The epidemiology of eye injuries in Western Australia: a retrospective 10-year study

ABSTRACT Clinical relevance Eye injuries constitute a significant cause of preventable lifelong visual impairment or blindness. It is important to identify the context in which these injuries occur to develop intervention programs to reduce the incidence and severity of injury. Background To evaluate the nature, external cause, place of occurrence and incidence rate of eye injuries treated at hospitals in Western Australia. Methods Retrospective, population-based study of patients presenting to all emergency departments or admitted to hospital with primary or secondary eye injuries between 2005 and 2014. Results The combined incidence rate of eye injuries requiring tertiary care was 278 per 100 000 person-years (95% CI 276–280). Significantly more males (79%, 44 569) presented to emergency departments (p < 0.001), and most injuries involved the cornea and conjunctiva (83%). The injury incidence rate was 248 per 100 000 person-years (95% CI 246–250). A total of 2823 and 3951 individuals were admitted to hospital for a primary or secondary eye injury, respectively. The most frequent primary diagnosis on admission was contusion (19%). Assault (24%) was the most common cause of injury requiring inpatient treatment. Indigenous individuals were hospitalised for an eye injury at a rate of 109 per 100 000 person-years (95% CI 102–116), compared to 27 (95% CI 26–27) for non-Indigenous individuals. Each year was associated with an increase in the mean number of eye injuries (7% and 5% for emergency department and hospital admission data, respectively). Conclusion Indigenous individuals and males experience eye injuries requiring tertiary management disproportionately. Indigenous female patients were conspicuously affected by eye injuries. Remedial intervention strategies should incorporate violence prevention as assault is a significant cause of eye injury.


Introduction
Ocular trauma represents a considerable preventable cause of monocular visual impairment or blindness. 1,2 The consequences of sustaining an eye injury resulting in long-term or permanent visual impairment is significant on both an individual and societal level, particularly as younger age groups are at highest risk. [3][4][5][6] The individual burden of injury is compounded by the economic cost of ongoing medical treatment and lost work time for affected individuals and their families. 3 Contemporaneous population-specific epidemiological data on the incidence and causes of ocular trauma are required to develop targeted public-health strategies to prevent injury in at-risk individuals and the broader population.
The epidemiology of eye injury in Australia is sparsely documented, with incidence rates differing according to residential status (urban versus rural) and regional location. Victorian rates of eye injury requiring admission to hospital increased from 15.2 per 100 000 person-years between 1989 and 1991 7 to 53.6 between 2001 and 2005. 8 For individuals residing in rural Victoria, the rate of injury requiring presentation to a hospital Emergency Department (ED) is even higher (950 per 100 000 in 2004). 9 In comparison, eye injuries occurred at a rate of 4.72 per 100 00 person-years 10 and 82.8 per 100 000 person-years for rural and regional New South Wales, 11 respectively. National age-standardised incidence rates of eye injury requiring hospitalisation have decreased (76.6 per 100 00 person-years between 1999 and 2006 12 to 11.9 between 2010-2011 and 2014-2015, 13 although data are not representative of all Australian states and Territories. Currently, there are no lifetime incidence rates of ocular trauma available for Western Australia, (WA), with existing data reporting exclusively on the paediatric population. 14,15 Previous studies have established significant variability in the reported incidence of eye injuries that require tertiary treatment in Australia. [7][8][9][10][11][12][13] This is likely the result of differing research methodologies and sampling intervals, coding and data collection standards, consideration of injuries at different levels of tertiary care and the inclusion of different communities within the population (e.g. rural and remote, Indigenous and non-Indigenous). The extrapolation of epidemiological data collected using discrete hospitals, care centres, regions and temporal intervals to a different state is therefore limited, and a comprehensive evaluation of the eye injury data in WA is warranted. Comprehensive epidemiological data are needed to inform approaches for preventing serious ocular trauma. The current study aims to identify the incidence rate and causative mechanisms of eye injury based on the retrospective analysis of patient records for hospitals in WA.

Methods
Eye injuries were identified using the Tenth edition of The International Statistical Classification of Diseases and Health Related Problems, Australian modification (Table  S1, supplementary material) in the WA Hospital Morbidity Data System and ED records between 2005 and 2014.
Hospital Morbidity Data System records represent patients admitted to all hospitals within WA, excluding outpatients. Patients were assigned a single, primary or principal diagnosis along with up to 20 additional or secondary comorbid diagnoses. Patients presenting to the ED during the study period were assigned a single diagnosis at the end of the episode of care. Both Hospital Morbidity Data System and ED datasets were provided by the WA Department of Health.
Cases within the hospital admission dataset with synchronised admission and separation dates and matching month and year of birth, age, sex, country of birth and residential postcodes were identified as probable inter-hospital transfers and were removed from total incidence counts. Duplicate cases within the ED dataset could not reliably be identified or excluded.
Descriptive analysis was performed to investigate age, sex, residential and Indigenous status (defined as Aboriginal or Torres Strait Islander, or both), mechanism and cause of eye injury. Residential status (metropolitan or rural/remote) was determined using the Rural, Remote and Metropolitan Area classification.
Age-weighted ocular injury incidence rates were calculated based on the distribution of the WA population as reported by the Australian Bureau of Statistics up to the 30 th of June each year. [16][17][18] Incidence rates were expressed in terms of 100 000 person-years. Unadjusted odds ratios were calculated to investigate the association between categorical variables and risk of eye injury. The denominator or atrisk population was the sum of all persons in the state. Injury incidence rates were evaluated using a negative binomial model with year and age as predictors to determine the rate of increase over the study period for both datasets. Further, categorical variables (age group, sex, residential and Indigenous status) were examined using a Chi-square goodness of fit test with an a priori significance of 0.05. Univariable and multivariable linear regression analysis performed on hospital admission data was used to investigate the association between all categorical variables with the length of hospital stay. Statistical analysis was completed using IBM

Results
The combined incidence rate of eye injuries presenting for tertiary management was 278 per 100 000 person-years (95% CI 276-280) over the study period, increasing from 233 (95% CI 226-240) to 302 (95% CI 295-309) per 100 000 personyears from 2005 to 2014. ED presentations accounted for approximately 89% of all cases of eye injury requiring tertiary care. ED and hospital admission datasets are presented separately herein.

ED presentations
There were 56 239 eye-injury presentations to EDs in WA between 2005 and 2014 ( Figure 1). Males presented significantly more frequently compared to females (44 569 to 11 656 respectively; unadjusted OR 3.77, 95% CI 3.70 to 3.85, p < 0.001). Demographic details for individuals in the ED data set are displayed in Table 1.
The majority of presenting injuries involved the cornea and conjunctiva, with or without the presence of a foreign body (46 682, 83% of all injuries). Traumatic cataract, penetrating wounds with a foreign body and avulsion of the eye were rare ( Table 2). The external cause of injury was recorded for 50% of presentations, and cutting, piercing and stabbing injuries were most common (11.6%) ( Table S2, supplementary material).

Hospital admissions
A total of 6774 individuals were admitted to hospital with an eye injury (66% males, 34% females) ( Figure 3). Of these, 2823 (71% males, 29% females) were diagnosed with a primary eye injury and 3951 (63% males, 37% females) received a secondary eye injury diagnosis. Demographic data for individuals admitted to hospital with a primary eye injury are displayed in Table 3.
Males were admitted to hospital with a primary eye injury significantly more frequently than females (unadjusted OR 2.42, 95% CI 2.23 to 2.62, p < 0.001). Female admissions only exceeded male admissions in elderly age categories (80 years and older). Indigenous individuals were significantly more likely to be admitted to hospital with a primary eye injury compared to non-Indigenous individuals (unadjusted OR 4.83, 95% CI 4.37 to 5.33, p < 0.001).
The most common primary eye injuries admitted to hospital included contusions (19%) and penetrating wounds of the eyeball and orbit (13%) ( Table 4). Primary injuries were most frequently caused by inanimate mechanical forces (42%) and assault (21%) ( Table S3, supplementary material). The most common mechanism of injury in the inanimate injury category was a foreign body entering the eye (13%); being struck by a thrown, projected or falling object (10%); and sporting equipment (5%). The location of primary injury was largely unspecified or unknown (62.5%), and most Table 1. Demographic information of persons attending emergency departments between 2005 and 2014. At-risk population calculated using data provided by the Australian Bureau of Statistics (total number of people residing in Western Australia over the study period in each category). Totals for each category may vary due to incomplete data at time of presentation. Unadjusted odds ratios (OR) and corresponding confidence intervals (95% CI) are displayed for each category. recorded injuries occurred at home (15%), within industrial and construction areas (4%) or at sporting and athletics venues (4%). A total of 4530 comorbid eye diagnoses secondary to other injuries were recorded. Injury of conjunctiva and corneal abrasion with no foreign body (28%), and contusion of eyeball and orbital tissues (25.4%) were common (Table 4). Patients with a comorbid eye injury (3951) were most frequently admitted due to assault (26%) or a fall (20%). Place of injury occurrence was not recorded for 33% of patients with a secondary eye injury. The majority of recorded injuries occurred at school (22%), home (16%) or on the street or highway (14%).
Assaults were the most frequent reason individuals with an eye injury were admitted to hospital (1615 cases, 24%) (Table S4, supplementary material). The frequency of eye injury caused by assault increased from 129 in 2005 (8%) to 202 (13%) in 2012. Assault-related eye injuries were more common in males, particularly those between 15 and 54 years of age (1003 cases, 62%). Indigenous females experienced assault-related eye injuries at a greater frequency compared to Indigenous males and non-Indigenous females (Figure 4). Exposure to inanimate mechanical forces was also a major cause of injury (1490 cases, 22%), followed by falls (1003 cases, 15%), particularly amongst elderly females.   Advanced age, residential status, primary eye injury and time spent in the intensive care unit were significantly associated with the length of hospital stay in the univariable model (p < 0.05). A primary eye injury diagnosis and length of admission to intensive care were significant predictors of hospital stay in the multivariable model (p < 0.05) (Table S5, supplementary material). Older patients (65 years of age and over) spent approximately 4 days longer in hospital compared to children under 16 years of age. Patients admitted with a primary eye injury spent an additional 4.8 days in hospital compared to those with a secondary injury, and each day of admission to the intensive care unit increased the total hospital stay by 2 days.
The age-adjusted incidence rate for hospitalisation due to an eye injury diagnosis was 30 per 100 000 person-years (95% CI 29-31) in the entire cohort. Indigenous patients were hospitalised for eye injuries at an adjusted rate of 109 per 100 000 person-years (95% CI 102-116) compared to 27 per 100 000 (95% CI 26-27) for non-Indigenous patients ( Figure 5). Age-adjusted hospitalisation rates for all individuals and eye injuries increased from 28 per 100 000 person years (95% CI 26-30) in 2005, to 31 (95% CI 29-33) in 2014. Each year was associated with a 5% increase (95% CI 1.03 to 1.07, p < 0.001) in the mean number of eye injuries, with age group being constant.

Discussion
The results of the present study demonstrate an increase in the incidence of eye injuries treated in WA hospitals between 2005 and 2014.
Males sustained eye injuries more frequently than females and had greater odds of injury. This is well-established in the literature. [8][9][10][19][20][21] Indigenous individuals were also more likely to experience an eye injury requiring hospitalisation beyond initial presentation to an ED. This may indicate that Indigenous individuals are sustaining more severe injuries that require on-going or specialised care and monitoring. The discrepancy between the odds of eye injury presentation and inpatient treatment within the Indigenous population may also be explained by direct admittance to hospital following ED presentation without being discharged, although this cannot be determined. Disclosure of Aboriginal and/or Torres Strait Islander heritage is self-reported and voluntary. This information was not recorded for 2553 individuals within the ED dataset, which may also account for contrasting ED and hospital admission results.
Patients over 65 years of age spent more time in hospital, which may be related to the severity of eye injuries and preexisting or comorbid conditions. Individuals with a primary eye injury were admitted for approximately 5 days longer than those with a secondary eye injury, which similarly may reflect the comparable severity of the injury. People from rural areas also experienced a longer hospital stay, potentially due to lack of access to specialist care closer to home. Rural residency has been identified as a risk factor for ocular injury and subsequent vision loss in previous studies. 19,22 Individuals residing in metropolitan areas are reported to have higher odds of sustaining an eye injury requiring ED presentation compared to hospital admission in the current study. This suggests that injuries occurring in metropolitan areas may be less severe and require outpatient treatment with potential follow-up with a primary care provider (e.g. optometrist, ophthalmologist), though this was not examined. Contusions to the eyeball and orbital tissues and injury to the cornea and conjunctiva were frequently observed in both ED and hospital admission data sets. The number of eye injuries secondary to a non-eye related diagnosis is likely due to patients being admitted for generalised head trauma. Given that the majority of known primary eye injuries admitted to hospital were caused by foreign bodies entering the eye, or the individual being struck by a falling object, and these injuries occurred at home or within construction areas, it is possible that these injuries resulted from inappropriate or a lack of use of eye protection. Previously between 24-30% and 40-60% of eye injuries in Australia occurred in the home or workplace, respectively. 8,11,19 McCarty et al. 19 reported lower rates of hospitalisation for individuals who indicated proper use of protective eyewear in the workplace compared to those who did not use any eyewear. Public health messages should identify high-risk behaviours and promote greater vigilance and the use of eye protection.
Although Indigenous individuals comprise only 4% of the WA population, they were found to disproportionately experience ocular trauma requiring hospitalisation compared to non-Indigenous individuals. This finding is similar to Smith et al., 20 where Indigenous individuals experienced a higher rate of eye injury (88.2 per 100 000 person-years compared to 11.8 per 100 000 for all recorded injuries) and accounted for 38% of eye injuries presenting to ED in northern Queensland, while only comprising 12% of the total population. While the current study did not examine the effect of ocular trauma on longterm visual outcomes, these findings are concerning as Indigenous patients experience higher rates of vision loss following injury. 22 Ongoing efforts to minimise the disparity in injury risk and access to specialist care between Indigenous and non-Indigenous communities is needed, particularly in rural and remote areas. 23 Assault was the leading cause of hospital admission in WA for known eye injuries. This is consistent with previous work in Australia; assault is a common cause of ocular injury among men (27% of presenting cases) 8 and Indigenous and non-Indigenous individuals alike (74% and 39% of cases, respectively). 21 Notably, female Indigenous patients between 10 and 39 years of age admitted to hospital due to an assaultrelated eye injury were overrepresented compared to Indigenous males and the non-Indigenous cohort as a whole. Smith et al. 20 similarly report higher rates of assault-related eye injuries in female (76%) and male (70%) Indigenous individuals. Alcohol consumption has been implicated in the incidence of assault-related eye injuries, 24,25 although the extent and effect of alcohol involvement has not been evaluated in the present study. This warrants further investigation as the implementation of preventative measures, such as the "lock out" laws introduced in Sydney during 2014 have reduced the incidence of injuries associated with alcohol and interpersonal violence. 25 While the economic and social cost of assault-related eye injury has not been calculated, it is likely to be substantial and should be explored in more detail.
Age-adjusted rates of ED presentation for an eye injury increased from 205 to 271 per 100 000 person-years from 2005 to 2014. High incidence rates of ocular trauma requiring ED attendance have been reported in regional New South Wales (between 257.5 and 449 per 100 000 person years) 11 and Victoria (950 per 100 000 person-years). 9 Eye-injury incidence rates across major metropolitan and rural hospitals across Australia are poorly documented, and further research is required to quantify associated costs and burden of injury.
Data on the incidence of eye injuries requiring hospitalisation are similarly scarce and highly varied. The age-adjusted rate of 30 per 100 000 person-years reported in the present study is double that observed by Fong 7 (15.2 per 100 000) and intermediate to rates reported by Smith et al. 20 (11.8 per 100 000) and Raymond et al. 8 (53.6 per 100 000). Variations in the reported eye injury incidence rate per state may be due to differences in data collection, coding systems and eye injury definitions over time; patient demographics; use of eye protection and whether injuries were recorded as a primary or secondary diagnosis.
This study was restricted to evaluating eye injuries managed in hospitals, meaning that less severe injuries were not included. Minor eye injuries treated in primary care settings (e.g. by a general practitioner, optometrist, or pharmacist) are not represented in either data set, so the results underestimate the true number and severity of all eye injuries that occurred during the study period. The completeness of data sets used is restricted by the quality of coding and extraction criteria, and some ED data are poorly documented in contrast to hospital admission data, which are generally more complete. For example, as a single diagnostic code was assigned to individuals attending an ED, it is possible that eye injuries occurring secondary to a non-eye related diagnosis were not captured. The underutilisation of external cause of injury coding within ED data limits opportunities to understand the context of eye injuries, and subsequently develop public health messages to reduce the occurrence and severity of injury. Improved data collection and coding practices are required, though this may be impeded by the anonymised nature of hospital data. De-identified data cannot be validated to evaluate the quality of coding, and systematic deficiencies in coding schemes used to identify eye-related conditions have previously been identified by Ng et al. 26 Further, patients attending multiple EDs or the same ED multiple times could not reliably be identified, so ED incidence rates may be overestimated. Despite the acknowledged limitations, the population-based nature of this study has important flow-on effects to the identification and care of ocular injuries in the wider Australian context as WA is representative of the nation as a whole. 27

Conclusion
This study presents seminal data on the epidemiology of ocular injury in the whole WA population over a 10-year window. The burden of eye injuries, particularly those requiring hospitalisation due to physical assault, increased between 2005 and 2014. Indigenous individuals are disproportionately overrepresented, and young Indigenous female patients are the most identifiable group to experience eye injuries. The remedial strategies required to address these findings may be more challenging than those promoted to minimise injuries sustained at home, within the workplace or while playing sport. The healthcare sector has an increasingly vital role in preventing violence and promoting safety, and new approaches to reduce the violencerelated incidence of eye injury are imperative.