Urban–rural disparities of all severity trauma in a region of Aotearoa New Zealand

Traumatic injuries are public health concerns of national significance in several countries, including Aotearoa New Zealand.^1,2 Rural–urban disparities exist in trauma incidence and trauma characteristics.³ Several studies across countries have observed that people living in rural areas may experience higher mortality rates and lower life expectancies.^4,5 Studies evaluating demographic and injury characteristics have shown increased incidence related to rurality: younger ages and social disadvantage, and there are higher rates of traffic-related and intentional injuries.^5,6 Earlier studies in Aotearoa New Zealand have shown that roads, homes and areas of recreation and sports are the most common places of serious injury, with significant socio-economic, ethnic and gender disparities.^7,8

While trauma registries in Aotearoa New Zealand capture information on geographical areas and injury locations, there is limited exploration of geographical disparities, to some extent driven by emerging variations in the definitions of rurality.^9,10 National statistics indicate over 80% of the people in Aotearoa New Zealand live in urban areas,¹¹ yet the incidence of major trauma is observed to be higher among residents of rural areas.¹² Gender and ethnic disparities in injury incidence have continued in Aotearoa New Zealand, with men and Indigenous people, Māori, in rural areas having a higher risk of injury,^8,13 and socio-economic status and ethnicity have been identified as contributing to the inequities in health outcomes in patients with road traffic injuries.¹⁴

This study aims to extend the current knowledge on geographic disparities of trauma, describing differences in injury incidence, ethnicity variations and injury characteristics of trauma admissions in the Te Manawa Taki (TMT) Region of Aotearoa New Zealand, applying the urban–rural definitions of Geographic Classification for Health (GCH).¹⁰ The TMT region is located in the middle of the North Island of Aotearoa New Zealand, stretching from Taranaki Maunga in the west to Te Tairāwhiti, and includes the major population centres Tauranga, Rotorua, Gisborne, Hamilton and New Plymouth. The TMT Region’s population is approximately 1 million, representing 20% of the Aotearoa New Zealand population,¹⁵ and displays demographic characteristics reflective of the country for median age (39 vs 37 years) and gender (48 vs 50% male),¹⁵ but a higher proportion of Māori (26.5 vs 17.1%)¹⁶ and areas with most (Q5) deprivation (27 vs 21%).¹⁷ This study examines trauma patients of all injury severities, the incidence by rurality classified with GCH and ethnic variations, and the associations of causes of injury with levels of rurality that can inform injury prevention interventions and trauma care pathways.

Methods

The study is a retrospective observational study of the cohort of trauma admissions in the TMT Region.

Data sources

Trauma hospitalisation data collected from the TMT trauma registry were analysed. The study included patients of all Injury Severity Score (ISS), all ages, and admitted to the hospitals in the TMT Region during the 10 years between 1 January 2013 and 31 December 2022.

The trauma registry collects comprehensive data on all patients across the TMT Region admitted to the hospital with an injury, independent of overall severity or admitting speciality. For the trauma registry, “admission” records are created for each period during which a trauma patient occupies an inpatient bed, or when treatment procedures occur within an emergency department (ED) without an inpatient stay at that facility, but the patient is subsequently transferred to an inpatient bed at another hospital or discharged home. Consistent with trauma registries internationally, patients were excluded from the registry if they sustained insufficiency or periprosthetic fractures, exertional injuries, hanging/drowning/asphyxiation without evidence of external force, poisoning, ingested foreign body, injury as a direct result of pre-existing medical conditions or late effects of injury, or the injury occurred more than 7 days prior to admission.¹⁸

In addition to injury characteristics, the TMT trauma registry contains health data routinely collected in hospitals, including patient demographic and injury event information from prehospital records, hospital information systems and, where required, directly from patients. Ethnicity is recorded in the registry at two levels: one as self-identified by the patient and the second at a high-level classification of Māori and non-Māori. Injury intent is defined in the registry as unintentional, by other, and self-inflicted, and cause and place of injury are recorded using the International Classification of Disease.¹⁹ Data for the trauma registry are collected by the trauma clinical nurse specialists and entered into the electronic trauma registry. Data quality is set by the TMT trauma registry data dictionary, and quarterly quality checks address completeness, validity, sequence, logic, coding accuracy, duplicate records and ethnicity coding.

The TMT trauma registry uses the Abbreviated Injury Scale (AIS), an anatomical scoring system used internationally to grade the severity of each and every individual injury on a scale of 1 (minor) to 6 (unsurvivable injury).²⁰ The ISS is calculated from the sum of the squares of the highest AIS grade in each of the three most severely injured body regions.²⁰ This summation value thus integrates and numerically describes the overall severity of multiple injuries ranging from ISS 1 to a maximum of 75. An ISS of 75 predicts the worst prognosis. An ISS >12 or death (any ISS) is considered major trauma in the National Major Trauma Registry and injury severity for this study is classified as major (ISS >12 or died) and non-major (ISS <13).²¹ The TMT trauma registry employs AIS Version 2008 for all diagnoses, thus ensuring the consistency of computed ISS throughout the extracted dataset.²⁰

Variables examined in this study included: patient demographic characteristics (age, gender, ethnicity and rurality of patient residence classified with GCH), injury event information (cause, and place of injury) and outcome (died/survived) at discharge from the hospital. Records with missing data in the variable examined are excluded from the analysis. Trauma data were extracted from the TMT trauma registry using DI Writer/Collector^TM.

Rurality

The patient’s home residence “rurality” variable was developed via mapping to the GCH, a rural–urban classification specifically developed and proposed for use in health research and policy.^10,17,22 The five-level GCH, a geographic classification with two urban (Urban 1 [U1], Urban 2 [U2]) and three rural categories (Rural 1 [R1], Rural 2 [R2], Rural 3 [R3]), was developed for the purpose of monitoring rural–urban differences in health outcomes.^17,22 While the GCH does not use access to specific services or rely on health statistics to define rurality, travel time to major population areas is a key determinant. For the urban categories, it is accessibility to the major urban areas within 25 (U1) and 20 minutes (U2). For rural categories, it is accessibility to a major urban area within 26 to 60 minutes (R1), within 61 to 120 minutes (R2) and more than 120 minutes (R3)^10,22 (see references 10 and 22 for detailed information on GCH). For this study, rurality was assigned based on patients’ residential addresses into the five GCH categories. While differences in socio-economic profiles may exist between the five GCH categories, including between U1 and U2 GCH classified Statistical Areas (SA), given the importance of travel time to trauma care, and to focus on comparing rurality classifications, we have not considered differences in socio-economic profiles in this study.

Anonymised patient residence addresses recorded in the TMT trauma registry were geocoded using the “ggmap” package²³ within R Studio. The resulting latitude and longitude were then mapped to Statistical Area 2 (SA2) via a spatial join using ArcGIS. Residence address SA2 was subsequently mapped to GCH classifications. Population data for the TMT catchment area were obtained from Stats NZ Tatauranga Aotearoa (Stats NZ), the New Zealand national statistics office, with a custom data extract at the SA2 level, based on the 2018 usually resident census. The SA2 area populations were subsequently mapped and aggregated to the GCH classifications to provide incidence population denominators. Due to low population volumes in some SA2s, and to protect privacy, the supplied population data for SA2 was at a high-level classification of Māori and non-Māori.

Statistical analysis

For the statistical analysis, age was classified into 0–14 years, 15–44 years, 45–64 years and 64 years and above; gender as male and female; and ethnicity as Māori and non-Māori. Annual incidence per 100,000 population was calculated, incorporating population figures drawn from the custom data extract provided by Stats NZ at the SA2 level. To calculate comparable rural and urban incidence rate ratios (IRRs), the five GCH classifications were used. The five classifications were collapsed to two categories—Urban (by collapsing U1 and U2) and Rural (by collapsing R1, R2 and R3)—for inferential statistics, an approach previously employed in similar rural health studies in Aotearoa New Zealand.^10,12,22 IRRs—and, when applicable, adjusted IRR (aIRR) corrected for population characteristics such as ethnicity—with associated p-values were calculated. IRR and aIRR were calculated with negative binomial regression using mean annualised events and population (at study midpoint) as offset, and using the log link. Descriptive statistics were produced, and Chi-squared tests were used for all categorical variables between the GCH categories. Odds ratio was calculated to analyse similarities and differences in the causes of injury between urban and rural classifications. Statistical significance was set at p< 0.05 for the statistical tests.

All statistical analyses were performed using RStudio 2023.06.0²⁴ with “ggmap”²³ and “MASS” ²⁵ packages.

Ethics

Ethical approval was deemed out of scope by the New Zealand Health and Disability Ethics Committee, and research approval was provided under the locality authorisation process by the Health New Zealand – Te Whatu Ora Waikato Research Office and the Māori Research Review Committee (RD023079).

Results

A total of 51,252 TMT trauma patient residence addresses were geocoded, spatially joined to SA2 polygons and subsequently classified into the five GCH categories of U1, U2, R1, R2 and R3 (Table 1). Approximately 69% of trauma patients were classified as urban dwellers (U1 and U2) and 31% as rural dwellers (R1, R2 and R3). The proportion of males injured was higher in rural areas compared with urban areas (χ²=166.1, p<0.01). The median age was 34 years, with most patients in the age group of 15–44 years in all areas. Compared with the rural areas, there were more trauma patients of the older life stage (65+ year-olds) in the urban areas (χ²=294.2, p<0.01). The proportion of Māori was higher in rural areas, and non-Māori were highest in U1 (χ²=1,039.7, p<0.01). The proportions remained relatively similar when comparing injury severity, major (ISS >12 or died) and non-major (ISS <13) trauma.

View Table 1–4, Figure 1.

When adjusted by population, patients dwelling in more rural areas (R2 and R3) were at significantly higher risk of trauma (p<0.01) compared with trauma patients in most urban areas (U1), and so were people in U2 areas (Table 2). The IRR for trauma was highest among R3 dwellers, with rates of 2.0 for overall trauma (p<0.01), 2.7 for major injuries (p<0.05) and 2.0 for non-major injuries (p<0.01) compared with U1 dwellers. The risk of trauma in general, and particularly non-major trauma, was also higher among R2 (p<0.01) and U2 (p<0.01) dwellers compared with U1; however, this increased risk was not observed for major trauma. There were no significant differences in the risk of trauma between U1 and R1 dwellers.

Risk of injury occurring at home was not significantly different across the GCH classifications. The IRR of non-medical death due to trauma was lower in all GCH classifications compared to U1, but the differences were not statistically significant (p>0.05) (Table 2).

The higher relative incidence of trauma among persons residing in R3 has persisted over time. The trend in the trauma incidence in all GCH classifications has remained more or less constant, with a gradual decrease since 2020 that coincides with the COVID-19 pandemic control interventions (Figure 1). This decrease continued in 2022 in U1, U2, R1 and R2 but had started to increase in R3.

Māori had a significantly higher risk of trauma admission compared with non-Māori in U1, U2, R1 and R2 areas (p<0.01), but not in R3 (Table 3).

Those living in rural areas (R1, R2, R3) had a significantly higher risk for motorcycle and road traffic crashes compared with urban-dwelling (U1, U2) persons, with almost double the risk for motorcycle-related injury and 1.5 times the risk for road traffic crashes for those living in rural/remote rural areas (Table 4).

Rural dwellers have a lower risk of trauma from falls and with sharps/glass/knives/hand tools. The risk of assault was similar irrespective of the geographic area of residence.

Discussion

Significant inequities in trauma incidence by rurality were observed in this study, with remote dwellers at a higher relative risk of trauma compared with urban trauma patients. This corresponds with several studies across different geographies, including Aotearoa New Zealand, that have observed health disadvantages in rural areas.^5,14,17 Importantly, we observed that while there was a gradual decrease in trauma incidence across all GCH classifications since 2020 that continued until 2022, the incidence in the most rural areas (R3) had started to increase in 2022. The reason for the continued decline across other GCH classifications after the COVID-19 pandemic and the upward trend in R3 indicates there are other factors at play that need further investigation.

We observed variations in injury incidence by injury severity, with dwellers in most rural areas (R3) at a slightly increased risk of hospitalisation for major trauma events, but not so in other areas. This difference was not reflected in the survival outcomes, and the risk of in-hospital non-medical death was not significantly different across the rurality classification. Existing knowledge from other studies, however, suggests a higher mortality for young rural males.⁵ The time to reach a hospital providing trauma service may be a factor influencing the findings, in that major trauma patients in rural areas may not survive long-distance transfers. The contributions of prehospital transfer times, and possibly health-seeking behaviour among urban, rural and remote rural dwellers for different injury severities on survival and health outcomes of trauma patients need further study.

Road trauma in rural areas of Australia and Aotearoa New Zealand has been higher compared with urban areas.²⁶ The findings of this study concur that the risk of motorcycle and road traffic crashes is higher among rural and remote rural dwellers compared to the urban, with almost double the risk for non-road motorcycle-related injury for those in remote areas. This could be related to road conditions, such as unsealed roads, and the use of off-road vehicles, which is more likely in rural areas, but this aspect needs further investigation. Work-related injury research in Aotearoa New Zealand has shown that farm workers are the occupational group commonly injured during work.²⁷ Job and Brodie²⁸ note that driving too fast for the conditions is one of the main causes of head-on road traffic crashes, and adherence to road-use safety standards is noted to be essential for trauma prevention.

Urban dwellers are at a higher risk of injury from falls and sharps/glass/knives/hand tools, perhaps related to the design features of housing and urban living environments, which need further exploration to ascertain contributing factors. Urban dwellers experience more pedal cycle injuries, which may be due to higher use of pedal cycles in urban areas.²⁹

This study observed that Māori had a significantly higher incidence of trauma compared with non-Māori in four GCH classifications (U1, U2, R1 and R2), and this is consistent with previous studies of trauma in Aotearoa New Zealand that show Māori have a significant disadvantage.³⁰ Although no ethnic variation was observed in the most remote areas (R3) compared with U1, this was not statistically significant, and needs further exploration to draw conclusions. Notwithstanding the lack of variation in incidence by ethnicity across rural classifications, the study shows that the major inequities in the incidence of trauma affecting Māori are pervasive and reinforces the need for urgent and specific attention to reduce risks for Māori.

Rurality as a determinant of health disadvantage may be contested on the premise that it is not the rural area, per se, but rather the underlying conditions, such as socio-economic disadvantages, ethnic and demographic variations of residents, availability and access to services, and environmental conditions.^13,14 However, this study did not observe a significantly higher risk of those in R1 compared with U1. Taking this premise, the differences in trauma incidence by rurality observed may be explained by the demographic and socio-economic profile of the residents, in addition to the level of deprivation of these areas.

Our study is constrained by a specific GCH classification and cannot be compared with other studies using different urban or rural classifications. Furthermore, the study explored only ethnic variations of trauma by rurality and was not able to explain the findings in relation to socio-economic status that could have mediated the outcomes. In defining admissions, this study did not differentiate “staging” admissions, where emergent treatment is provided in the ED before onward transfer to a trauma service. It also did not capture possible prehospital deaths. Further studies are required to examine the extent and nature of these aspects. Despite these limitations, this study provides evidence on the observed variations in the patterns of trauma in the different urban and rural areas of the TMT Region that can meaningfully inform trauma prevention programmes in Aotearoa New Zealand.

Conclusion

Rurality was associated with a higher relative risk of trauma admissions, particularly associated with non-major traumatic injury. Māori showed increased risk in all GCH categories except R3. Falls were dominant as a cause across all classifications, followed by road traffic crashes, which disproportionately affected rural and remote rural dwellers. Future research into rurality may explore the role of socio-economic factors that mediate the risk of trauma across the GCH areas.

The study reinforces the need for injury prevention strategies that are customised to the known variations in patterns of urban and rural trauma.