Assault-related admissions to hospital in Central Australia

A perceived increase in serious assault and violence in Australia has received periodic intense media and public scrutiny. The focus is often on marginalised or minority groups, including Aboriginal people.1-3 The personal, social and community effects of serious assault are well described and may be associated with substance misuse, poverty and cultural fragmentation.4,5

National surveys of assault in Australia demonstrate a range of demographic and social differences. Of the 618 300 people over the age of 15 assaulted in a 12-month period before 1998, 54% were male.6 The Australian Capital Territory (6.9%) and the Northern Territory (6.3%) had the greatest proportions of their population assaulted, whereas the national rate was 4.3%.6 Force or violence was associated with assaults in 245 000 (39.7%) cases, and 13 300 (2.2%) assaulted people were admitted to hospital.6

Data relating to the Indigenous population are generally of poorer quality and most reports are anecdotal.7 In one study, the Indigenous age-adjusted rate of hospitalisation per 100 000 was 17 times the national rate.8 Hospitalisation resulting from interpersonal violence was generally similar for male and female Aboriginals, but the rate of hospitalisation both for unarmed fights or brawls and for attack by cutting or piercing instruments was higher for men aged 15–49 years than for women of the same age.8

The purpose of our study was to describe the number of admissions to hospital that were directly related to assault (using a six-year retrospective analysis of computerised hospital data) and to examine factors associated with these admissions.

Our data are from Alice Springs Hospital (ASH) and Tennant Creek Hospital (TCH), the only two hospitals in the Central Australia region of the NT. ASH is a 160-bed, general regional hospital, providing emergency, intensive care, and most other acute services. TCH is a 20-bed community hospital, 500 km north of Alice Springs. The catchment area for these hospitals is more than 1 million km2 and has a residential population of 47 000, of whom about 40% are Aboriginal.9

Electronic records of all admissions to ASH and TCH between 1 July 1995 and 30 June 2001 were examined, and admission with the primary cause of injury as "assault" were extracted. Demographic data were collected using the Business Objects program, a hospital-based software system. Additional information from the electronic records identified alcohol use, repeated admissions and, for ASH, the cause of the assault. Data were de-identified and grouped to preserve the anonymity of subjects and communities, as required by the Institutional Ethics Committee, and Aboriginal subcommittee, who approved our study.

In July 1998, the NT moved from ICD-9-CM to ICD-10-AM coding. The codes used for the two time periods are displayed in Box 1. Data entered into the ICD system are dependent on hospital coders examining individual patient records. The coders identify medical staff entries that refer to specific events (eg, alcohol present on admission, or blunt trauma). This information is translated into ICD codes and entered into an electronic database. At ASH, the same two hospital coders were responsible for all patient data entries for the six years of the study. These coders also monitored TCH data entry.

When patients were admitted to TCH and then transferred to ASH for a single assault-related admission, the episode was counted only once, and for TCH.

There were 2994 assault-related admissions between July 1995 and June 2001: 2449 (81.8%) at ASH and 545 (18.2%) at TCH (Box 2). People aged 25–34 years, women, and Aboriginals accounted for significantly greater proportions of assault-related admissions than expected from their proportions in the NT population in 2001 (Box 2).

Box 3 displays crude assault-related admission rates per 10 000 people at ASH and TCH combined. While the non-Aboriginal rate was constant over the study period (test for non-zero linear slope, P = 0.96), the Aboriginal rate was much higher and significantly increased at an average rate of 2.4 (SE, 0.5) per 10 000 people per year (P = 0.006). The relative ratio for Aboriginal/non-Aboriginal crude rates was 31.6 in 1995–96, increasing to 43.8 in 2000–01.

The total number of admissions due to all causes (excluding day-only cases) at the two hospitals during the study period was only accessible from July 1997. The overall proportion of hospital admissions resulting from assault significantly increased from 4.2% in 1997–98 to 4.8% in 2000–01 (P = 0.02). This increase in proportions was evident in ASH admissions, increasing from 3.8% to 4.7% (P = 0.003), but not in TCH admissions, decreasing from 6.9% to 5.0% (P = 0.05).

Box 4 provides age-adjusted assault-related admission rates per 10 000 for the two hospitals combined over the study period. Each of these rates had a significant linear increase over time: females, 2.4 (SE, 0.3) per 10 000 people per year (P = 0.001); males, 0.9 (SE, 0.3) per 10 000 people per year (P = 0.05); and overall, 1.6 (SE, 0.2) per 10 000 people per year (P = 0.002).

The proportion of assault-related admissions involving alcohol increased from 29.9% in 1995–96 to 43.0% in 2000–01 (P < 0.001); however, the rates have remained steady since 1997–98 (P = 0.92) (Box 5).

The primary causes of assault-related admissions by year for ASH (TCH information was not accessible) appear in Box 6. Prior to 1999–00, blunt trauma was the most frequent cause of these admissions, although significantly decreasing over the study period (P < 0.001). A trend of increasing use of sharp objects was evident (P < 0.001), with physical force (P = 0.25), other (P = 0.31), and not recorded (P = 0.14) remaining stable. Fire was involved in 29 (1.2%) ASH assault-related admissions and rape in 19 (0.8%). Primary cause data were missing in less than 10% of observations.

Adults aged between 25 and 44 years were most likely to be admitted to hospital for assault-related causes. A study in Victoria in 1997 found that 40% of people presenting to emergency departments because of assault were aged 25–44 years,10 somewhat lower than the 68% in our study.

Women were admitted for assault-related injuries more frequently than expected from their population frequency, a finding contrary to previous reports.4,6,8,10 In provincial Victoria, males were more likely to present to the emergency department because of assault.10 In the NT, males comprised 53% of the population from 1979 to 1995, but made up 77% of the violence-related deaths, while a third of all injury-related deaths among Aboriginal women in the NT are due to homicide.11 There are several possible reasons why women appeared to be more at risk of assault in our study than in previous ones. The Central Australian population, with 40% Aboriginals, is likely to be quite different to populations in other studies. Alcohol consumption is high in this population, and this might have an influence.12,13 Finally, our data do not include individuals who were assaulted but not admitted to hospital, or those who died before hospitalisation.

It is striking that the proportion of hospital admissions due to assault fell at TCH while it rose at ASH. This warrants further consideration.

Aboriginals greatly outnumbered non-Aboriginals, comprising more than 90% of the admissions, significantly higher than the 40% they represent in the Central Australian population. Although the assault-related admissions per 10 000 people is only an estimate and is based on the predicted population at one point in time, it does support the proposition that Aboriginals make up a disproportionate number of hospital admissions for assault. Our study shows that serious assault resulting in admission to hospital is of real concern in the Aboriginal population, especially among women.

The association between alcohol use and violence is well known.14,15 Importantly, our data demonstrate a significant increase over time in the proportion of assault admissions associated with alcohol. Alcohol is the overriding substance-misuse issue of concern for people in Central Australia. Consensus as to how to address this problem has been difficult to achieve, and the problem continues to escalate.13 Many resources have been developed to assist healthcare workers, communities and individuals with alcohol and violence problems in the NT.16-19 That the rate of alcohol-related assault admissions has stopped increasing since 1997–98 may reflect positive effects of these health interventions. However, this pattern could also be explained, in part, by the adoption of ICD-10-AM codes for alcohol use in July 1998. The reasons for the levelling since 1997–98 need to be clarified.

Most assaults involved blunt objects; however, a substantial rise in the use of sharp objects is evident since 1999–00.

Our study uses "admissions to hospital" data. These data are useful because:

• hospital admission data generally exclude more trivial assaults and injuries and are determined clinically by a qualified healthcare professional;

• presentations to the emergency department can be skewed by the availability of general practice services and other primary healthcare services, which can vary significantly over time;

• alternatives such as self-report surveys contain inherent biases and are not suited to a large proportion of the Central Australian community, who are itinerant, poorly literate and "non-compliant" with such approaches.

Limitations of secondary data are well known and apply to our data20 — charting and coding may contain inaccuracies, be incomplete and be subject to coding bias,21-23 although false-positive comorbidities in the Australian population have been shown to be low.24

Despite these limitations, the data reveal major potential problems, suggesting the need for more rigorous, prospective investigation. The trend of increasing assault-related admissions is likely to continue if new measures are not implemented to curtail it. Clearly, assault-related admissions to hospital in the proportions described here suggest a significant public health problem that requires attention.

3: Crude assault-related admission rates at Alice Springs Hospital and Tennant Creek Hospital, July 1995 to June 2001