Periods of successive extreme heat and cold temperature have major effects on human health and increase rates of health service utilisation. The severity of these events varies between geographic locations and populations. This study aimed to estimate the effects of heat waves and cold waves on health service utilisation across urban, regional and remote areas in New South Wales (NSW), Australia, during the 10-year study period 2005-2015. We divided the state into three regions and used 24 over-dispersed or zero-inflated Poisson time-series regression models to estimate the effect of heat waves and cold waves, of three levels of severity, on the rates of ambulance call-outs, emergency department (ED) presentations and mortality. We defined heat waves and cold waves using excess heat factor (EHF) and excess cold factor (ECF) metrics, respectively. Heat waves generally resulted in increased rates of ambulance call-outs, ED presentations and mortality across the three regions and the entire state. For all of NSW, very intense heat waves resulted in an increase of 10.8% (95% confidence interval (CI) 4.5, 17.4%) in mortality, 3.4% (95% CI 0.8, 7.8%) in ED presentations and 10.9% (95% CI 7.7, 14.2%) in ambulance call-outs. Cold waves were shown to have significant effects on ED presentations (9.3% increase for intense events, 95% CI 8.0-10.6%) and mortality (8.8% increase for intense events, 95% CI 2.1-15.9%) in outer regional and remote areas. There was little evidence for an effect from cold waves on health service utilisation in major cities and inner regional areas. Heat waves have a large impact on health service utilisation in NSW in both urban and rural settings. Cold waves also have significant effects in outer regional and remote areas. EHF is a good predictor of health service utilisation for heat waves, although service needs may differ between urban and rural areas.

BackgroundDeath certificates provide an invaluable source for mortality statistics which can be used for surveillance and early warnings of increases in disease activity and to support the development and monitoring of prevention or response strategies. However, their value can be realised only if accurate, quantitative data can be extracted from death certificates, an aim hampered by both the volume and variable nature of certificates written in natural language. This study aims to develop a set of machine learning and rule-based methods to automatically classify death certificates according to four high impact diseases of interest: diabetes, influenza, pneumonia and HIV.MethodsTwo classification methods are presented: i) a machine learning approach, where detailed features (terms, term n-grams and SNOMED CT concepts) are extracted from death certificates and used to train a set of supervised machine learning models (Support Vector Machines); and ii) a set of keyword-matching rules. These methods were used to identify the presence of diabetes, influenza, pneumonia and HIV in a death certificate. An empirical evaluation was conducted using 340,142 death certificates, divided between training and test sets, covering deaths from 2000–2007 in New South Wales, Australia. Precision and recall (positive predictive value and sensitivity) were used as evaluation measures, with F-measure providing a single, overall measure of effectiveness. A detailed error analysis was performed on classification errors.ResultsClassification of diabetes, influenza, pneumonia and HIV was highly accurate (F-measure 0.96). More fine-grained ICD-10 classification effectiveness was more variable but still high (F-measure 0.80). The error analysis revealed that word variations as well as certain word combinations adversely affected classification. In addition, anomalies in the ground truth likely led to an underestimation of the effectiveness.ConclusionsThe high accuracy and low cost of the classification methods allow for an effective means for automatic and real-time surveillance of diabetes, influenza, pneumonia and HIV deaths. In addition, the methods are generally applicable to other diseases of interest and to other sources of medical free-text besides death certificates.

BackgroundSmoking during pregnancy increases the risk of adverse health outcomes for both the mother and the child. Rates of smoking during pregnancy, and rates of smoking cessation during pregnancy, vary between demographic groups. This study describes demographic factors associated with smoking cessation during pregnancy in New South Wales, Australia, and describes trends in smoking cessation in demographic subgroups over the period 2000 – 2011.MethodsData were obtained from the New South Wales Perinatal Data Collection, a population-based surveillance system covering all births in New South Wales. Multivariate logistic regression was used to explore associations between smoking cessation during pregnancy and demographic factors.ResultsBetween 2000 and 2011, rates of smoking cessation in pregnancy increased from 4.0% to 25.2%. Demographic characteristics associated with lower rates of smoking cessation during pregnancy included being a teenage mother, being an Aboriginal person, and having a higher number of previous pregnancies.ConclusionsBetween 2000 and 2011, rates of smoking cessation during pregnancy increased dramatically across all demographic groups. However, specific demographic groups remain significantly less likely to quit smoking, suggesting a need for targeted efforts to promote smoking cessation in these groups.

No abstract

Introduction and AimsAcute harm from heavy drinking episodes is an increasing focus of public health policy, but capturing timely data on acute harms in the population is challenging. This study aimed to evaluate the precision of readily available administrative emergency department (ED) data in public health surveillance of acute alcohol harms.Design and MethodsWe selected a random sample of 1000 ED presentations assigned an ED diagnosis code for alcohol harms (the ‘alcohol syndrome’) in the New South Wales, Australia, automatic syndromic surveillance system. The sample was selected from 68 public hospitals during 2014. Nursing triage free‐text fields were independently reviewed to confirm alcohol consumption and classify each presentation into either an ‘acute’ or ‘chronic’ harm. Positive predictive value (PPV) for acute harm was calculated, and predictors of acute harm presentations were estimated using logistic regression.ResultsThe PPV of the alcohol syndrome for acute alcohol harm was 53.5%. Independent predictors of acute harm were ambulance arrival [adjusted odds ratio (aOR) = 3.4, 95% confidence interval (CI) 2.4–4.7], younger age (12–24 vs. 25–39 years: aOR = 3.4, 95% CI 2.2–5.3), not being admitted (aOR 2.2, 95% CI 1.5–3.2) and arriving between 10 pm and 5.59 am (aOR 2.1, 95% CI 1.5–2.8). PPV among 12 to 24‐year‐olds was 82%.Discussion and Conclusions The alcohol syndrome provides moderate precision as an indicator of acute alcohol harms presenting to the ED. Precision for monitoring acute harm in the population is improved by filtering the syndrome by the strongest independent predictors of acute alcohol harm presentations. [Whitlam G, Dinh M, Rodgers C, Muscatello DJ, McGuire R, Ryan T, Thackway S. Diagnosis‐based emergency department alcohol harm surveillance: What can it tell us about acute alcohol harms at the population level? Drug Alcohol Rev 2016;35:693–701]