Serious injury outcome indicators

Data sources for serious injury outcome indicators The two main data sources for the serious injury outcome indicators are the Ministry of Health’s (MoH) Mortality Collection and the National Minimum Dataset (NMDS) of hospital discharges. In addition to these, data from the Accident Compensation Corporation (ACC), Coronial Services of New Zealand, the Ministry of Transport, WorkSafe NZ, and Water Safety New Zealand are used to provide counts for the relevant injury priority areas. Fatal injury indicators The Mortality Collection (Ministry of Health, nd(a)) contains information on all deaths registered in New Zealand, and is the primary data source for the fatal indicators. The majority of indicators are based on Mortality Collection data, with the exception of the following indicators: fatal work-related injury indicators, which are based on ACC fatal claims data and WorkSafe NZ fatal notifications. provisional fatal suicide indicators, which are based on Coronial Services of New Zealand Annual Provisional Suicide Statistics. provisional fatal motor-vehicle traffic crash indicators, which are based on Police Traffic Crash Report data supplied by the Ministry of Transport. provisional fatal drowning indicators, which are based on Water Safety New Zealand DrownBase data. Serious non-fatal injury indicators The MoH’s NMDS is the primary data source for the serious non-fatal injury indicators (Ministry of Health, nd(b)). The NMDS is a national administrative database that records information on all publicly funded hospital discharges in New Zealand. Although the NMDS only reports on public hospital discharges, research has shown that only a small number of relevant cases are not identified in the serious injury outcome indicators because of the exclusion of privately funded cases (Langley, 1995). Alternative data sources are used for the work-related injury indicators. These indicators are based on ACC work related entitlement claims data linked to the NMDS data. This data linkage is required as work-relatedness is not comprehensively recorded in the NMDS, and severity cannot be calculated from the ACC data. Serious (fatal and serious non-fatal) injury indicators The separate fatal and serious non-fatal indicators are supplemented by the serious (fatal and serious non-fatal) injury indicators. The data for these indicators is the sum of the relevant fatal and serious non-fatal indicator counts. The serious indicators are included to present a more comprehensive picture of serious injury trends in New Zealand. Examining fatal and serious non-fatal injury events together allows an understanding of the potential impact of independent factors, such as improved medical treatment, to be accounted for. For example, improvements in treatment may mean that more people are surviving their injuries than before, resulting in a decline in the number or rate of fatal injuries. However, the result would be a shift of cases from fatal to serious non-fatal, rather than an overall reduction in injury cases. Rates and denominator data sources Denominator data provide counts for the population of interest and are used to estimate injury rates. For the serious injury outcome indicators, age standardised rates are used to provide an estimate of an individual’s risk of injury, as well as to allow for changes in the population age structure over time. Adjusting for the change in age structure over time means that trends independent of this change in the population can be presented. Population data are obtained from national population estimates (Statistics NZ, 2023). These are year-end estimated resident population values and include only the usually resident population. This excludes short-term overseas visitors, which may create a small difference between some numerator and denominator rates, as the NMDS will include tourists hospitalised with any injury. However, this is unlikely to have a significant impact on the indicators and trends. There are a few exceptions to the use of whole population estimates for the denominators. These include: Motor-vehicle traffic crash indicators. These indicators are based on Ministry of Transport data and are expressed as either rate per billion vehicle kilometres, or per 100,000 registered vehicles. Work-related injury indicators. These indicators are based on Statistics NZ’s Household Labour Force Survey data, which collects information on full-time and part-time workers aged 15 years and older. These indicators are expressed as a rate per 100,000 full-time equivalent employees (FTEs). Injury rates are expressed as per 100,000 person-years at risk (or per 100,000 population per year of exposure). Person-years are the natural units used when presenting rates, and where an indicator is based on moving averages (see three-year moving average), person-years easily take into account the effect of using multiple years to construct the rate estimates. The serious non-fatal injury severity threshold A serious non-fatal injury case is defined as one that is hospitalised and has a probability of death (at admission) of at least 6.9 percent. By applying the severity threshold of 6.9 percent, the level at which an injury is classified as ‘serious’ can be quantified. This is achieved by calculating an estimate for survival probability or an Injury Severity Score for each NMDS injury case. International Classification of Disease (ICD)-based Injury Severity Score The ICD-based Injury Severity Score (ICISS) is a survival probability measure. It is a useful tool for estimating injury severity based on administrative data (Cryer et al, 2004). ICISS method ICISS methodology involves estimating the probability of surviving an injury. This is calculated by estimating a survival probability score for each injury event, which is based on the ICD injury diagnosis code. This is known as the diagnosis-specific survival probability (DSP). The individual DSPs are calculated by dividing the number of people who survive with a specific diagnosis by the total number of people assigned that diagnosis over a defined time period. (The term diagnosis-specific survival probability replaces the previously used term survival risk ratios (SRR), to reflect international standard terminology). The ICISS is the product of all the DSPs associated with the injury diagnoses listed for a hospital event. For example, a patient admitted with a diagnosis of concussion alone (ICD-10 code S0600) will have an ICISS equal to the DSP for that diagnosis (DSP = 0.993). However, the ICISS for a patient admitted with the diagnoses of concussion and a skull fracture (ICD-10 code S021, ICISS = 0.903) will be the product of the DSPs for each of these diagnoses. In this case, the ICISS is 0.897 (0.993 x 0.903). For the serious injury indicators, a hospitalised patient is identified as having a serious non-fatal injury if they have an estimated ICISS of ≤ 0.931. This ICISS equates to a risk of death of 6.9 percent or greater, as stated in the definition of serious non-fatal injury. From the example above, the patient with the concussion diagnosis alone would not meet the threshold for a serious non-fatal injury (ICISS of 0.993 is greater than 0.931), However, the patient with a concussion and skull fracture has an ICISS estimate that is less than the threshold, and so this injury event is considered a serious non-fatal event (ICISS of 0.897 is less than 0.931). Injury events that are more serious in nature will have a low DSP and probability of survival, together resulting in a low ICISS. For a discussion of the merits of ICISS relative to other severity scales, please see appendix 1 of Developing valid injury indicators (Cryer et al, 2004). ICISS and serious injury outcome indicators In the NMDS all diagnoses are coded according to the World Health Organization’s ICD. The serious injury outcome indicators present serious injury counts and rates from 2000 to the most current annual data available. During this time, NMDS diagnoses have been classified using the International Statistical Classification of Diseases and Health Related Problems, 10th Revision, Australian Modification (ICD-10-AM). This publication is referred to throughout this report as the ICD-10. Injury events are identified using the principal diagnosis (main reason for the hospital event) and the first listed e-code. Currently the ICD-10 based DSPs are calculated using NMDS discharge data for the period 2000 to 2017. The DSPs are then applied to all hospitalisations coded with a diagnosis code in the range S00–T78. From here the severity threshold of 6.9 percent or greater probability of death (ICISS of ≤ 0.931) is applied to define which non-fatal injury events are classified as serious. Work-related injury cases in more than one priority area Serious non-fatal work-related status is defined using data from the ACC claims dataset (see work-related injury). These claims are then linked to the NMDS to obtain more accurate diagnostic information. Because the cause of injury (work-relatedness) is obtained from an external data source (ACC), and there may be another cause of injury listed in the NMDS (eg fall) some work-related injuries will also be assigned to other priority groups. These cases appear in both sets of tables. They are counted only once in the 'all injury’ indicators. Work-related cases are not excluded from this process on the basis that they may have already been categorised into another priority group. For example, in the 2014 release, there were 400 serious non-fatal work-related injuries in the 2013 calendar year. Of these, 141 (35 percent) were also assigned to the falls priority group, 54 (14 percent) to motor-vehicle traffic crashes, and 9 (2 percent) to assault. Three-year moving averages The exact number of deaths and injuries fluctuates randomly from year to year, even when there is no underlying trend. For most injury priority areas, these fluctuations are small compared with the overall numbers of injuries and deaths, so the trends are still clear. However, when the overall number is small, these fluctuations may hide trends in the numbers and rates of injury. To overcome this effect, we estimate three-year moving averages. This means, for example, that we use data from 2000, 2001, and 2002 to estimate an indicator value for 2001-03. Household Labour Force Survey HLFS weighting, imputation, and seasonal adjustment methods are described in the HLFS methodology: Household Labour Force Survey - Stats NZ DataInfo+ en-NZ