Bridging the gap between different statistical approaches: An integrated framework for modelling

This paper proposes a template for modelling complex datasets that integrates traditional statistical modelling approaches with more recent advances in statistics and modelling through an exploratory framework. Our approach builds on the well-known and long standing traditional idea of 'good practice in statistics' by establishing a comprehensive framework for modelling that focuses on exploration, prediction, interpretation and reliability assessment, a relatively new idea that allows individual assessment of predictions. The integrated framework we present comprises two stages. The first involves the use of exploratory methods to help visually understand the data and identify a parsimonious set of explanatory variables. The second encompasses a two step modelling process, where the use of non-parametric methods such as decision trees and generalized additive models are promoted to identify important variables and their modelling relationship with the response before a final predictive model is considered. We focus on fitting the predictive model using parametric, non-parametric and Bayesian approaches. This paper is motivated by a medical problem where interest focuses on developing a risk stratification system for morbidity of 1,710 cardiac patients given a suite of demographic, clinical and preoperative variables. Although the methods we use are applied specifically to this case study, these methods can be applied across any field, irrespective of the type of response.

Rural and urban attitudes to the conservation of Asian elephants in Sri Lanka: Empirical evidence

There has been little study of economic and general attitudes towards the conservation of the Asian elephant. This paper reports and analyses results from surveys conducted in Sri Lanka of attitudes of urban dwellers and farmers towards nature conservation in general and the elephant conservation in particular. The analyses are based on urban and a rural sample. Contingent valuation techniques are used as survey instruments. Multivariate logit regression analysis is used to analyze the respondents’ attitudes towards conservation of elephants. It is found that, although some variations occurred between the samples, the majority of the respondents (both rural and urban) have positive attitudes towards nature conservation in general. However, marked differences in attitudes toward elephant conservation are evident between these two samples: the majority of urban respondents were in favour of elephant conservation; rural respondents expressed a mixture of positive and negative attitudes. Overall, considerable unrecorded and as yet unutilised economic support for conservation of wild elephants exists in Sri Lanka.

Lipid peroxidation in hepatic steatosis in humans is associated with hepatic fibrosis and occurs predominately in acinar zone 3

Background and Aims: Hepatic steatosis has been shown to be associated with lipid peroxidation and hepatic fibrosis in a variety of liver diseases including non-alcoholic fatty liver disease. However, the lobular distribution of lipid peroxidation associated with hepatic steatosis, and the influence of hepatic iron stores on this are unknown. The aim of this study was to assess the distribution of lipid peroxidation in association with these factors, and the relationship of this to the fibrogenic cascade. Methods: Liver biopsies from 39 patients with varying degrees of hepatic steatosis were assessed for evidence of lipid peroxidation (malondialdehyde adducts), hepatic iron, inflammation, fibrosis, hepatic ;stellate cell activation (alpha-smooth muscle actin and TGF-beta expression) and collagen type I synthesis (procollagen a 1 (I) mRNA). Results: Lipid peroxidation occurred in and adjacent to fat-laden hepatocytes and was maximal in acinar zone 3. Fibrosis was associated with steatosis (P < 0.04), lipid peroxidation (P < 0.05) and hepatic iron stores (P < 0.02). Multivariate logistic regression analysis confirmed the association between steatosis and lipid peroxidation within zone 3 hepatocytes (P < 0.05), while for hepatic iron, lipid peroxidation was seen within sinusoidal cells (P < 0.05), particularly in zone 1 (P < 0.02). Steatosis was also associated with acinar inflammation (P < 0.005). α-Smooth muscle actin expression was present in association with both lipid peroxidation and fibrosis. Although the effects of steatosis and iron on lipid peroxidation and fibrosis were additive, there was no evidence of a specific synergistic interaction between them. Conclusions: These observations support a model where steatosis exerts an effect on fibrosis through lipid peroxidation, particularly in zone 3 hepatocytes. (C) 2001 Blackwell Science Asia Pty Ltd.

Myocardial abnormalities in hypertensive patients with normal and abnormal left ventricular filling: a study of ultrasound tissue characterization and strain

Abnormal left ventricular (LV) filling is common, but not universal, in hypertensive LV hypertrophy (LVH). We sought to elucidate the relative contributions of myocardial structural changes, loading and hypertrophy to LV dysfunction in 113 patients: 85 with hypertensive LVH and 28 controls without LVH and with normal filling. Patients with normal dobutamine stress echocardiography and no history of coronary artery disease were selected, in order to exclude a contribution from ischaemia or scar. Abnormal LV filling was identified in 65 LVH patients, based on Doppler measurement of transmitral filling and annular velocities. All patients underwent grey-scale and colour tissue Doppler imaging from three apical views, which were stored and analysed off line. Integrated backscatter (113) and strain rate imaging were used to detect changes in structure and function; average cyclic variation of 113, strain rate and peak systolic strain were calculated by averaging each segment. Calibrated 113 intensity, corrected for pericardial 113 intensity, was measured in the septum and posterior wall from the parasternal long-axis view. Patients with LVH differed significantly from controls with respect to all backscatter and strain parameters, irrespective of the presence or absence of abnormal LV filling. LVH patients with and without abnormal LV filling differed with regard to age, LV mass and incidence of diabetes mellitus, but also showed significant differences in cyclic variation (P < 0.01), calibrated 113 in the posterior wall (P < 0.05) and strain rate (P < 0.01), although blood pressure, heart rate and LV systolic function were similar. Multivariate logistic regression analysis demonstrated that age, LV mass index and calibrated IB in the posterior wall were independent determinants of abnormal LV filling in patients with LVH. Thus structural and functional abnormalities can be detected in hypertensive patients with LVH with and without abnormal LV filling. In addition to age and LVH, structural (not functional) abnormalities are likely to contribute to abnormal LV filling, and may be an early sign of LV damage. 113 is useful for the detection of myocardial abnormalities in patients with hypertensive LVH.

Risk factors for non-haemorrhagic stroke in patients with coronary heart disease and the effect of lipid-modifying therapy with pravastatin

Objective To determine the relative importance of recognised risk factors for non-haemorrhagic stroke, including serum cholesterol and the effect of cholesterol-lowering therapy, on the occurrence of non-haemorrhagic stroke in patients enrolled in the LIPID (Long-term Intervention with Pravastatin in Ischaemic Disease) study. Design The LIPID study was a placebo-controlled, double-blind trial of the efficacy on coronary heart disease mortality of pravastatin therapy over 6 years in 9014 patients with previous acute coronary syndromes and baseline total cholesterol of 4-7 mmol/l. Following identification of patients who had suffered non-haemorrhagic stroke, a pre-specified secondary end point, multivariate Cox regression was used to determine risk in the total population. Time-to-event analysis was used to determine the effect of pravastatin therapy on the rate of non-haemorrhagic stroke. Results There were 388 non-haemorrhagic strokes in 350 patients. Factors conferring risk of future non-haemorrhagic stroke were age, atrial fibrillation, prior stroke, diabetes, hypertension, systolic blood pressure, cigarette smoking, body mass index, male sex and creatinine clearance. Baseline lipids did not predict non-haemorrhagic stroke. Treatment with pravastatin reduced non-haemorrhagic stroke by 23% (P= 0.016) when considered alone, and 21% (P= 0.024) after adjustment for other risk factors. Conclusions The study confirmed the variety of risk factors for non-haemorrhagic stroke. From the risk predictors, a simple prognostic index was created for nonhaemorrhagic stroke to identify a group of patients at high risk. Treatment with pravastatin resulted in significant additional benefit after allowance for risk factors. (C) 2002 Lippincott Williams Wilkins.

Abnormal left ventricular filling with increasing age reflects abnormal myocardial characteristics independent of ischemia or hypertrophy

Abnormal left ventricular (IV) filling may occur with increasing age despite apparently normal IV size and function, and is usually attributed to IV hypertrophy and coronary artery disease. The purpose of this study was to determine whether myocardial abnormalities could be identified in 67 such patients (36 men, mean age 57 +/- 9 years) whose IV hypertrophy and coronary artery disease were excluded by dobutamine echocardiography. All patients underwent gray scale and color tissue Doppler imaging from 3 apical views, which were stored and analyzed off line. Disturbances in structure and function were assessed by averaging the cyclic variation of integrated backscatter, strain rate, and peak systolic strain from each myocardial segment. Calibrated integrated backscatter (corrected for pericardial backscatter intensity) was measured in the septum and posterior wall from the parasternal long-axis view. Abnormal IV filling was present in 36 subjects (54%). Subjects with and without abnormal IV filling had similar IV mass, but differed in age (p <0.01), cyclic variation (p = 0.001), strain rate (p <0.01), and peak systolic strain (p <0.001). Multivariate logistic regression analysis demonstrated that age (p = 0.016) and cyclic variation (p = 0.042) were the most important determinants of abnormal IV filling in these apparently normal subjects. (C) 2003 by Excerpta Medica, Inc.

Comparison of rural and urban attitudes to the conservation of Asian elephants in Sri Lanka: empirical evidence

There has been little study of economic and general attitudes towards the conservation of the Asian elephant. This paper reports and analyses results from surveys conducted in Sri Lanka of attitudes of urban dwellers and farmers towards nature conservation in general and the elephant conservation in particular. The analyses are based on urban and a rural sample. Contingent valuation techniques are used as survey instruments. Multivariate logit regression analysis is used to analyse the respondents' attitudes towards conservation of elephants. It is found that, although some variations occurred between the samples, the majority of the respondents (both rural and urban) have positive attitudes towards nature conservation in general. However, marked differences in attitudes toward elephant conservation are evident between these two samples: the majority of urban respondents were in favour of elephant conservation; rural respondents expressed a mixture of positive and negative attitudes. Overall, considerable unrecorded and as yet unutilised economic support for conservation of wild elephants exists in Sri Lanka. (C) 2002 Elsevier Science Ltd. All rights reserved.

The relative importance of vascular structure and function in predicting cardiovascular events

OBJECTIVES We sought to assess the prognostic utility of brachial artery reactivity (BAR) in patients at risk of cardiovascular events. BACKGROUND Impaired flow-mediated vasodilation measured by BAR is a marker of endothelial dysfunction. Brachial artery reactivity is influenced by risk factors and is responsive to various pharmacological and other treatments. However, its prognostic importance is uncertain, especially relative to other predictors of outcome. METHODS A total of 444 patients were prospectively enrolled to undergo BAR and follow-up. These patients were at risk of cardiovascular events, based on the presence of risk factors or known or suspected cardiovascular disease. We took a full clinical history, performed BAR, and obtained carotid intima-media thickness (IMT) and left ventricular mass and ejection fraction. Patients were followed up for cardiovascular events and all-cause mortality. Multivariate Cox regression analysis was performed to assess the independent association of investigation variables on outcomes. RESULTS The patients exhibited abnormal BAR (5.2 +/- 6.1% [mean +/- SD]) but showed normal nitrate-mediated dilation (9.9 +/- 7.2%) and normal mean IMT (0.67 +/- 0.12 mm [average]). Forty-nine deaths occurred over the median follow-up period of 24 months (interquartile range 10 to 34). Patients in the lowest tertile group of BAR (<2%) had significantly more events than those in the combined group of highest and mid-tertiles (p = 0.029, log-rank test). However, mean IMT (rather than flow-mediated dilation) was the vascular factor independently associated with mortality, even in the subgroup (n = 271) with no coronary artery disease and low risk. CONCLUSIONS Brachial artery reactivity is lower in patients with events, but is not an independent predictor of cardiovascular outcomes in this cohort of patients. (C) 2004 by the American College of Cardiology Foundation.

Prevalence of whole-body skin self-examination in a population at high risk for skin cancer (Australia)

Objective: Whole-body skin self-examination (SSE) with presentation of suspicious lesions to a physician may improve early detection of melanoma. The aim of this study was to establish the prevalence and determinants of SSE in a high-risk population in preparation for a community-based randomised controlled trial of screening for melanoma. Methods: A telephone survey reached 3110 residents older than 30 years (overall response rate of 66.9%) randomly selected from 18 regional communities in Queensland, Australia. Results: Overall, 804 (25.9%) participants reported whole-body SSE within the past 12 months and 1055 (33.9%) within the past three years. Whole-body SSE was associated in multivariate logistic regression analysis with younger age (< 50 years); higher education; having received either a whole-body skin examination, recommendation or instruction on SSE by a primary care physician; giving skin checks a high priority; concern about skin cancer and a personal history of skin cancer. Conclusion: Overall, the prevalence of SSE in the present study is among the highest yet observed in Australia, with about one-third of the adult population reporting whole-body SSE in the past three years. People over 50 years, who are at relatively higher risk for skin cancer, currently perform SSE less frequently than younger people.

Pre-term delivery and periodontal disease: a case-control study from Croatia

Aim: The aim of this report was to assess the strength and influence of periodontitis as a possible risk factor for pre-term birth (PTB) in a cohort of 81 primiparous Croatian mothers aged 18-39 years. Methods: PTB cases (n=17; mean age 25 +/- 2.9 years; age range 20-33 years) were defined as spontaneous delivery after less than 37 completed weeks of gestation that were followed by spontaneous labour or spontaneous rupture of membranes. Controls (full-time births) were normal births at or after 37 weeks of gestation (n=64; mean age 25 +/- 2.9 years; age range 19-39 years). Information on known risk factors and obstetric factors included the current pregnancy history, maternal age at delivery, pre-natal care, nutritional status, tobacco use, alcohol use, genitourinary infections, vaginosis, gestational age, and birth weight. Full-mouth periodontal examination was performed on all mothers within 2 days of delivery. Results: PTB cases had significantly worse periodontal status than controls (p=0.008). Multivariate logistic regression model, after controlling for other risk factors, demonstrated that periodontal disease is a significant independent risk factor for PTB, with an adjusted odds ratio of 8.13 for the PTB group (95% confidence interval 2.73-45.9). Conclusion: Periodontal disease represents a strong, independent, and clinically significant risk factor for PTB in the studied cohort. There are strong indicators that periodontal therapy should form a part of preventive prenatal care in Croatia.

Poverty and the economic effects of TB in rural China

OBJECTIVE: To investigate the economic effects of illness on individual tuberculosis (TB) cases in rural China and to use a case-control study to show a strong TB-poverty link. SETTING: In 2002-2004 we studied 160 new smear-positive pulmonary tuberculosis (PTB) cases and 320 age- and sex-matched controls living in neighbouring houses in four rural counties of Henan Province. DESIGN: Cases and controls were interviewed 1-3 months after patients were diagnosed. We used matched multivariate logistic regression to compare cases with controls for poverty status using household income, household assets and relative wealth within the village. We conducted follow-up interviews of patients 10-12 months later to assess economic effects by collecting data on treatment costs, income losses, coping strategies and treatment completion. RESULTS: Poverty is strongly associated with TB incidence even after controlling for smoking and other risk factors. Excluding income losses, direct out-of-pocket treatment costs (medical and non-medical) accounted for 55.5 % of average annual household income, and most TB cases fell into heavy debt. The DOTS cure rate was 91 %. When DOTS was incomplete or not done, mortality was high. CONCLUSIONS: Poverty is both a cause and a devastating outcome of TB. Ongoing poverty reduction schemes in China must also include reducing TB.

Interaction of psychosocial risk factors explain increased neck problems among female office workers

This study investigated the relationship between psychosocial risk factors and (1) neck symptoms and (2) neck pain and disability as measured by the neck disability index (NDI). Female office workers employed in local private and public organizations were invited to participate, with 333 completing a questionnaire. Data were collected on various risk factors including age, negative affectivity, history of previous neck trauma, physical work environment, and task demands. Sixty-one percent of the sample reported neck symptoms lasting greater than 8 days in the last 12 months. The mean NDI of the sample was 15.5 out of 100, indicating mild neck pain and disability. In a hierarchical multivariate logistic regression, low supervisor support was the only psychosocial risk factor identified with the presence of neck symptoms. Similarly, low supervisor support was the only factor associated with the score on the NDI. These associations remained after adjustment for potential confounders of age, negative affectivity, and physical risk factors. The interaction of job demands, decision authority, and supervisor support was significantly associated with the NDI in the final model and this association increased when those with previous trauma were excluded. Interestingly, and somewhat contrary to initial expectations, as job demands increased, high decision authority had an increasing effect on the NDI when supervisor support was low. Crown copyright (c) 2006 Published by Elsevier B.V. All rights reserved.

Adaptive multiscale principal components analysis for online monitoring of wastewater treatment

Fault detection and isolation (FDI) are important steps in the monitoring and supervision of industrial processes. Biological wastewater treatment (WWT) plants are difficult to model, and hence to monitor, because of the complexity of the biological reactions and because plant influent and disturbances are highly variable and/or unmeasured. Multivariate statistical models have been developed for a wide variety of situations over the past few decades, proving successful in many applications. In this paper we develop a new monitoring algorithm based on Principal Components Analysis (PCA). It can be seen equivalently as making Multiscale PCA (MSPCA) adaptive, or as a multiscale decomposition of adaptive PCA. Adaptive Multiscale PCA (AdMSPCA) exploits the changing multivariate relationships between variables at different time-scales. Adaptation of scale PCA models over time permits them to follow the evolution of the process, inputs or disturbances. Performance of AdMSPCA and adaptive PCA on a real WWT data set is compared and contrasted. The most significant difference observed was the ability of AdMSPCA to adapt to a much wider range of changes. This was mainly due to the flexibility afforded by allowing each scale model to adapt whenever it did not signal an abnormal event at that scale. Relative detection speeds were examined only summarily, but seemed to depend on the characteristics of the faults/disturbances. The results of the algorithms were similar for sudden changes, but AdMSPCA appeared more sensitive to slower changes.

Experimental evidence for multivariate stabilizing sexual selection

Stabilizing selection is a fundamental concept in evolutionary biology. In the presence of a single intermediate optimum phenotype (fitness peak) on the fitness surface, stabilizing selection should cause the population to evolve toward such a peak. This prediction has seldom been tested, particularly for suites of correlated traits. The lack of tests for an evolutionary match between population means and adaptive peaks may be due, at least in part, to problems associated with empirically detecting multivariate stabilizing selection and with testing whether population means are at the peak of multivariate fitness surfaces. Here we show how canonical analysis of the fitness surface, combined with the estimation of confidence regions for stationary points on quadratic response surfaces, may be used to define multivariate stabilizing selection on a suite of traits and to establish whether natural populations reside on the multivariate peak. We manufactured artificial advertisement calls of the male cricket Teleogryllus commodus and played them back to females in laboratory phonotaxis trials to estimate the linear and nonlinear sexual selection that female phonotactic choice imposes on male call structure. Significant nonlinear selection on the major axes of the fitness surface was convex in nature and displayed an intermediate optimum, indicating multivariate stabilizing selection. The mean phenotypes of four independent samples of males, from the same population as the females used in phonotaxis trials, were within the 95% confidence region for the fitness peak. These experiments indicate that stabilizing sexual selection may play an important role in the evolution of male call properties in natural populations of T. commodus.

Studying phenotypic evolution using multivariate quantitative genetics

Quantitative genetics provides a powerful framework for studying phenotypic evolution and the evolution of adaptive genetic variation. Central to the approach is G, the matrix of additive genetic variances and covariances. G summarizes the genetic basis of the traits and can be used to predict the phenotypic response to multivariate selection or to drift. Recent analytical and computational advances have improved both the power and the accessibility of the necessary multivariate statistics. It is now possible to study the relationships between G and other evolutionary parameters, such as those describing the mutational input, the shape and orientation of the adaptive landscape, and the phenotypic divergence among populations. At the same time, we are moving towards a greater understanding of how the genetic variation summarized by G evolves. Computer simulations of the evolution of G, innovations in matrix comparison methods, and rapid development of powerful molecular genetic tools have all opened the way for dissecting the interaction between allelic variation and evolutionary process. Here I discuss some current uses of G, problems with the application of these approaches, and identify avenues for future research.