Search

Daily counts of computer records of hospital emergency department arrivals grouped according to diagnosis (called here syndrome groupings) can be monitored by epidemiologists for changes in frequency that could provide early warning of bioterrorism events or naturally occurring disease outbreaks and epidemics. This type of public health surveillance is sometimes called syndromic surveillance. We used transitional Poisson regression models to obtain one-day-ahead arrival forecasts. Regression parameter estimates and forecasts were updated for each day using the latest 365 days of data. The resulting time series of recursive estimates of parameters such as the amplitude and location of the seasonal peaks as well as the one-day-ahead forecasts and forecast errors can be monitored to understand changes in epidemiology of each syndrome grouping. The counts for each syndrome grouping were autocorrelated and non-homogeneous Poisson. As such, the main methodological contribution of the article is the adaptation of Cumulative Sum (CUSUM) and Exponentially Weighted Moving Average (EWMA) plans for monitoring non-homogeneous counts. These plans were valid for small counts where the assumption of normally distributed one-day-ahead forecasts errors, typically used in other papers, breaks down. In addition, these adaptive plans have the advantage that control limits do not have to be trained for different syndrome groupings or aggregations of emergency departments. Conventional methods for signaling increases in syndrome grouping counts, Shewhart, CUSUM, and EWMA control charts of the standardized forecast errors were also examined. Shewhart charts were, at times, insensitive to shifts of interest. CUSUM and EWMA charts were only reasonable for large counts. We illustrate our methods with respiratory, influenza, diarrhea, and abdominal pain syndrome groupings. Keywords: Control charts, CUSUM, EWMA, monitoring, Poisson regression, Shewhart charts, epidemiology, 1. Introduction This article develops flexible methods to obtain early warning of disease outbreaks by prospectively monitoring data streams collected from computer information systems used for routine patient management at [...]

Objectives: To investigate individual factors associated with an asthma outbreak among children aged one to 14 years in Sydney in February 1999. Methods: A case control study was undertaken with cases (n=92) defined as all children admitted to Sydney Children's Hospital for asthma in February 1999. Unmatched controls (n=76) were all children admitted for asthma in the previous three months. We obtained information by a structured telephone survey of parents. Logistic regression analyses were used to determine odds ratios for risk factors for hospital admission. Results: Mean age for hospital admission of 4.7 years for cases and 4.4 years for controls. The presence of one or more siblings reduced the risk of admission during an asthma outbreak (OR=0.59, 95% CI 0.37 to 0.93). Children with older siblings aged 10 to 14 years were also less likely to be admitted (OR=0.3, 95% CI 0.12 to 0.74). An age effect was observed. Other demographic, clinical and environmental characteristics, including smoking, were not associated with admission during the outbreak. Conclusions: The main findings of this study are the protective effect of siblings and an age-dependent effect in risk of hospital admission during an asthma outbreak. These findings are consistent with an infective cause of the outbreak. Implications: Children without siblings, particularly older siblings, appear to be at highest risk of hospital admission during an asthma outbreak. Environmental and other factors need to be examined to further explain the episodicity of such outbreaks and to determine means of predicting and preventing future episodes.

Abstract Background In a climate of concern over bioterrorism threats and emergent diseases, public health authorities are trialling more timely surveillance systems. The 2003 Rugby World Cup (RWC) provided an opportunity to test the viability of a near real-time syndromic surveillance system in metropolitan Sydney, Australia. We describe the development and early results of this largely automated system that used data routinely collected in Emergency Departments (EDs). Methods Twelve of 49 EDs in the Sydney metropolitan area automatically transmitted surveillance data from their existing information systems to a central database in near real-time. Information captured for each ED visit included patient demographic details, presenting problem and nursing assessment entered as free-text at triage time, physician-assigned provisional diagnosis codes, and status at departure from the ED. Both diagnoses from the EDs and triage text were used to assign syndrome categories. The text information was automatically classified into one or more of 26 syndrome categories using automated "naïve Bayes" text categorisation techniques. Automated processes were used to analyse both diagnosis and free text-based syndrome data and to produce web-based statistical summaries for daily review. An adjusted cumulative sum (cusum) was used to assess the statistical significance of trends. Results During the RWC the system did not identify any major public health threats associated with the tournament, mass gatherings or the influx of visitors. This was consistent with evidence from other sources, although two known outbreaks were already in progress before the tournament. Limited baseline in early monitoring prevented the system from automatically identifying these ongoing outbreaks. Data capture was invisible to clinical staff in EDs and did not add to their workload. Conclusion We have demonstrated the feasibility and potential utility of syndromic surveillance using routinely collected data from ED information systems. Key features of our system are its nil impact on clinical staff, and its use of statistical methods to assign syndrome categories based on clinical free text information. The system is ongoing, and has expanded to cover 30 EDs. Results of formal evaluations of both the technical efficiency and the public health impacts of the system will be described subsequently.

Abstract Background We conducted a retrospective study to determine if potential past exposure to dioxin had resulted in increased incidence of cancer in people living near a former manufacturing plant in New South Wales, Australia. During operation, from 1928 to 1970, by-products of the manufacturing process, including dioxin and other chemical waste, were dumped into wetlands and mangroves, discharged into a nearby bay and used to reclaim land along the foreshore, leaving a legacy of significant dioxin contamination. Methods We selected 20 Census Collector Districts within 1.5 kilometres of the former manufacturing plant as the study area. We obtained data on all cases of cancer and deaths from cancer in New South Wales from 1972 to 2001. We also compared rates for some cancer types that have been associated with dioxin exposure. Based on a person's residential address at time of cancer diagnosis, or at time of death due to cancer, various geo-coding software and processes were used to determine which collector district the case or death should be attributed to. Age and sex specific population data were used to calculate standardised incidence ratios and standardised mortality ratios, to compare the study area to two comparison areas, using indirect standardisation. Results During the 30-year study period 1,106 cases of cancer and 524 deaths due to cancer were identified in the study area. This corresponds to an age-sex standardised rate of 3.2 cases per 1,000 person-years exposed and 1.6 deaths per 1,000 person-years exposed. The study area had a lower rate of cancer and deaths from cancer than the comparison areas. The case incidence and mortality due to lung and bronchus carcinomas and haematopoietic cancers did not differ significantly from the comparison areas for the study period. There was no obvious geographical trend in ratios when comparing individual collector districts to New South Wales according to distance from the potential source of dioxin exposure. Conclusion This investigation found no evidence that dioxin contamination from this site resulted in increased cancer rates in the potentially exposed population living around the former manufacturing plant.