Impact of legislative reform on worker fatalities in New Zealand workplaces: a 30-year retrospective population-level analysis.

OBJECTIVES: To determine the impact of major legislative changes to New Zealand's Occupational Health and Safety (OHS) legislation with the adoption of the Robens model as a means to control occupational risks on the burden and risk of work-related fatal injury (WRFI). METHODS: Population-based comparison of WRFI to workers aged 15-84 years occurring during three periods: before (pre:1985-1992), after legislative reform (post-1:1993-2002) and after subsequent amendment (post-2:2003-2014). Annual age-industry standardised rates were calculated with 95% CI. Multivariable Poisson regression was used to estimate age-adjusted annual percentage changes (APC) for each period, overall and stratified by high-risk industry and occupational groups. RESULTS: Over the 30-year period, 2053 worker deaths met the eligibility criteria. Age-adjusted APC in rates of worker WRFI changed little between periods: pre (-2.8%, 95% CI 0.0% to -5.5%); post-1 (-2.9%, 95% CI -1.3% to -4.5%) and post-2 (-2.9%, 95% CI -1.3% to -4.4%). There was no evidence of differences in slope. Variable trends in worker WRFI were observed for historically high-risk industry and occupational groups. CONCLUSIONS: The rate of worker WRFI decreased steadily over the 30-year period under examination and there was no evidence that this pattern of declining WRFI was substantially altered with the introduction of Robens-styled OHS legislative reforms. Beyond headline figures, historically high-risk groups had highly variable progress in reducing worker WRFI following legislative reform. This study demonstrates the value in including prereform data and high-risk subgroup analysis when assessing the performance of OHS legislative reforms to control occupational risks.

Missing cases of injury death: use of quantitative methods and case reviews to inform future improvements in case definition.

BACKGROUND: Knowledge of fatal injuries is required to inform prevention activities. Where hospital patients with an injury principal diagnosis (PDx) died and were certified to a medical underlying cause of death (UCoD), there is the potential to underestimate injury mortality. We sought to characterise injury/non-injury (NI) mismatches between PDx and UCoD by identifying which subgroups had small/large mismatches, and to understand why mismatches had occurred using informative examples. METHOD: Hospital records (n=10 234) with a PDx of injury were linked to the mortality collection using a unique personal identifier. Percentages UCoD coded to a NI were tabulated, for three follow-up periods and by selected variables. Additionally, we reviewed a sample of 70 records for which there was a mismatch. RESULTS: %NIs were 39%, 66% and 77% for time from injury to death of <1 week, <90 days and <1 year, respectively. Variations in %NI were found for all variables. Illustrative examples of 70 medical UCoD deaths showed that for 12 cases the injury event was unequivocally judged to have resulted in premature death. A further 16 were judged as injury deaths using balance of probability arguments. CONCLUSION: There is variation in rates of mismatch between PDx of injury and UCoD of NI. While legitimate reasons exist for mismatches in certain groups, a material number of injury deaths are not captured using UCoD alone; a new operational definition of injury death is needed. Early solutions are proposed. Further work is needed to investigate operational definitions with acceptable false positive and negative detection rates.

Use of mixed methods to investigate case definitions to improve the identification of serious injury cases from hospital episode data.

INTRODUCTION: It has been commonplace internationally, when using hospital data, to use the principal diagnosis to identify injury cases and the first external cause of injury code (E-code) to identify the main cause. Our purpose was to investigate alternative operational definitions of serious non-fatal injury to identify cases of interest for injury surveillance, both overall and for four common causes of injury. METHODS: Serious non-fatal injury cases were identified from New Zealand (NZ) hospital discharge data using an alternative definition: that is, case selection using principal and additional diagnoses. Separately, identification of cause used all E-codes on the discharge record. Numbers of cases identified were contrasted with those captured using the usual definition. Views of NZ government stakeholders were sought regarding the acceptability of the additional cases found using these alternative definitions. Views of international experts were also canvassed. RESULTS: When using all diagnoses there was a 7% increase in 'all injury' cases identified, a 17% increase in self-harm cases and 8% increase in falls cases. Use of all E-codes resulted in a 4% increase in self-harm cases, 2% increase in assault cases and 1% increase in both falls and motor vehicle traffic crash cases. DISCUSSION: A case definition based solely on principal diagnosis fails to count a material number of serious non-fatal injury cases that are of interest to the injury prevention community. There is a need, therefore, to use an alternative case definition that includes additional diagnoses. Use of multiple E-codes to classify cause of injury should be considered.

Using hospital discharge data for injury research or surveillance? An observational study illustrating the impact of administrative change.

INTRODUCTION: Hospital discharge data provide an important basis for determining priorities for injury prevention and monitoring trends in incidence. This study aims to illustrate the impact of a recent change in administrative practice on estimates of hospitalised injury incidence and to investigate the extent to which different case selection affects trends in injury incidence rates. METHODS: New Zealand (NZ) hospital discharges (2000-2014) with a primary diagnosis of injury were identified. Additional case selection criteria included first admissions only, and for serious injury, a high threat-to-life estimate. Comparisons were made, over time and by District Health Board, between hospitalised injury incidence estimates that included, or not, short-stay emergency department (SSED) discharges. RESULTS: Of the 1 229 772 injury hospital discharges, 365 114 were SSED; 16% of the annual total in 2000, 38% in 2014. Identification of readmissions prior to the exclusion of SSED discharges resulted in 30 724 cases being erroneously removed. Age-standardised rates of hospitalised injury over the 15-year period increased by, on average, 2.7% per year when SSED discharges were included; there was minimal secular change (-0.2%) when SSEDs were excluded. For serious hospitalised injury, the annual increase was 2.3% when SSED was included compared with 1.1% when SSEDs were excluded. CONCLUSION: Spurious trends in hospitalised injury incidence can result when administrative practices are not appropriately accounted for. Exclusion of SSED discharges before the identification of readmissions and the use of a severity threshold are recommended to minimise the reporting bias in NZ hospitalised injury incidence estimates.

Empirical validation of the New Zealand serious non-fatal injury outcome indicator for 'all injury'.

Our purpose was to empirically validate the official New Zealand (NZ) serious non-fatal 'all injury' indicator. To that end, we aimed to investigate the assumption that cases selected by the indicator have a high probability of admission. Using NZ hospital in-patient records, we identified serious injury diagnoses, captured by the indicator, if their diagnosis-specific survival probability was ≤0.941 based on at least 100 admissions. Corresponding diagnosis-specific admission probabilities from regions in Canada, Denmark and Greece were estimated. Aggregate admission probabilities across those injury diagnoses were calculated and inference made to New Zealand. The admission probabilities were 0.82, 0.89 and 0.90 for the regions of Canada, Denmark and Greece, respectively. This work provides evidence that the threshold set for the official New Zealand serious non-fatal injury indicator for 'all injury' captures injuries with high aggregate admission probability. If so, it is valid for monitoring the incidence of serious injuries.

Towards valid 'serious non-fatal injury' indicators for international comparisons based on probability of admission estimates.

BACKGROUND: Governments wish to compare their performance in preventing serious injury. International comparisons based on hospital inpatient records are typically contaminated by variations in health services utilisation. To reduce these effects, a serious injury case definition has been proposed based on diagnoses with a high probability of inpatient admission (PrA). The aim of this paper was to identify diagnoses with estimated high PrA for selected developed countries. METHODS: The study population was injured persons of all ages who attended emergency department (ED) for their injury in regions of Canada, Denmark, Greece, Spain and the USA. International Classification of Diseases (ICD)-9 or ICD-10 4-digit/character injury diagnosis-specific ED attendance and inpatient admission counts were provided, based on a common protocol. Diagnosis-specific and region-specific PrAs with 95% CIs were calculated. RESULTS: The results confirmed that femoral fractures have high PrA across all countries studied. Strong evidence for high PrA also exists for fracture of base of skull with cerebral laceration and contusion; intracranial haemorrhage; open fracture of radius, ulna, tibia and fibula; pneumohaemothorax and injury to the liver and spleen. Slightly weaker evidence exists for cerebellar or brain stem laceration; closed fracture of the tibia and fibula; open and closed fracture of the ankle; haemothorax and injury to the heart and lung. CONCLUSIONS: Using a large study size, we identified injury diagnoses with high estimated PrAs. These diagnoses can be used as the basis for more valid international comparisons of life-threatening injury, based on hospital discharge data, for countries with well-developed healthcare and data collection systems.

Internationally comparable diagnosis-specific survival probabilities for calculation of the ICD-10-based Injury Severity Score.

BACKGROUND: The International Statistical Classification of Diseases, 10th Revision (ICD-10)-based Injury Severity Score (ICISS) performs well but requires diagnosis-specific survival probabilities (DSPs), which are empirically derived, for its calculation. The objective was to examine if DSPs based on data pooled from several countries could increase accuracy, precision, utility, and international comparability of DSPs and ICISS. METHODS: Australia, Argentina, Austria, Canada, Denmark, New Zealand, and Sweden provided ICD-10-coded injury hospital discharge data, including in-hospital mortality status. Data from the seven countries were pooled using four different methods to create an international collaborative effort ICISS (ICE-ICISS). The ability of the ICISS to predict mortality using the country-specific DSPs and the pooled DSPs was estimated and compared. RESULTS: The pooled DSPs were based on a total of 3,966,550 observations of injury diagnoses from the seven countries. The proportion of injury diagnoses having at least 100 discharges to calculate the DSP varied from 12% to 48% in the country-specific data set and was 66% in the pooled data set. When compared with using a country's own DSPs for ICISS calculation, the pooled DSPs resulted in somewhat reduced discrimination in predicting mortality (difference in c statistic varied from 0.006 to 0.04). Calibration was generally good when the predicted mortality risk was less than 20%. When Danish and Swedish data were used, ICISS was combined with age and sex in a logistic regression model to predict in-hospital mortality. Including age and sex improved both discrimination and calibration substantially, and the differences from using country-specific or pooled DSPs were minor. CONCLUSION: Pooling data from seven countries generated empirically derived DSPs. These pooled DSPs facilitate international comparisons and enables the use of ICISS in all settings where ICD-10 hospital discharge diagnoses are available. The modest reduction in performance of the ICE-ICISS compared with the country-specific scores is unlikely to outweigh the benefit of internationally comparable Injury Severity Scores possible with pooled data. LEVEL OF EVIDENCE: Prognostic and epidemiological study, level III.

The epidemiology of life-threatening work-related injury--a demonstration paper.

BACKGROUND: Workers' compensation (WC) data traditionally provides information to stakeholders on work-related disabling injuries. It is important to complement this with information on serious threat to life (TTL) injury, which is the focus of this paper. METHODS: In this cross-sectional descriptive epidemiological study, based on New Zealand's WC data linked to hospital discharge data, TTL was measured using the ICD10-based Injury Severity Score (ICISS); ICISS ≤ 0.941 was used to define serious TTL injury. RESULTS: During 2002-2004, there was an average of 368 serious TTL work-related injury cases annually. The distribution of these injuries was very different from those traditionally found using WC data to describe disabling injury. For example, for serious TTL injury the main injury types included traumatic brain injury, whereas for disabling injury it was sprains and dislocations. CONCLUSIONS: The method presented provides the opportunity for government agencies to produce a national description of the epidemiology of serious TTL work-related injuries.

An outcome evaluation of a New Zealand farm safety intervention: a historical cohort study.

BACKGROUND: There is limited evidence that farm safety-related interventions based solely on an educational element have an effect on injury rates. Our aim was to evaluate a New Zealand national educational program, FarmSafe™ Awareness, for its effect on injury rates. METHODS: We used a before-after design followed by a historical cohort study of sheep, beef, and dairy farmers/workers. The outcomes were work-related injuries, identified from workers compensation data. Cox regressions were used to compare intervention with matched control group rates. RESULTS: FarmSafe™ Awareness was associated with significantly higher rates of work-related injury, than matched controls. CONCLUSIONS: It is difficult to see how FarmSafe™ Awareness could be causing an increased rate of work-related injury. We detected no reporting bias, and selection bias is likely to act in the opposite direction to the observed results. We conclude that there is no evidence that FarmSafe™ Awareness prevents farm injury.

Monitoring trends in violence: a delayed response to Estrada (2006).

Reliable and valid indicators of assault are required to effectively monitor population trends and ensure that resources are targeted effectively. Trends in assault, reported by the media, based on crime statistics, or on victim surveys, are substantively affected by extraneous factors. In 2006, Estrada offered up solutions to the difficulties posed by crime statistics and victim surveys, by proposing the development of indicators based on hospital discharge data, albeit with identified limitations. This article is a response to Estrada's proposition, and works through each of Estrada's identified limitations of hospital discharge data. Potential problems with Estrada's suggestions are highlighted in our article and solutions, based on the current evidence, are proposed.

Measuring the population burden of fatal and nonfatal injury.

The value of measuring the population burden of fatal and nonfatal injury is well established. Population health metrics are important for assessing health status and health-related quality of life after injury and for integrating mortality, disability, and quality-of-life consequences. A frequently used population health metric is the disability-adjusted life-year. This metric was launched in 1996 in the original Global Burden of Disease and Injury study and has been widely adopted by countries and health development agencies alike to identify the relative magnitude of different health problems. Apart from its obvious advantages and wide adherence, a number of challenges are encountered when the disability-adjusted life-year is applied to injuries. Validation of disability-adjusted life-year estimates for injury has been largely absent. This paper provides an overview of methods and existing knowledge regarding the population burden of injury measurement. The review of studies that measured burden of injury shows that estimates of the population burden remain uncertain because of a weak epidemiologic foundation; limited information on incidence, outcomes, and duration of disability; and a range of methodological problems, including definition and selection of incident and fatal cases, choices in selection of assessment instruments and timings of use for nonfatal injury outcomes, and the underlying concepts of valuation of disability. Recommendations are given for methodological refinements to improve the validity and comparability of future burden of injury studies.

The New Zealand serious non-fatal self-harm indicators: how valid are they for monitoring trends?

BACKGROUND: To monitor accurately injury incidence trends, indicators should measure incidence independently of extraneous factors. Frequencies and rates of New Zealand's serious non-fatal self-harm indicators may be prone to fluctuations in reporting owing, for example, to changing social norms. Hence, they have been considered provisional. AIM: To validate empirically the serious non-fatal self-harm indicators. Methods All serious non-fatal first admissions to hospital were identified and classified according to whether principal diagnosis (PDx) was injury or mental disorder, and conversely whether contributing diagnoses were mental disorder or injury. The proportion assigned self-harm external cause of injury code (E-code) was calculated for each year from 2001 to 2007. Subsequently, all cases with a self-harm E-code were identified, and the proportion with a PDx of injury and contributing diagnosis of mental disorder, or PDx of mental disorder and contributing diagnosis of injury over time, were determined. RESULTS: No linear changes over time were detected in the proportion of cases assigned an injury PDx, or the proportion assigned a mental disorder PDx, or the proportion with a self-harm E-code. The estimated maximum observed increase in the frequency of serious non-fatal self-harm hospitalisation explained by changes in reporting was 19- 40%. CONCLUSION: Identification of serious non-fatal self-harm events using an operational definition of PDx of injury, a self-harm first listed E-code, and an appropriate severity cut-off point, is a valid method of monitoring incidence and rates in New Zealand.

Identifying Maori ethnicity for estimating trends in fatal and serious non-fatal injury.

OBJECTIVES: Counts of mortality and morbidity based on routinely collected national datasets have undercounted Maori, the indigenous people of New Zealand. To correct for the undercount, when estimating fatal and serious non-fatal injury incidence, the 'ever-Maori' method has been used. This study sought to determine how well the ever-Maori method corrects for the undercount. METHODS: Trends in frequencies and age-standardised rates for fatal injury indicators were compared using: (a) ever-Maori classification; (b) New Zealand Census Mortality Study adjustment ratios applied to Total Maori counts from the Mortality Collection; and (c) Total Maori counts from the Mortality Collection. For serious non-fatal injury, trends using ever-Maori were compared with Total Maori from hospital discharge data. RESULTS: The absolute number of injuries attributable to Maori varied depending on the method used to adjust for ethnicity status, but trends over time were comparable. CONCLUSIONS AND IMPLICATIONS: At present, there is no optimal method for adjusting for the undercount of Maori in routinely collected health databases. Reassuringly, trends in fatal and serious non-fatal injury are similar across the methods of adjusting for the undercount.

Injury mortality indicators: recommendations from the International Collaborative Effort on Injury Statistics.

BACKGROUND: The International Collaborative Effort (ICE) on Injury Statistics called for an effort 'to reach consensus on what are the 10 most important indicators of injury incidence that offer the potential for international comparisons and for regional or global monitoring.' OBJECTIVES: To describe the process of developing the ICE indicators and to present the specifications of selected injury mortality indicators, along with comparisons between selected countries for those specified indicators. METHODS: Participants on the ICE list had been asked to send to one of the authors what they considered the most important five indicators of injury incidence. These were synthesised and presented under six themes: mortality indicators (general); mortality indicators (motor vehicle); mortality indicators (other); hospital data-based (overall); hospital data-based (traumatic brain injury (TBI)); long-term disability (overall). Following two work group discussions and after drafting and revising indicator specifications, agreement was reached on mortality indicators and specifications. Specifications for those mortality indicators are presented. Morbidity indicators are still to be agreed. RESULTS: The mortality indicators proposed were age-adjusted rates of injury death; motor vehicle traffic crash-related death; self-harm/suicide; assault/homicide; and TBI death. The empirical work highlighted difficulties in identifying TBI in countries where mortality data do not capture multiple injuries, prompting us to drop the mortality indicator related to TBI and moving us instead to introduce an indicator to monitor the use of undetermined intent in the classification of injury mortality. CONCLUSION: The ICE has reached a consensus on what injury mortality indicators should be used for comparison between countries. Specifications for each of these have been applied successfully to the mortality data of seven countries.

An investigation into methods to develop indicators to measure injury related impairment.

OBJECTIVE: To investigate whether valid indicators of injury-related impairment could be developed based on data from a New Zealand national no-fault accident insurance scheme. DESIGN: The feasibility of (1) developing impairment related indicators using insurance data directly, and (2) deriving diagnosis specific probability of impairment measures based on hospital discharge data were investigated. SETTING: The source data were lump sum payment data for permanent impairment arising from an injury event between 1 April 2002 and 31 October 2006, and hospital discharge data for injury events between 1 April 2002 and 31 December 2005. MAIN OUTCOME MEASURE: A threat of impairment measure was developed that was based on a list of International Classification of Diseases codes (version 10) which, if assessed would attract a lump sum payment for permanent impairment, and would almost always be expected to be admitted to hospital. RESULTS: A group of approximately 80 diagnoses satisfying the above criteria were identified. The trend in age standardised injury rates for the threat of impairment indicator is consistent with the trend for the New Zealand threat to life indicator. CONCLUSIONS: This work has provided a method for the development of hospital discharge-based serious threat of impairment injury indicators. This is an important first step in developing a comprehensive package of (threat of) disability indicators.