Model specification tests in nonparametric stochastic regression models

In this paper, we consider testing for additivity in a class of nonparametric stochastic regression models. Two test statistics are constructed and their asymptotic distributions are established. We also conduct a small sample study for one of the test statistics through a simulated example. (C) 2002 Elsevier Science (USA).

Forecasting with serially correlated regression models

In this article we investigate the asymptotic and finite-sample properties of predictors of regression models with autocorrelated errors. We prove new theorems associated with the predictive efficiency of generalized least squares (GLS) and incorrectly structured GLS predictors. We also establish the form associated with their predictive mean squared errors as well as the magnitude of these errors relative to each other and to those generated from the ordinary least squares (OLS) predictor. A large simulation study is used to evaluate the finite-sample performance of forecasts generated from models using different corrections for the serial correlation.

Optimal crossover designs for logistic regression models in pharmacodynamics

Pharmacodynamics (PD) is the study of the biochemical and physiological effects of drugs. The construction of optimal designs for dose-ranging trials with multiple periods is considered in this paper, where the outcome of the trial (the effect of the drug) is considered to be a binary response: the success or failure of a drug to bring about a particular change in the subject after a given amount of time. The carryover effect of each dose from one period to the next is assumed to be proportional to the direct effect. It is shown for a logistic regression model that the efficiency of optimal parallel (single-period) or crossover (two-period) design is substantially greater than a balanced design. The optimal designs are also shown to be robust to misspecification of the value of the parameters. Finally, the parallel and crossover designs are combined to provide the experimenter with greater flexibility.

Social-cognitive and perceived environment influences associated with physical activity in older Australians

Background, Regular physical activity in older adults can facilitate healthy aging, improve functional capacity, and prevent disease. However, factors associated with physical inactivity in older populations are poorly understood. This study attempts to identify social-cognitive and perceived environmental influences associated with physical activity participation in older populations. Methods. In a randomly selected sample of 449 Australian adults age 60 and older, we assessed self-reported physical activity and a range of social-cognitive and perceived environmental factors. Respondents were classified as sufficiently active and inactive based on energy expenditure estimates (kcal/week) derived from self-reported physical activity. Two logistic regression models, with and without self-efficacy included, were conducted to identify modifiable independent predictors of physical activity. Results. Significantly more males than females were physically active. Physical activity participation was related to age with a greater proportion of those age 65-69 being active than those age 60-64 or 70 or older. High self-efficacy, regular participation of friends and family, finding footpaths safe for walking, and access to local facilities were significantly associated with being active. Conclusion. Identifying predictors of physical activity in older populations, particularly social support, facility access, and neighbourhood safety, can inform the development of policy and intervention strategies to promote the health of older people. (C) 2000 American Health Foundation and Academic Press.

Increased rates of ENT surgery among young children: Have clinical guidelines made a difference?

Objectives: To examine the association between introduction of paediatric ear, nose and throat (ENT) surgery guidelines and population procedure rates. To determine changes in children's risk of undergoing ENT surgery. Methods: Trend analysis of incidence of myringotomy, tonsillectomy and adenoidectomy among New South Wales (NSW) children aged 0-14 between 1981 and mid 1999. Poisson regression models were used to estimate annual rates of change pre and postguidelines introduction and age/gender specific rates, and lifetable methods to determine risk of undergoing an ENT procedure by age 15. Results: ENT surgery rates increased by 21% over the study period. Children's risk of surgery increased from 17.9% in 1981 to 20.2% in 1998/99. Guideline introduction was associated with moderate short-term decreases in rates. For tonsillectomy, rates decreased between 1981 and 1983, but then rose continually until the introduction of myringotomy guidelines in 1993, when they fell, only to recommence rising until the end of the study period. For myringotomy, rates rose annually from 1981 to 1992/93 and fell in the 3 years following guideline introduction, after which they rose again. Increases were almost exclusively restricted to children aged 0-4 and correspond with increased use of formal childcare. The prevalence of myringotomy by the age of 5 years rose from 5.6% of children born in 1988/89 to 6.4% of those born in 1994/95, and the prevalence of tonsillectomy from 2.4% to 2.7%. Conclusions: The risk of young Australian children undergoing ENT surgery increased significantly over the last two decades despite the introduction of guidelines and no evidence of an increase in otitis media, one condition prompting surgery. Surgery increased most among the very young. We hypothesize this is related to increasing use of childcare.

Mental health and socio-economic variations in Australian suicide

This paper investigates the relationship between suicide rates and prevalence of mental disorder and suicide attempts, across socio-economic status (SES) groups based on area of residence. Australian suicide data (1996-1998) were analysed in conjunction with area-based prevalences of mental disorder derived from the National Survey of Mental Health and Well-Being (1997). Poisson regression models of suicide risk included age, quintile of area-based SES, urban-rural residence, and country of birth (COB), with males and females analysed separately. Analysis focussed on the association between suicide and prevalences of (ICD-10) affective disorders, anxiety disorders, substance use disorders and suicide attempts by SES group. Prevalences of other psychiatric symptomatology, substance use problems, health service utilisation, stressful life-events and personality were also investigated. Significant increasing gradients were evident from high to low SES groups for prevalences of affective disorders, anxiety disorders (females only), and substance use disorders (males only); sub-threshold drug and alcohol problems and depression; and suicide attempts and suicide (males only). Prevalences of mental disorder, other sub-threshold mental health items and suicide attempts were significantly associated with suicide, but in most cases associations were reduced in magnitude and became statistically non-significant after adjustment for COB, urban-rural residence, and SES. For male suicide the relative risk (RR) in the lowest SES group compared to the highest was 1.40 (95% CI 1.29-1.52, p < 0.001) for all ages, and 1.46 (95% CI 1.27-1.67, p < 0.001) for male youth (20-34 years). This relationship was not substantially modified in males when regression models included prevalences of affective disorders, and other selected mental health variables and demographic factors. From a population perspective, SES remained significantly associated with suicide after controlling for the prevalence of mental disorders and other psychiatric symptomatology. Mental conditions and previous suicidal behaviour may play an intermediary role between SES and suicide, but this study suggests that an independent relationship between suicide and SES also exists. (c) 2005 Elsevier Ltd. All rights reserved.

Breastfeeding, menopause, and epithelial ovarian cancer

No previous study has examined the modifying effect of menopausal status on the association between lactation and ovarian cancer risk. We recruited 824 epithelial ovarian cancer cases and 855 community controls in three Australian states, collecting reproductive and lactation histories by means of a contraceptive calendar and pregnancy and breastfeeding record. We report results in women with at least one liveborn infant for unsupplemented breastfeeding, in line with a biological model linking suppression of ovulation to reduction in ovarian cancer risk. We derived odds ratios from multiple logistic regression models including number of liveborn children, age, age at first or last birth, and other potential confounders, overall and by menopausal status. Estimates of relative risk of ovarian cancer per month of full lactation were 0.99 [95% confidence interval(CI) = 0.97-1.00] overall and 1.00 (95% CI = 0.99-1.01) and 0.98 (95% CI = 0.95-1.01) among post- and premenopausal women, respectively. We tailored a lactation variable to the incessant ovulation hypothesis by progressively discounting breastfeeding the longer after birth it occurred, finding odds ratios similar to those for the unmodified duration variable. We found no association of note among postmenopausal women. Breastfeeding seems to be somewhat protective against ovarian cancer, but only before menopause.

Dietary influences on survival after ovarian cancer

We evaluated the effects of various food groups and micronutrients in the diet on survival among women who originally participated in a population-based case-control study of ovarian cancer conducted across 3 Australian states between 1990 and 1993. This analysis included 609 women with invasive epithelial ovarian cancer, primarily because there was negligible mortality in women with borderline tumors. The women's usual diet was assessed using a validated food frequency questionnaire. Deaths in the cohort were identified using state-based cancer registries and the Australian National Death Index (NDI). Crude 5-year survival probabilities were estimated using the Kaplan-Meier technique, and adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) were obtained from Cox regression models. After adjusting for important confounding factors, a survival advantage was observed for those who reported higher intake of vegetables in general (HR = 0.75, 95% CI = 0.57-0.99, p-value trend 0.01 for the highest third, compared to the lowest third), and cruciferous vegetables in particular (HR = 0.75, 95% CI = 0.57-0.98, p-value trend 0.03), and among women in the upper third of intake of vitamin E (HR = 0.76, 95% CI = 0.58-1.01, p-value trend 0.04). Inverse associations were also seen with protein (p-value trend 0.09), red meat (p-value trend 0.06) and white meat (p-value trend 0.07), and modest positive trends (maximum 30% excess) with lactose (p-value trend 0.04), calcium and dairy products. Although much remains to be learned about the influence of nutritional factors after a diagnosis of ovarian cancer, our study suggests the possibility that a diet high in vegetable intake may help improve survival. (C) 2003 Wiley-Liss, Inc.