Inpatient Statistics Collection Research Articles

Preventing Q fever endocarditis: a review of cardiac assessment in hospitalised Q fever patients

Acute Q fever is an important zoonotic disease in some parts of rural Australia. Q Fever can lead to chronic disease such as endocarditis, this complication occurring more commonly in patients with underlying heart valve pathology or an impaired immune system. Untreated Q fever endocarditis has a high mortality rate, but even with appropriate therapy, 10% of patients will die. Cardiac assessment can identify patients at risk. The aim of this review was to examine recorded cardiac assessment of hospitalised Q fever patients within the regional area of Hunter New England (HNE), New South Wales (NSW). Medical records of patients with Q fever admitted to hospitals in HNE during the period 2005-2009 were identified through the NSW Notifiable Diseases Database and the NSW Inpatient Statistics Collection. A standardised medical record review tool was used to undertake the review. Eighty-nine records were reviewed. Over 50% of patients were admitted to a district hospital staffed by local GPs. Cardiac assessment was not routinely documented and for 91% there was no record of a cardiac history being taken. Approximately 25% had no record of a cardiac physical examination and only 6 cases had a record of a complete cardiac examination. Q Fever remains an important disease in some parts of rural Australia. Q Fever endocarditis is a serious sequel to acute Q fever and underlying heart valve pathology. Due to its indolent progression and poor outcome when diagnosis is delayed, a thorough cardiac assessment of all patients with suspected or confirmed Q fever is important. The level of documentation of cardiac assessment for Q fever patients is of concern because it may indicate cardiac assessments were not performed. General practitioners, especially in rural and regional areas, are encouraged to conduct cardiac assessments for all patients with acute Q fever to identify patients at risk of developing Q fever endocarditis.

Trends of spinal fusion surgery in Australia: 1997 to 2006

This study aims to explore the trend in spine fusion surgery in Australia over the past 10 years, and to explore the possible influence of health insurance status (private versus public) on the rate of surgery. Data pertaining to the rate of lumbar spine fusion from 1997 to 2006 were collected from Inpatient Statistics Collection of NSW Health, Medicare Australia Statistics and the Australian Bureau of Statistics. Data on total hip and total knee arthroplasties were collected to provide a comparator. The number of publicly performed spinal fusion procedures increased by 2% from 1997 to 2006. In comparison, privately performed spinal fusion procedures increased by 167% over the same 10-year period. In 2006, spine fusion surgery was 10.8 times more likely to be done in the private sector than in the public sector, compared with corresponding figures of 4.2 times and 3.0 times for knee replacement and hip replacement, respectively. Waiting list data showed no increase in demand for spine fusion surgery in the public sector. There is a disproportionately high rate of lumbar spine fusion surgery performed in the private sector, given the rate of private insurance. The rate of increase was found to be higher than that for hip or knee arthroplasty procedures. Possible explanations for this difference include: over-servicing in the private sector, under-servicing in the public sector, differences in medical referral patterns, surgeon and patient preferences and financial incentives.

Comparison of methods to correct the miscounting of multiple episodes of care when estimating the incidence of hospitalised injury in child motor vehicle passengers

This study evaluates the performance of different case selection criteria to account for multiple episodes of care when estimating the hospitalisation rate due to road trauma amongst children. The internally linked NSW Inpatient Statistics Collection (ISC) dataset for the period between 1st July, 2000 and 30th June, 2003 was used to identify the “single” episode of care for each hospitalised child motor vehicle passenger residing in NSW. We used two hospitalised injury definitions of a case based on (1) all-diagnoses and (2) principal diagnosis only. We then developed case selection criteria, based on (a) linkage methods only available from linked ISC datasets; (b) selected variables available in both the linked and unlinked ISC datasets, to exclude repeat episodes of care for an injury. Changes in the estimated hospitalisation rate, and sensitivity and specificity, were calculated for each selection criteria compared to the findings from linkage methods as the “gold standard”. None of the correction methods for multiple episodes of care was clearly superior in terms of incidence estimation, sensitivity, and specificity concurrently. However, the correction criterion which is optimal may vary depending on different study objectives and different types of hospitalised injuries.

Admission to hospital for effects of heat and light: NSW, 1993-94 to 2003-04

The study examined the hospital admission rates and characteristics of patients experiencing severe heat-related morbidity in NSW using data from the NSW Health Inpatient Statistics Collection. The study covered the 11-year period from July 1993 to June 2004. ICD-10-AM. codes examined included T67 (effects of heat and light). There was an average of 91 admissions for each year due to a principal diagnosis of the effects of heat and light, with consistently more males than females admitted (1.7 : 1). Many of the admissions (39%) were of people 65 years of age or older. Most admissions (49%) occurred in the summer months of December and January.

Trends in hospitalisation rates for road traffic injuries in child motor vehicle passengers in New South Wales, July 1998 – June 2005

To analyse changes in the incidence of injuries requiring hospitalisation for child passengers in motor vehicle crashes. Population-based study of children (aged 0 -15 years) residing in New South Wales and admitted to hospital for injuries resulting from a traffic crash in the period 1 July 1998 - 30 June 2005, identified from the NSW Inpatient Statistics Collection. Age-standardised rates of hospitalisation for injuries, and trends by inpatient demographics, severity of injuries, and injury sites and types. 2297 children were hospitalised for injuries sustained in a motor vehicle crash over the study period. The overall hospitalisation rate for injuries was relatively constant, with a non-significant decline of - 0.4% (95% CI, - 3.1% to 2.3%). The rate of hospitalisation for serious injuries also declined non-significantly (- 5.5% [95% CI, -11.8% to 1.1%]). Only hospitalisation rates for traumatic brain injuries declined significantly (-11.1% [95% CI, -19.0% to - 2.8%]) over the study period. The rate of hospitalisation for injuries to NSW-resident child motor vehicle passengers due to traffic crashes has not significantly decreased. High hospitalisation rates and the subsequent burden to the community and public health system make further injury prevention efforts for child motor vehicle passengers a priority.

Differences in injury rates in child motor vehicle passengers in rural and urban areas in New South Wales, July 2000 to June 2004

To investigate whether the pattern of hospitalised injuries in injured child motor vehicle passengers involved in traffic crashes differs in rural and urban residents of New South Wales (NSW). This study compared injuries of hospitalised child motor vehicle passengers resident in rural areas with those from urban areas. The NSW Inpatient Statistics Collection (ISC), a population-based dataset, was used to select cases for the period of July 2000 to June 2004. The hospitalised injury rate was calculated according to urban/rural status using Poisson regression. Injury rate ratios (IRR) comparing rural and urban children were computed overall and for specific injury types. Overall, 1,286 children (aged 0-15 years) residing in NSW were identified from the NSW ISC internally linked datasets as being separated from hospital for injuries resulting from a motor vehicle crash. The overall hospitalised injury incidence rates for child motor vehicle passengers resident in rural and urban NSW areas were 46.75 (95% CI 36.63-59.66) and 20.13 (95% CI 17.94-22.58) per 100,000 children respectively. The rural/urban IRR for comparing the incidence of hospitalisation was significantly elevated (IRR=2.10, 95% CI 1.78-2.48). The IRR was also significantly elevated across most injury types. The largest risk disparity between rural and urban children was in 9-12 year-olds (IRR=2.33, 95% CI 1.73-3.13). There is an elevated injury incidence rate in rural resident children, compared with their urban counterparts. This differential should be addressed in future road safety initiatives.

A discussion of the potential benefits to injury surveillance through inclusion of date of injury in hospitalisation data in New South Wales and Australia

Most hospitalised injury cases have sudden onset at a specific time and date, due to a well-defined external cause (eg, a road crash). Date of injury is not presently recorded in routine hospital separations data in NSW or Australia, though it is in New Zealand. Benefits of adding date of injury to the Inpatient Statistics Collection would include: more accurate estimation of the population incidence of serious injury; better assessment of the health system utilisation and costs attributable to injurious events; and better linkage of hospital data with other data relevant to injury measurement and control (eg, road crash data).

Methadone in pregnancy: treatment retention and neonatal outcomes

To examine the association between retention in methadone treatment during pregnancy and key neonatal outcomes. Client data from the New South Wales Pharmaceutical Drugs of Addiction System was linked to birth information from the NSW Midwives Data Collection and the NSW Inpatient Statistics Collection from 1992 to 2002. Obstetric and perinatal characteristics of women who were retained continuously on methadone maintenance throughout their pregnancy were compared to those who entered late in their pregnancies (less than 6 months prior to birth) and those whose last treatment episode ended at least 1 year prior to birth. There were 2993 births to women recorded as being on methadone at delivery, increasing from 62 in 1992 to 459 births in 2002. Compared to mothers who were maintained continuously on methadone throughout their pregnancy, those who entered treatment late also presented later to antenatal services, were more likely to arrive at hospital for delivery unbooked, were more often unmarried, indigenous and smoked more heavily. A higher proportion of neonates born to late entrants were born at less than 37 weeks gestation and were admitted to special care nursery more often. Continuous methadone treatment during pregnancy is associated with earlier antenatal care and improved neonatal outcomes. Innovative techniques for early engagement in methadone treatment by pregnant heroin using women or those planning to become pregnant should be identified and implemented.

Spatio-temporal Analysis of the Impact of Socioeconomic Status on Admissions for Acute Myocardial Infarction and Related Procedures, in New South Wales, Australia

P-385 Introduction: Our objective was to use spatio-temporal models to examine the relationship between socio-economic status and hospital admissions for heart disease and related procedures. We also quantified the effect of including other variables such as remoteness, Sydney postal areas and Indigenous status. Methods: The main source of outcome data was the New South Wales (NSW) Inpatient Statistics Collection, which is a census of all hospital admissions. Estimated population size from the Australian Bureau of Statistics (ABS) 2001 census, was used to calculate indirect sex and age standardised counts of the outcomes at the postal-area level for the period 1996–2002. Data on Index of Relative Social Disadvantage (IRSD), Accessibility and Remoteness Index for Australia (ARIA-plus), and proportion of Aboriginal & Torres Strait Islander (ATSI) data was also obtained from the ABS. We applied the Bernardinelli spatio-temporal model in our analysis, where both the area-specific intercepts and temporal trends were modelled as random effects. Results: We found an exposure-response relationship between IRSD and acute myocardial infarction (AMI) admissions. i.e. those living in socio-economically disadvantaged postal areas were more likely to be admitted for AMI. We also found that including ARIAP-plus to IRSD actually resulted in a lower percentage of spatial correlation (from 85% to 78%), indicating that accessibility and remoteness does seem to explain away some of the spatial variation in the admissions data, after adjusting for socio-economic status. There was no overall linear temporal increase in relative risks from 1996 to 2002, but we did find a small and significant interaction random effect, indicating that there could be some regions showing a temporal trend over time. We also examined the spatio-temporal distribution of related procedures such cardiac catheterisations, Coronary Artery Bypass Grafts and stents. These demonstrated an inverse relationship with disadvantage and, in some regions, this disparity appeared to worsen rather than improve over the study period. Discussion: We found strong spatially correlated effects in the risk estimates of heart disease and related procedures hospital admissions in NSW, Australia from 1996 to 2002. While low socio-economic status was associated with higher rates of admissions for AMI, it was also associated with low intervention rates. This disparity appears to be increasing in some regions.

The use of record linkage to examine illicit drug use in pregnancy

To examine the obstetric and perinatal outcomes for women with a drug-related hospital admission during pregnancy. Antenatal and birth admissions to New South Wales (NSW) hospitals from the NSW Inpatient Statistics Collection were linked to birth information from the NSW Midwives Data Collection over a 5-year period (1998-2002). Birth admissions were flagged as positive for drug use where a birth admission or any pregnancy admission for that birth involved an opioid, cannabis or stimulant-related International Classification of Diseases version 10 Australian modification (ICD-10AM) code. A total of 416 834 live births were analysed over a 5-year period (1998-2002). Of these, 1974 pregnancies had an opioid ICD-10AM diagnosis recorded, 552 a stimulant-related ICD-10AM and 2172 a cannabis ICD-10AM diagnosis. Births in each of the drug groups were to women who were younger, had a higher number of previous pregnancies, were indigenous, smoked heavily and were not privately insured. These women also presented later in their pregnancy to antenatal services and were more likely to arrive at hospital unbooked. Neonates born to women in each of the drug groups were more likely to be premature and were admitted to neonatal intensive care and special care nursery more often, with neonates born to women in the opioid group admitted most often. Linked population level administrative data is a powerful method for examining the maternal and neonatal outcomes associated with the use of specific illicit drugs during pregnancy.

Use of Record Linkage to Examine Alcohol Use in Pregnancy

To date, no population-level data have been published examining the obstetric and neonatal outcomes for women with an alcohol-related hospital admission during pregnancy compared with the general obstetric population. This information is critical to planning and implementing appropriate services. Antenatal and delivery admissions to New South Wales (NSW) hospitals from the NSW Inpatient Statistics Collection were linked to birth information from the NSW Midwives Data Collection over a 5-year period (1998-2002). Birth admissions were flagged as positive for maternal alcohol use where a birth admission or any pregnancy admission for that birth involved an alcohol-related International Statistical Classification of Diseases and Related Health Problems, 10th Revision, Australian Modification (ICD-10-AM) code. Key demographic, obstetric, and neonatal variables were compared for births to mothers in the alcohol group with births where no alcohol-related ICD10-AM was recorded. A total of 416,834 birth records were analyzed over a 5-year period (1998-2002). In this time, 342 of these were coded as positive for at least 1 alcohol-related ICD-10-AM diagnosis. Mothers in the alcohol group had a higher number of previous pregnancies, smoked more heavily, were not privately insured, and were more often indigenous. They also presented later on in their pregnancy to antenatal services and were more likely to arrive at hospital unbooked for delivery. Deliveries involved less epidural and local and more general anesthesia. Cesarean sections were more common to women in the alcohol group and were performed more often for intrauterine growth retardation. Neonates born to women in the alcohol group were smaller for gestational age, had lower Apgar scores at 5 minutes, and were admitted to special care nursery more often. This study shows that linked population-level administrative data provide a powerful new source of information for examining the maternal and neonatal outcomes associated with alcohol use in pregnancy.

Patterns of morbidity and rehospitalisation following spinal cord injury.

Longitudinal, descriptive design. The aim of this study was to investigate the frequency, cause and duration of rehospitalisations in individuals with spinal cord injury (SCI) living in the community. Australian spinal cord injury unit in collaboration with State Health Department. A data set was created by linking records from the NSW Department of Health Inpatient Statistics Collection between 1989-1990 and 1999-2000 with data from the Royal North Shore Hospital (RNSH) Spinal Cord Injuries Database using probabilistic record linkage techniques. Records excluded were nontraumatic injuries, age <16 years, spinal column injury without neurological deficit, full recovery (ASIA Grade E) and index admission not at RNSH. Descriptive statistics and time to readmission using survival analysis, stratified by ASIA impairment grade, were calculated. Over the 10-year period, 253 persons (58.6%) required one or more spinal-related readmissions, accounting for 977 rehospitalisations and 15,127 bed-days (average length of stay (ALOS) 15.5 days; median 5 days). The most frequent causes for rehospitalisation were genitourinary (24.1% of readmissions), gastrointestinal (11.0%), further rehabilitation (11.0%), skin-related (8.9%), musculoskeletal (8.6%) and psychiatric disorders (6.8%). Pressure sores accounted for only 6.6% of all readmissions, however, contributed a disproportionate number of bed-days (27.9%), with an ALOS of 65.9 (median 49) days and over 50% of readmissions (33 out of 64) occurred in only nine individuals aged under 30 years. Age, level and completeness of neurological impairment, all influenced differential rates of readmission depending on the type of complication. Overall rehospitalisation rates were high in the first 4 years after initial treatment episode, averaging 0.64 readmissions (12.6 bed-days) per person at risk in the first year and fluctuating between 0.52 and 0.61 readmissions (5.1-8.3 bed-days) per person at risk per year between the second to fourth years, before trending downwards to reach 0.35 readmissions (2.0 bed-days) as 10th year approaches. Time to readmission was influenced by degree of impairment, with significantly fewer people readmitted for ASIA D (43.2%) versus ASIA A, B and C (55.2-67.0%) impairments (P<0.0001). The mean duration to first readmission was 46 months overall, however, differed significantly between persons with ASIA A-C impairments (26-36 months) and ASIA D impairment (60 months). Identifying rates, causes and patterns of morbidity is important for future resource allocation and targeting preventative measures. For instance, the late complication of pressure sores in a small subgroup of young males, consuming disproportionately large resources, warrants further research to better understand the complex psychosocial and environmental factors involved and to develop effective countermeasures.

Should There Be an Age Split for Stroke DRGs? Analysing a Large Clinical Data Set of a Principal Teaching Hospital over a Five-Year Period

The aim of this study was to analyse the inpatient statistics collection relating to stroke patients admitted to a major teaching hospital, with particular reference to length of stay, and to assess the adequacy of the diagnosis related group (DRG) as a predictor of length of stay. The study subjects were selected by DRG to identify all stroke inpatients admitted and discharged between 1 July 1995 and 30 June 2000. There were 1365 stroke discharges (half of whom were over 75 years of age at discharge) over the period of the study. The median length of stay was 8 days, and 67% of the subjects experienced complications and/or comorbidities. Age was significantly associated with increased length of stay of stroke patients, independent of complications or comorbidities. These findings raise the question of whether casemix-based funding should be based solely on DRGs for complicated conditions such as stroke, or whether additional measures such as age should be used for funding allocation. This study provides a model that health information managers and other researchers could use to analyse inpatient statistics collections at state, territory or national levels.

Neonatal surgery in New South Wales—What is performed where?

Objectives The aim of this study was to describe what surgical procedures are performed in the neonatal period in New South Wales (NSW) and where they are performed. Methods Population-based descriptive study was conducted in NSW in a 2-year period from July 1, 1996 to June 30, 1998, inclusive, using information from the NSW Health Department’s Inpatient Statistics Collection. All neonates undergoing major surgery (excluding circumcisions) in NSW. Results In the first 4 weeks of life, 990 (0.6%) neonates underwent surgery. The most common surgical procedures were gastrointestinal, cardiovascular, hernia, genitourinary, and neurosurgical. Frenotomy accounted for 5% of all surgical procedures. Whereas 75% of neonatal surgery (including 88% of gastrointestinal and 97% cardiovascular surgery) occurs in children’s hospitals, only 13% of the babies requiring surgery are born in the co-located obstetric hospitals. Perinatal centers accounted for 5.3% of surgery; urban hospitals for 8.4%; rural hospitals, 5.5%, and private hospitals, 6.4%. The mortality rate in the neonatal period was 3.0% overall. Conclusions This is the first review of major neonatal surgery in Australia and provides baseline data for future comparisons. Whereas most neonates had surgery in a children’s hospital, few of them were born in the most appropriate place, the co-located obstetric hospital. Parents should be informed of the level of institutional surgical expertise and be involved in the decision-making regarding the place of surgery for their infant. Parents and children have a right to expect the best possible results.

The epidemiology of cardiac arrests in a Sydney hospital

Aim: To examine the epidemiology of cardiac arrest (CA) in New South Wales (NSW), Australia, and a large teaching hospital in Sydney and to identify predictors of survival. Methods: Data from the 1996/97 NSW inpatient statistics collection were analysed. Logistic regression was used to determine predictors of mortality from CA. Results: In 1996/97 in public hospitals, Eastern Sydney Area Health Service (AHS) (47.5), and in private hospitals of Macleay-Hastings District Health Services (DHSs), recorded the highest crude CA rates in NSW with figures of 47.5 and 21.3 per 10,000 hospital separations. Standardised CA rates were highest in Western Sydney, Illawara, Macleay-Hasting, Mid North Coast and Orana DHSs. Most CAs in hospital ‘X’ occurred after admission and 55.7% occurred in patients admitted with a non-cardiac principal diagnosis, mainly pneumonia. Acute myocardial infarction was the leading diagnosis (28.9%) in patients who suffered CA after admission. Only 32% of CA patients survived to discharge. Age above 65 (odds ratio (OR)=2.284, P=0.006) had the highest effect on cardiac mortality. The longer the patients were in the intensive care unit (OR=0.997, P=0.037), the more likely they were to survive. Conclusion: We describe considerable variation in CA rates in NSW. The majority of hospital CAs occur at or soon after admission, and are associated with underlying non-cardiac conditions. Survival following hospital CA is low, but there may be preventable elements. Knowledge of risk factors and epidemiology of hospital CAs may help identify patients at risk of CA.

Antenatal transfer of rural women: how does the NSW Inpatient Statistics Collection compare with an audit of hospital records?

The aim of this study was to determine whether the NSW Inpatient Statistics Collection (ISC), a census of hospital admissions, could be used to estimate the magnitude of, and reasons for, antenatal transfer of rural women. Data from the ISC were compared with results of a clinical audit of all antenatal admissions of rural women to perinatal centres in NSW during 1997-1998. While the overall number of perinatal centre admissions identified by the ISC and the audit were similar, the ISC identified only about 70% of antenatal transfers. Rural hospitals identified 12% of women as indigenous compared with 9% at perinatal centres. The ISC showed 28% of rural women admissions and 42% of transfers were for threatened preterm labour compared with 21% and 30% respectively from the audit.

Antenatal hospitalisations in New South Wales, 1995-96.

To provide a population-based estimate of the prevalence of antenatal hospitalisations and to determine the reasons for admission. Descriptive study. Data sources were primarily the New South Wales inpatient Statistics Collection (ISC) and also the linked population-based New South Wales Midwives Data Collection. All women resident in New South Wales (NSW) admitted to public and private hospitals in NSW for pregnancy complications from 1 July 1995 to 30 June 1996. Antenatal hospitalisations for pregnancy complications per 100 confinements. There were a total of 25,710 antenatal non-delivery admissions in NSW hospitals among 86,263 confinements, yielding a ratio of 30 antenatal admissions per 100 confinements (21 per 100 excluding day-stay admissions). Women without private health insurance, Aboriginal women and women under the age of 20 had significantly higher rates of antenatal admissions. The principal admitting diagnoses were hypertension (17.6%), threatened preterm labour (14.1%), antepartum haemorrhage (14.0%), threatened labour (after 37 weeks' gestation) (11.1%) and excessive vomiting in pregnancy (9.4%). A majority (58%) of antenatal admissions (including admissions to day-stay facilities) were for one day. Significant pregnancy-related morbidity, as measured by high ratios of antenatal admissions, is evident in NSW. Sociodemographic factors and health service management protocols appear to account for some of the variability in antenatal admissions.

Evaluation of linked cancer registry and hospital records of breast cancer

BackgroundInformation on the treatment of women with breast cancer in Australia is generally available only from special surveys. Analysis of routinely collected datasets may be more timely and cost effective, if the data are sufficiently accurate and complete. ObjectiveTo evaluate the accuracy and completeness of data on treatment in linked records of breast cancer from two routinely collected datasets. MethodsThe NSW Department of Health linked NSW Central Cancer Registry (CCR) records for 2,636 women diagnosed with breast cancer in NSW in 1992 to all hospital admission records in the NSW In-patient Statistics Collection (ISC) from January 1991 to June 1994. We queried the original paper records of subsets of women to identify missing or miscoded information and cases not notified to the CCR. We also compared the treatment data with data collected independently from the medical records of 19% of the women. ResultsISC records linked to 89% of the CCR records. The CCR had identified 94.9% of women with breast cancer treated as hospital in-patients and 83% of these women had surgical treatment recorded in the ISC. The linked dataset under-estimated the percentage of women having breast-conserving therapy (-4%) and slightly over-estimated the percentage having mastectomy (+1%).We estimated that 42% of women treated surgically for breast cancer had actually had breast-conserving surgery, compared with 39% in the original dataset. There was no evident bias by age or by urban or rural residence in the under-recording of breast conservation. There was 94% agreement on the type of surgery between the linked dataset and the independent dataset.

Clonidine overdose in childhood: implications of increased prescribing.

To highlight the increase in the number of cases of clonidine overdose admitted to a specialist paediatric hospital, with particular reference to the clinical features, clinical course and circumstances surrounding the incident. Cases of clonidine overdose were identified by review of the emergency department attendance register, the intensive care unit database and inpatient statistics collection. Case notes were reviewed to determine the clinical features, history and clinical course in each case. Fifteen patients experienced 16 overdoses during the period 1990-97 inclusive. Only one case occurred before 1994. Depressed level of consciousness and bradycardia were the most common clinical manifestations, and were observed in 75 and 88% of cases respectively. There were no fatalities. Five patients received naloxone. Other treatment modalities included gastrointestinal decontamination, atropine, ventilation and inotropic support. Fourteen cases occurred in association with medication prescribed for attention-deficit hyperactivity disorder (ADHD). Clonidine overdose is a potentially serious condition, often requiring intensive care management. Our experience suggests that it is a growing problem, related in part to its increased use in the treatment of ADHD. Preventive strategies, including raising the level of awareness of risks, changes to packaging and appropriate selection of patients for treatment, need consideration if further overdoses are to be prevented.

Surgical treatment of breast cancer in New South Wales 1991, 1992.

The purpose of this study was to examine existing data on women diagnosed with breast cancer in New South Wales in 1991 and 1992 and to describe surgical treatments received on an inpatient basis. Analyses were based on the linkage of two databases: the New South Wales Central Cancer Registry and the New South Wales Health Department's Inpatient Statistics Collection. Data were limited to women who were resident and treated in New South Wales. Main analyses were restricted to definitive surgical procedures. Thirty-six per cent of women treated surgically for breast cancer in 1991 had breast-conserving therapy. This had increased to 39% in 1992. There were substantial geographical variations in the use of breast-conserving therapy in New South Wales which could not be explained by patient characteristics. Age, degree of spread at diagnosis, and area health service/ health region were all found to have an independent association with the probability of having a mastectomy. Women with a localized degree of spread living in non-metropolitan areas (Health Regions) were almost twice as likely to have a mastectomy as compared with similar women who were resident in metropolitan areas (Area Health Services). The concentration of radiotherapy services may have contributed to the urban/rural variation in breast-conserving therapy in New South Wales, but it is also likely that some of the variations that were observed may be a reflection of the failure of clinicians to use best current practice.