients during the 10-day study period. The main outcomes were the prevalence of acute alcohol intake, as determined by blood ethanol, and recent use over 2-4 weeks, as determined by PEth concentrations, for all ED presentations and different diagnostic groups.

RESULTS

The overall prevalence for blood ethanol was 9.3% (95% confidence interval [CI] 7.8%, 11.1%), 5.3% for general medical presentations, increasing four-fold to 22.2% for injury presentations. The overall prevalence of PEth positive samples was 32.5% (95% CI 29.9%, 35.3%) and 41.4% for injury presentations. There were 263 (25.3%) cases that tested negative for acute blood ethanol but positive for PEth concentrations indicative of significant to heavy medium-term alcohol consumption.

DISCUSSION AND CONCLUSIONS

This novel surveillance approach demonstrates that using blood ethanol tests in isolation significantly underestimates the prevalence of medium-term alcohol consumption in ED presentations. Prevalence of alcohol use was higher for key diagnostic groups such as injury presentations. Performing periodic measurement of both acute and medium-term alcohol consumption accurately and objectively in ED presentations, would be valuable for informing targeted public health prevention and control strategies.

Application of a Machine Learning-based Decision Support Tool to Improve an Injury Surveillance System Workflow

Abstract Background Emergency department (ED)-based injury surveillance systems across many countries face resourcing challenges related to manual validation and coding of data. Objective This paper describes the evaluation of a machine learning-based Decision Support Tool (DST) to assist injury surveillance departments in the validation, coding and use of their data, comparing outcomes in coding time and accuracy pre- and post-implementation. Methods Manually coded injury surveillance data has been used to develop, train and iteratively refine a machine learning-based classifier to enable semi-automated coding of injury narrative data. This paper describes a trial implementation of the machine learning-based DST in the Queensland Injury Surveillance Unit (QISU) workflow using a major pediatric hospital's emergency department data comparing outcomes in coding time and accuracy pre- and post-implementation. Results The study found a 10% reduction in manual coding time after the DST was introduced. The Kappa statistics analysis in both DST-assisted and unassisted data shows increases in accuracy across three data fields; injury intent (85.4% unassisted vs. 94.5% assisted), external cause (88.8% unassisted vs. 91.8% assisted) and injury factor (89.3% unassisted vs. 92.9% assisted). The classifier was also used to produce a timely report monitoring injury patterns during the COVID-19 pandemic. Hence, it has the potential for near real-time surveillance of emerging hazards to inform public health responses. Conclusions The integration of the DST into the injury surveillance workflow shows benefits as it facilitates timely reporting and acts as a DST in the manual coding process.

DINED (Delivery-related INjuries in the Emergency Department) part 2: A chart review of risk factors and injuries affecting food delivery riders

OBJECTIVE

This study aimed to examine patterns of injury, the impact of these injuries on patients and identification of potentially modifiable contributing factors through industry regulation reforms and education.

METHODS

Food delivery rider (FDR)-related presentations to the Royal Brisbane and Women's Hospital Emergency and Trauma Centre for a year from September 2020 were identified. Data collected included patient and incident demographics: time, type and location of injuries, investigations and care required, length of stay, admission requirements and follow up.

RESULTS

The cohort included 81.8% male with a mean age of 25.2 years. Most injuries occurred on the road from a collision with a vehicle. The most common injury was fractures. Incidence increased on weekends and during the evening. More than half the cohort were admitted to hospital. Only 22.7% of patients were eligible for workers compensation and less than half were covered by Medicare. The majority (72.7%) of cases involved non-resident riders from other countries.

CONCLUSION

The majority of FDRs presenting with injuries are not Australian citizens and less than half were Medicare eligible potentially contributing to inadequate access to care especially fracture follow up. There were spikes in injuries occurring at night, weekends and during periods of pandemic associated lockdowns demonstrating an increased usage of delivery services during these times. Results highlight injury patterns experienced by delivery riders and potentially modifiable risk factors for this rapidly growing area of employment within the gig economy.

Applying a public health approach to identify priorities for regulating child product safety

OBJECTIVE

To identify leading injury risk factors and jurisdictional differences in Australian and US child-related product safety regulatory responses to inform the development of Australian policy and reform priorities.

METHODS

The study established and evaluated a knowledge base of child-related product safety regulatory responses (recalls, bans, standards and warnings) made in Australia and the US over the period 2011-17 to identify risk factors and potential regulatory gaps.

RESULTS

The research identified 1,540 Australian and US child-related product safety regulatory responses with the most common response type being product safety recall, and the leading product hazards in responses being choking, fire, fall, strangulation and chemical hazards. Jurisdictional differences identified potential regulatory gaps in Australia related to chemical hazards and high-risk durable infant and toddler products, and some data deficiencies in Australian responses.

CONCLUSIONS

Priorities include the need to improve the prevention orientation of the Australian product safety framework, to create an intelligence platform to assess injury risks more precisely and to address regulatory gaps related to the use of toxic chemicals in children's products and high-risk durable infant and toddler products. Implications for public health: The study demonstrates the identification of policy and reform priorities for child product safety using a public health lens.

Community Opioid Dispensing after Injury (CODI): Protocol for a retrospective population-based cohort study (Preprint)

Background

There is an urgent need to reduce preventable deaths and hospitalizations from prescription opioid harms and minimize the negative effect opioid misuse can have on injured individuals, families, and the wider community. Data linkage between administrative hospitalization records for injured patients and community opioid dispensing can improve our understanding of the health and surgical trajectories of injured persons and generate insights into corresponding opioid dispensing patterns.

Objective

The Community Opioid Dispensing after Injury (CODI) study aims to link inpatient hospitalization data with opioid dispensing data to examine the distribution and predictive factors associated with high or prolonged community opioid dispensing among adults, for 2 years following an injury-related hospital admission.

Methods

This is a retrospective population-based cohort study of adults aged 18 years or older hospitalized with an injury in Queensland, Australia. The study involves the linkage of statewide hospital admissions, opioid prescription dispensing, and mortality data collections. All adults hospitalized for an injury between January 1, 2014, and December 31, 2015, will be included in the cohort. Demographics and injury factors are recorded at the time of the injury admission. Opioid dispensing data will be linked and extracted for 3 months prior to the injury admission date to 2 years after the injury separation date (last date December 31, 2017). Deaths data will be extracted for the 2-year follow-up period. The primary outcome measure will be opioid dispensing (frequency and quantity) in the 2 years following the injury admission. Patterns and factors associated with community opioid dispensing will be examined for different injury types, mechanisms, and population subgroups. Appropriate descriptive statistics will be used to describe the cohort. Regression models will be used to examine factors predictive of levels and duration of opioid use. Nonparametric methods will be applied when the data are not normally distributed.

Results

The project is funded by the Royal Brisbane and Women’s Hospital Foundation. As of November 2021, all ethics and data custodian approvals have been granted. Data extraction and linkage has been completed. Data management and analysis is underway with results relating to an analysis for blunt chest trauma patients expected to be published in 2022.

Conclusions

Little is currently known of the true prevalence or patterns of opioid dispensing following injury across Queensland. This study will provide new insights about factors associated with high and long-term opioid dispensing at a population level. This information is essential to inform targeted public policy and interventions to reduce the risk of prolonged opioid use and dependence for those injured. The novel work undertaken for this project will be vital to planning, delivering, monitoring, and evaluating health care services for those injured. The findings of this study will be used to inform key stakeholders as well as clinicians and pain management services.

International Registered Report Identifier (IRRID)

RR1-10.2196/36357

Prevalence of Alcohol Consumption in Emergency department presentations (PACE) in Queensland, Australia, using alcohol biomarkers ethanol and phosphatidylethanol: an observational study protocol

INTRODUCTION

Alcohol use in patients presenting to the emergency department (ED) is a significant problem in many countries. There is a need for valid and reliable surveillance of the prevalence of alcohol use in patients presenting to the ED, to provide a more complete picture of the risk factors and inform targeted public health interventions. This PACE study will use two biomarkers, blood ethanol and phosphatidylethanol (PEth), to determine the patterns, presence and level of alcohol use in patients presenting to an Australian ED.

METHODS AND ANALYSIS

This is an observational prevalence study involving the secondary use of routinely collected blood samples from patients presenting to the Royal Brisbane and Women's Hospital (RBWH) Emergency and Trauma Centre (ETC). Samples will be tested for acute and medium-term alcohol intake using the two biomarkers blood ethanol and PEth respectively, over one collection period of 10-12 days. Descriptive statistics such as frequencies, percentages, means, SD, medians and IQRs, will be used to describe the prevalence, pattern and distribution of acute and medium-term alcohol intake in the study sample. The correlation between acute and medium-term alcohol intake levels will also be examined.

ETHICS AND DISSEMINATION

This study has been approved by the RBWH Human Research Ethics Committee (reference, LNR/2019/QRBW/56859). Findings will be disseminated to key stakeholders such as RBWH ETC, Australasian College for Emergency Medicine, Royal Australasian College of Surgeons, Statewide Clinical Networks, and used to inform clinicians and hospital services. Findings will be submitted for publication in peer-reviewed journals and presentation at appropriate conferences.

Comparing the accuracy of ICD-based severity estimates to trauma registry-based injury severity estimates for predicting mortality outcomes

INTRODUCTION

Trauma registries have been used internationally for several decades to measure the quality of trauma care between hospitals. Given the significant costs involved in establishing and maintaining trauma registries, and increasing availability of routinely collected, linked health data describing a patient's journey (and inherent cost savings in data re-use), there is significant interest in development of integrated, comprehensive trauma data repositories. However, approaches to estimating injury severity using routinely collected data would need to be developed if routinely collected hospital data were to be used as an alternative/supplement to registries.

OBJECTIVES

This study aimed to compare the accuracy of registry-based injury severity estimates with ICD-based injury severity estimates in predicting mortality outcomes in a cohort of minor and major trauma patients in Queensland, using retrospectively linked trauma registry and hospital admissions data.

METHODS

Queensland Trauma Registry (QTR) data with an admission date between 1 January 2005 and 31 December 2011 was linked with all acute care patients included in the Queensland Hospital Admitted Patient Data Collection (QHAPDC) with a Principal Diagnosis coded with an ICD-10-AM code within Chapter 19 (S00-T98). Abbreviated Injury Scale coding was undertaken manually by QTR trauma data nurses for the registry data. ICD-based injury severity scores (ICISS) were calculated automatically using all injury-related diagnoses captured in the QHAPDC data using the ICISS multiplicative and worst injury method.

RESULTS

There were 92,140 QTR patients admitted between January 2005 and December 2011 with a valid ISS with a matching QHAPDC record (98.4% survived, 1.6% died). ICISS (multiplicative and worst injury approach) showed marginally better predictive accuracy than ISS when predicting mortality across minor and major injury and ICISS showed marginally better predictive accuracy to ISS when restricted to major trauma/high threat to life cases. Both ICISS and ISS restricted to major trauma/high threat to life showed poorer accuracy compared to the predictive performance when both minor and major cases were included.

CONCLUSION

ICD-based predictions were as accurate as ISS-based predictions for this cohort and this study provides evidence to support the potential for using routinely coded hospital data for risk adjustment within State-based trauma data repositories.

Trauma by-pass guideline: A data-driven conformance analysis for road trauma cases in Queensland

OBJECTIVE

Study objectives were to (i) develop and test a whole-of-system method for identifying patients who meet a major trauma by-pass guideline definition; (ii) apply this method to assess conformance to the current 2006 guideline for a road trauma cohort; and (iii) leverage relevant findings to propose improvements to the guideline.

METHODS

Retrospective analysis of existing, routinely collected data relating to Queensland road trauma patients July 2015 to June 2017. Data from ambulance, aero-medical retrievals, ED, hospital and death registers were linked and used for analysis. Processes of care measured included: frequency of pre-hospital triage criteria, distribution of destination (trauma service level), compliance with guideline (recommended vs actual destination), trauma service level by threat to life (injury severity) (all modes of transport and aero-medical in particular), proportion of patients requiring only ED, transport pathway (direct vs inter-hospital transfer).

RESULTS

3847 cases were identified from data as meeting criteria for major trauma by-pass. The top five most frequently used criteria for qualifying patients as meeting the major trauma by-pass guideline were pulse rate, vehicle rollover, possible spinal cord injury, respiration rate and entrapment. The study demonstrates a 65% conformance to the clinical guideline. Overtriaged patients (transported to higher trauma service than recommended) generally reveal International Classification of Disease Injury Severity Score representing a high threat to life.

CONCLUSION

Overall, the present study found good conformance, with overtriage rate as expected by clinicians. It is recommended to include data values to capture paramedics assessment of trauma level to enable more accurate assessment of conformance to guideline and future revision of the thresholds.

Three Es of linked injury data: Episodes, Encounters and Events

BACKGROUND

Treatment and recovery times following injury can be lengthy, comprising multiple interactions with the hospital system for initial acute care, subsequent rehabilitation and possible re-presentation due to complications.

AIMS

This article aims to promote the use of consistent terminology in injury data linkage studies, suggest important factors to consider when managing linked injury data, and encourage thorough documentation and a robust discourse around different approaches to data management to ensure reproducibility, consistency and comparability of analyses arising from linked injury data.

APPROACH

This paper is presented in sections describing: (1) considerations for identifying injury cohorts, (2) considerations for grouping Episodes into Encounters and (3) considerations for grouping Encounters into Events. Summary tools are provided to aid researchers in the management of linked injury data.

DISCUSSION

Careful consideration of decisions made when identifying injury cohorts and grouping data into units of analysis (Episodes/Encounters/Events) is essential when using linked injury data. Choices made have the potential to significantly impact the epidemiological and clinical findings derived from linked injury data studies, which ultimately affect the quality of injury prevention initiatives and injury management policy and practice. It is intended that this paper will act as a call to action for injury linkage methodologists, and those using linked data, to critique approaches, share tools and engage in a robust discourse to further advance the use of linked injury data, and ultimately enhance the value of linked injury data for clinicians and health and social policymakers.

Identifying the priority challenges in trauma care delivery for Australian and New Zealand trauma clinicians

INTRODUCTION

Injury is a leading cause of death and disability world-wide. Little is known about the day-to-day challenges the trauma clinicians face in their practice that they feel could be improved through an increased evidence base. This study explored and ranked the trauma clinical practice research priorities of trauma care professionals across Australia and New Zealand.

METHODS

A modified-Delphi study was conducted between September 2019 and January 2020. The study employed two rounds of online survey of trauma professionals from relevant Australia and New Zealand professional organisations using snowballing method. Participants were asked to rank the importance of 29 recommendations, each corresponding to a key challenge in trauma care delivery. Decisions on the priorities of the challenges were determined by a consensus of >70% of respondents ranking the challenge as important or very important.

RESULTS

One hundred and fifty-five participants completed Round One, and 106 participants completed Round Two. A total of 15 recommendations reached >70% in Round One. Nine recommendations also reached >70% consensus in Round Two. Recommendations ranked highest were 'Caring for elderly trauma patients', 'Identifying and validating key performance indicators for trauma system benchmarking and improvement', and 'Management of traumatic brain injury'.

CONCLUSION

This study identified the priority areas for trauma research as determined by clinician ranking of the most important for informing and improving their practice. Addressing these areas generates potential to improve the quality and safety of trauma care in Australian and New Zealand.

Artificial intelligence or manufactured stupidity? The need for injury informaticians in the big data era

The volume, velocity and variety of data collected about individuals have increased exponentially over the last decade, presenting new injury surveillance opportunities to identify risk factors, monitor trends, and evaluate the efficacy of interventions. But does the hype around big data and artificial intelligence (AI) apply to the injury prevention space, and how veracious is surveillance in this era? This commentary discusses the digital transformation of health as applied to injury prevention, but cautions on the challenges of maintaining data quality in integrated systems and discusses the need for an injury informatics strategy moving forward.

Long term impact of ladder-related injuries as measured by the AQoL instrument

Introduction

Ladder-related falls are a common cause of patients presenting to emergency departments (ED) with serious injury. The impacts of ladder-related injuries were assessed at six-months post-injury using the quality of life, AQoL 4D Basic (AQoL) instrument.

Materials and methods

This was a prospective observational study, conducted and reported according to the STROBE statement. All adult patients with ladder-related injuries who presented to two EDs in southeast Queensland, Australia between October 2015 and October 2016 were approached. Initial participant interviews took place at the time of ED presentation or shortly thereafter, with follow-up telephone interview at six-months.

Results

There were 177 enrolments, 43 (24%) were lost to follow up. There were statistically significant changes post-injury for three of the four AQoL dimensions: independence, social relationships and psychological wellbeing, as well as the global AQoL. Twenty-four (18%) participants reported a clinically significant deterioration in independence, 26 (20%) participants reported a clinically significant deterioration in their social relationships, and 34 participants (40%) reporting a clinically significant deterioration in their psychological wellbeing. Nine of the twelve individual items (in AQoL dimension) deteriorated after injury, there was no change in two items (vision and hearing) and an improvement reported in one (communication). The largest changes (> 25% of participants) were reported with sleeping, anxiety worry and depression, and pain. Across the global AQoL dimension, 65 (49%) participants reported a clinically significant deterioration. The severity of injury as measured by the ISS was an independent predictor of the change in AQoL scores (p<0.001).

Conclusions

Injuries related to falls from ladders continue to have a profound impact on patients at six-months post-injury as measured using the AQoL instrument. This adds to previous research which has demonstrated considerable morbidity and mortality at the time of injury.

Prevention

Older males using ladders at home are at high risk for serious long-term injury. Injury prevention strategies and the safety instructions packaged with the ladder need to be targeted to this at-risk community group. There may also be a role for regulatory bodies to mandate a stabilising device to be included with the ladder at the time of purchase.

A Comparative Process Mining Analysis of Road Trauma Patient Pathways

In this paper we report on key findings and lessons from a process mining case study conducted to analyse transport pathways discovered across the time-critical phase of pre-hospital care for persons involved in road traffic crashes in Queensland (Australia). In this study, a case is defined as being an individual patient’s journey from roadside to definitive care. We describe challenges in constructing an event log from source data provided by emergency services and hospitals, including record linkage (no standard patient identifier), and constructing a unified view of response, retrieval, transport and pre-hospital care from interleaving processes of the individual service providers. We analyse three separate cohorts of patients according to their degree of interaction with Queensland Health’s hospital system (C1: no transport required, C2: transported but no Queensland Health hospital, C3: transported and hospitalisation). Variant analysis and subsequent process modelling show high levels of variance in each cohort resulting from a combination of data collection, data linkage and actual differences in process execution. For Cohort 3, automated process modelling generated ’spaghetti’ models. Expert-guided editing resulted in readable models with acceptable fitness, which were used for process analysis. We also conduct a comparative performance analysis of transport segment based on hospital ‘remoteness’. With regard to the field of process mining, we reach various conclusions including (i) in a complex domain, the current crop of automated process algorithms do not generate readable models, however, (ii) such models provide a starting point for expert-guided editing of models (where the tool allows) which can yield models that have acceptable quality and are readable by domain experts, (iii) process improvement opportunities were largely suggested by domain experts (after reviewing analysis results) rather than being directly derived by process mining tools, meaning that the field needs to become more prescriptive (automated derivation of improvement opportunities).

Challenges to trauma care delivery for Australian and New Zealand trauma clinicians

INTRODUCTION

The availability and implementation of evidence-based care is essential to achieving safe, quality trauma patient outcomes. Little is documented, however, about the challenges trauma clinicians face in their day-to day practice, or their views on the availability of evidence. This paper presents the most significant clinical practice challenges reported by multidisciplinary trauma care professionals in Australia and New Zealand, in particular those that may be resolved with focussed research or enhanced implementation activity.

METHODS

An exploratory survey of trauma professionals from relevant Australia and New Zealand professional organisations was conducted between September 2018 and February 2019 using the Snowballing Method. Participants were recruited via a non-random sampling technique to complete an online survey. Thematic analyses were conducted.

RESULTS

There were nine significant clinical practice challenge themes in trauma care, arising from 287 individual clinical practice challenges reported. The most reported being clinical management (bleeding, spinal, older patients) and operationalisation of the trauma system. There was no consensus as to the availability of evidence to guide each theme.

CONCLUSION

Future research should seek to address the clinical practice challenge of Australian and New Zealand trauma community to enable safe, quality trauma patient outcomes.

Hazardous children's products on the Australian and US market 2011-2017: an empirical analysis of child-related product safety recalls

Objective

While there is evidence that unsafe children’s products are entering the Australian market, with increasing product safety recalls, no research has examined the nature of recalls or their trends over time. This research analyses Australian and US child-related product safety recall data to better understand the frequency and nature of unsafe children’s products, emerging hazard trends and cross-jurisdictional similarities and differences. Results can inform improved childhood injury prevention policy and regulation strategies in Australia.

Method

Empirical analysis of child-related product safety recalls in Australia and the USA over the period 2011–2017.

Results

Cross-jurisdictional comparison revealed similarities in Australia and the USA, with over 80% of recalled products occurring in four industry segments (toys/games, household furniture/furnishings, clothing and sports equipment) and a common leading hazard of choking. Australia and the USA also had a similar number of child-related recalls over the study period (Australia: 652, USA: 668). Disparate trends included a 21% decrease in US child-related recalls over the study period, with most recalled products still complying with mandated safety requirements. In contrast, Australian child-related recalls increased by 88% over the study period, with the majority of recalled products failing to comply with mandated safety requirements. Based on US child-related recall data, the leading cause of injuries was the child falling, the most severe injuries related to furniture/furnishings and the most frequent injuries related to sports equipment.

Conclusion

Analysing recall data provides new insights into hazardous children’s products. Cross-jurisdictional comparison of data on recalls highlights disparities and indicates a need for reforms to improve regulation of children’s products in Australia.

Characteristics of accidental injuries from power tools treated at two emergency departments in Queensland

OBJECTIVE

Injuries are a major burden on the Australian healthcare system. Power tool usage is a common cause of accidental injury. A better understanding of the trends of power tool injuries will inform prevention strategies and potentially mitigate costs.

METHODS

The ED databases from two level 1 hospitals were reviewed for presentations between 2005 and 2015 resulting from accidental injury with power tools. A subgroup of patients presenting to one hospital between 2016 and 2017 were interviewed about the activities and circumstances that led to their injuries, and followed up 3 months later to assess outcomes.

RESULTS

A total of 4057 cases of accidental injury from power tool use were identified. Power saws and grinders contributed to 54% of injuries. Most injuries were located on an upper limb (48%) or the head and neck (30%). Over half (54%) of all head injuries were associated with metal and wood fragments to the eye from grinders, drills and saws. Hospital admission rates were highest for patients aged >60 years. Injuries to females were <5% of all presentations, but 40% of those caused by lawnmowers. Among the 200 patients interviewed, lapses in concentration during use, and modification and inappropriate use of a power tool were the main contributors to injury. Recovery periods >3 months were common.

CONCLUSIONS

Accidental injuries from power tool use have a considerable impact on ED resources and can affect the long-term quality of life of those injured. Effective education about safe usage and protection may prevent many injuries.

Value of emergency department triage data to describe and understand patterns and mechanisms of cycling injuries

OBJECTIVE

To characterise patients presenting to EDs for a bicycle-related injury, identify contributing factors to the injuries and describe the data gaps.

METHODS

A retrospective study of bicycle-related injury presentations over the 5 year period 2010-2014 to two major metropolitan EDs. Data collected from the emergency presentation database consisted of patient demographics, presenting complaint, discharge diagnosis and details about the circumstances and mechanism of the accident.

RESULTS

Bicycle injuries (n = 4144) increased from 20.0 to 25.2 per 1000 injury presentations over 5 years. Patients were 80% male across all age groups and 35% of presentations occurred at the weekend. Fractures accounted for one-third (34%) of all injuries, while injuries to the upper extremities and head resulted in 36% and 19% of cases, respectively. Admission rate was 17%. The number of falls exceeded collisions (1611 vs 937), and in the 13.3% of the fall cases where documentation was present for which how the fall occurred, over half were attributed to bike handling errors. Information related to type of bicycle, accident location, type of activity, protective clothing worn (including helmets) and visibility aids was not recorded for over 95% of the cases.

CONCLUSION

Bicycle injuries carry a considerable burden to the ED and the incidence of presentations appears to be rising. The current triage data, designed to provide a rapid assessment for medical urgency, are limited to describing broad demographics, trends and causes.

Improving autocoding performance of rare categories in injury classification: Is more training data or filtering the solution?

INTRODUCTION

Classical Machine Learning (ML) models have been found to assign the external-cause-of-injury codes (E-codes) based on injury narratives with good overall accuracy but often struggle with rare categories, primarily due to lack of enough training cases and heavily skewed nature of injurdata. In this paper, we have: a) studied the effect of increasing the size of training data on the prediction performance of three classical ML models: Multinomial Naïve Bayes (MNB), Support Vector Machine (SVM) and Logistic Regression (LR), and b) studied the effect of filtering based on prediction strength of LR model when the model is trained on very-small (10,000 cases) and very-large (450,000 cases) training sets.

METHOD

Data from Queensland Injury Surveillance Unit from years 2002-2012, which was categorized into 20 broad E-codes was used for this study. Eleven randomly chosen training sets of size ranging from 10,000 to 450,000 cases were used to train the ML models, and the prediction performance was analyzed on a prediction set of 50,150 cases. Filtering approach was tested on LR models trained on smallest and largest training datasets. Sensitivity was used as the performance measure for individual categories. Weighted average sensitivity (WAvg) and Unweighted average sensitivity (UAvg) were used as the measures of overall performance. Filtering approach was also tested for estimating category counts and was compared with approaches of summing prediction probabilities and counting direct predictions by ML model.

RESULTS

The overall performance of all three ML models improved with increase in the size of training data. The overall sensitivities with maximum training size for LR and SVM models were similar (∼82%), and higher than MNB (76%). For all the ML models, the sensitivities of rare categories improved with increasing training data but they were considerably less than sensitivities of larger categories. With increasing training data size, LR and SVM exhibited diminishing improvement in UAvg whereas the improvement was relatively steady in case of MNB. Filtering based on prediction strength of LR model (and manual review of filtered cases) helped in improving the sensitivities of rare categories. A sizeable portion of cases still needed to be filtered even when the LR model was trained on very large training set. For estimating category counts, filtering approach provided best estimates for most E-codes and summing prediction probabilities approach provided better estimates for rare categories.

CONCLUSIONS

Increasing the size of training data alone cannot solve the problem of poor classification performance on rare categories by ML models. Filtering could be an effective strategy to improve classification performance of rare categories when large training data is not available.

Age-related trends in injury and injury severity presenting to emergency departments in New South Wales Australia: Implications for major injury surveillance and trauma systems

OBJECTIVES

To describe population based trends and clinical characteristics of injury related presentations to Emergency Departments (EDs).

DESIGN AND SETTING

A retrospective, descriptive analysis of de-identified linked ED data across New South Wales, Australia over five calendar years, from 2010 to 2014.

PARTICIPANTS

Patients were included in this analysis if they presented to an Emergency Department and had an injury related diagnosis. Injury severity was categorised into critical (triage category 1-2 and admitted to ICU or operating theatre, or died in ED), serious (admitted as an in-patient, excluding above critical injuries) and minor injuries (discharged from ED).

MAIN OUTCOME MEASURES

The outcomes of interest were rates of injury related presentations to EDs by age groups and injury severity.

RESULTS

A total of 2.09 million injury related ED presentations were analysed. Minor injuries comprised 85.0%, and 14.1% and 1.0% were serious and critical injuries respectively. There was a 15.8% per annum increase in the rate of critical injuries per 1000 population in those 80 years and over, with the most common diagnosis being head injuries. Around 40% of those with critical injuries presented directly to a major trauma centre.

CONCLUSION

Critical injuries in the elderly have risen dramatically in recent years. A minority of critical injuries present directly to major trauma centres. Trauma service provision models need revision to ensure appropriate patient care. Injury surveillance is needed to understand the external causes of injury presenting to hospital.

The Extent of Consumer Product Involvement in Paediatric Injuries

A challenge in utilising health sector injury data for Product Safety purposes is that clinically coded data have limited ability to inform regulators about product involvement in injury events, given data entry is bound by a predefined set of codes. Text narratives collected in emergency departments can potentially address this limitation by providing relevant product information with additional accompanying context. This study aims to identify and quantify consumer product involvement in paediatric injuries recorded in emergency department-based injury surveillance data. A total of 7743 paediatric injuries were randomly selected from Queensland Injury Surveillance Unit database and associated text narratives were manually reviewed to determine product involvement in the injury event. A Product Involvement Factor classification system was used to categorise these injury cases. Overall, 44% of all reviewed cases were associated with consumer products, with proximity factor (25%) being identified as the most common involvement of a product in an injury event. Only 6% were established as being directly due to the product. The study highlights the importance of utilising injury data to inform product safety initiatives where text narratives can be used to identify the type and involvement of products in injury cases.

Harnessing information from injury narratives in the 'big data' era: understanding and applying machine learning for injury surveillance

OBJECTIVE

Vast amounts of injury narratives are collected daily and are available electronically in real time and have great potential for use in injury surveillance and evaluation. Machine learning algorithms have been developed to assist in identifying cases and classifying mechanisms leading to injury in a much timelier manner than is possible when relying on manual coding of narratives. The aim of this paper is to describe the background, growth, value, challenges and future directions of machine learning as applied to injury surveillance.

METHODS

This paper reviews key aspects of machine learning using injury narratives, providing a case study to demonstrate an application to an established human-machine learning approach.

RESULTS

The range of applications and utility of narrative text has increased greatly with advancements in computing techniques over time. Practical and feasible methods exist for semiautomatic classification of injury narratives which are accurate, efficient and meaningful. The human-machine learning approach described in the case study achieved high sensitivity and PPV and reduced the need for human coding to less than a third of cases in one large occupational injury database.

CONCLUSIONS

The last 20 years have seen a dramatic change in the potential for technological advancements in injury surveillance. Machine learning of 'big injury narrative data' opens up many possibilities for expanded sources of data which can provide more comprehensive, ongoing and timely surveillance to inform future injury prevention policy and practice.

Current profile of cycling injuries: A retrospective analysis of a trauma centre level 1 in Queensland

OBJECTIVES

One out of 50 injury-related presentations to an ED is a transport-related cycling injury. Detailed information about the most frequent mechanism of cycling injuries, sustained injuries and patterns are under-reported. The objective of this research was to examine the pattern of injuries sustained by cyclists at a level 1 trauma centre.

METHODS

A retrospective review of data of injured cyclists admitted and treated at a level 1 trauma centre between 2011 and 2012 evaluated demographic data, mechanism of injury, injury pattern, economic costs and outcome.

RESULTS

Data of 261 patients (mean age of 39 years) were reviewed, of which 88% was male patients with an average age of 38 years. Non-collision cycling injuries were reported in 55% of cases followed by collisions with other motor vehicles in 25.6% of cases. The mean injury severity score (ISS) was 9, but an ISS ≥ 12 was documented in 24%. Predominantly upper limb injuries (24.8%) were found, followed by injuries to the head and lower limb (each with 16.8%). Traumatic brain and chest injuries were equally seen in 8%. The overall length of stay was 4 days, and survival rate was 98%.

CONCLUSION

This current data review reveals that non-collision traffic crashes accounted for the majority of injuries in cyclists treated in this facility, and the upper limb has replaced the head as the most injured body part. With a growing number of cyclists, this information contributes to considerations to improve road safety and trauma management.

Infant product-related injuries: comparing specialised injury surveillance and routine emergency department data

OBJECTIVE

To explore the potential for using a basic text search of routine emergency department data to identify product-related injury in infants and to compare the patterns from routine ED data and specialised injury surveillance data.

METHODS

Data was sourced from the Emergency Department Information System (EDIS) and the Queensland Injury Surveillance Unit (QISU) for all injured infants between 2009 and 2011. A basic text search was developed to identify the top five infant products in QISU. Sensitivity, specificity, and positive predictive value were calculated and a refined search was used with EDIS. Results were manually reviewed to assess validity. Descriptive analysis was conducted to examine patterns between datasets.

RESULTS

The basic text search for all products showed high sensitivity and specificity, and most searches showed high positive predictive value. EDIS patterns were similar to QISU patterns with strikingly similar month-of-age injury peaks, admission proportions and types of injuries.

CONCLUSIONS

This study demonstrated a capacity to identify a sample of valid cases of product-related injuries for specified products using simple text searching of routine ED data.

IMPLICATIONS

As the capacity for large datasets grows and the capability to reliably mine text improves, opportunities for expanded sources of injury surveillance data increase. This will ultimately assist stakeholders such as consumer product safety regulators and child safety advocates to appropriately target prevention initiatives.

Estimating under-reporting of road crash injuries to police using multiple linked data collections

The reliance on police data for the counting of road crash injuries can be problematic, as it is well known that not all road crash injuries are reported to police which under-estimates the overall burden of road crash injuries. The aim of this study was to use multiple linked data sources to estimate the extent of under-reporting of road crash injuries to police in the Australian state of Queensland. Data from the Queensland Road Crash Database (QRCD), the Queensland Hospital Admitted Patients Data Collection (QHAPDC), Emergency Department Information System (EDIS), and the Queensland Injury Surveillance Unit (QISU) for the year 2009 were linked. The completeness of road crash cases reported to police was examined via discordance rates between the police data (QRCD) and the hospital data collections. In addition, the potential bias of this discordance (under-reporting) was assessed based on gender, age, road user group, and regional location. Results showed that the level of under-reporting varied depending on the data set with which the police data was compared. When all hospital data collections are examined together the estimated population of road crash injuries was approximately 28,000, with around two-thirds not linking to any record in the police data. The results also showed that the under-reporting was more likely for motorcyclists, cyclists, males, young people, and injuries occurring in Remote and Inner Regional areas. These results have important implications for road safety research and policy in terms of: prioritising funding and resources; targeting road safety interventions into areas of higher risk; and estimating the burden of road crash injuries.

Machine learning approaches to analysing textual injury surveillance data: a systematic review

OBJECTIVE

To synthesise recent research on the use of machine learning approaches to mining textual injury surveillance data.

DESIGN

Systematic review.

DATA SOURCES

The electronic databases which were searched included PubMed, Cinahl, Medline, Google Scholar, and Proquest. The bibliography of all relevant articles was examined and associated articles were identified using a snowballing technique.

SELECTION CRITERIA

For inclusion, articles were required to meet the following criteria: (a) used a health-related database, (b) focused on injury-related cases, AND used machine learning approaches to analyse textual data.

METHODS

The papers identified through the search were screened resulting in 16 papers selected for review. Articles were reviewed to describe the databases and methodology used, the strength and limitations of different techniques, and quality assurance approaches used. Due to heterogeneity between studies meta-analysis was not performed.

RESULTS

Occupational injuries were the focus of half of the machine learning studies and the most common methods described were Bayesian probability or Bayesian network based methods to either predict injury categories or extract common injury scenarios. Models were evaluated through either comparison with gold standard data or content expert evaluation or statistical measures of quality. Machine learning was found to provide high precision and accuracy when predicting a small number of categories, was valuable for visualisation of injury patterns and prediction of future outcomes. However, difficulties related to generalizability, source data quality, complexity of models and integration of content and technical knowledge were discussed.

CONCLUSIONS

The use of narrative text for injury surveillance has grown in popularity, complexity and quality over recent years. With advances in data mining techniques, increased capacity for analysis of large databases, and involvement of computer scientists in the injury prevention field, along with more comprehensive use and description of quality assurance methods in text mining approaches, it is likely that we will see a continued growth and advancement in knowledge of text mining in the injury field.