Corrado Lagazio

Disease mapping in veterinary epidemiology: a Bayesian geostatistical approach

Model-based geostatistics and Bayesian approaches are useful in the context of veterinary epidemiology when point data have been collected by appropriate study design. We take advantage of an example of Epidemiological Surveillance on urban settings where a two-stage sampling design with first stage transects is applied to study the risk of dog parasite infection in the city of Naples, 2004-2005. We specified Bayesian Gaussian spatial exponential models and Bayesian kriging were performed to predict the continuous risk surface of parasite infection on the study region. We compared the results with those obtained by the application of hierarchical Bayesian models on areal data (proportion of positive specimens by transect). The models results were consistent with each other and the Bayesian geostatistical approach proved to be more accurate in identifying areas at risk of zoonotic parasitic diseases. In general, larger risk areas were identified at the city border where wild dogs mixed with domestic dogs and human or urban barriers were less present.

Anthropometric characteristics of full-term infants: effects of varying degrees of "normal" glucose metabolism.

Aim of this study was to examine the maternal-neonatal outcome and the neonatal anthropometric characteristics of a full-term mother-infant pairs group with a positive oral glucose challenge test (GCT) without gestational diabetes mellitus (GDM). Our study involved 1615 white women with singleton pregnancies who underwent universal screening for GDM in two periods of pregnancy. This population was divided into three groups according to GCT results: 1) 172 patients with abnormal GCT in both periods; 2) 391 patient with normal GCT in the early period and abnormal GCT in the late period; 3) 1052 patients with normal GCT in both periods (control group). The incidence of LGA (large for gestational age) infants was higher in Group (40.7%) and Group 2 (22.0%) respect to control group (8.3%) (p < 0.00001 and p < 0.0001 respectively) and was significantly different in the two groups (p < 0.0008). Comparison among the three groups of LGA infants showed the following results: male and female newborns of Group I were heavier than those of Group 2 and of the control group, while males and females of the control group had significantly greater length and cranial circumference means. A significant decrease in ponderal index, choracic circumference, weight/length ratio means could be seen as well as a significative increase in cranial/thoracic circumference ratio means from Group I to the control group. These data confirm the involvement of fetal development in terms of weight and anthropometric characteristics in the presence of alterations in maternal glucose metabolism which are not currently classified as gestational diabetes.

A hierarchical Bayesian model for space-time variation of disease risk

In this paper we propose a hierarchical Bayesian model to study the variation in space and time of disease risk. We represent spatial effects following the usual Bayesian specification of a Gaussian convolution of unstructured and structured components, while we adopt the birth cohort (instead of the commonly used period of death) as the main time scale. The model also includes space-time interaction terms to take into account structured inseparable space-time variability. The model is applied to lung cancer death certificate data in Tuscany, for males during the period 1971-94. While a calendar period analysis points out a general increase of mortality levelling off in the last period (1990-94), the cohort model shows a general and substantial decrease of the relative risk for the youngest cohorts born after 1930. Moreover, the pattern of the epidemic by birth cohort presents a maximum which varies by municipalities, with a strong north-west/south-east gradient.

Mitochondrial DNA in day 3 embryo culture medium is a novel, non-invasive biomarker of blastocyst potential and implantation outcome

In assisted reproduction technology, embryo competence is routinely evaluated on morphological criteria. Over the last decade, efforts in improving non-invasive embryo assessment have looked into the secretome of embryos. Human embryos release genomic DNA (gDNA) and mitochondrial DNA (mtDNA) into the culture medium, and the mtDNA/gDNA ratio is significantly correlated with embryo fragmentation. Here, we investigate whether mtDNA/gDNA ratio in embryo spent medium is correlated with blastulation potential and implantation. The mtDNA/gDNA ratio was assessed in 699 Day 3 culture media by quantitative polymerase chain reaction (qPCR) to investigate its correlation with embryo morphology, blastocyst development and implantation. A logistic regression model evaluated whether mtDNA/gDNA ratio in the secretome may improve the prediction of blastulation. We found that embryos that successfully developed into blastocysts exhibited a significantly higher mtDNA/gDNA ratio in the culture medium compared with those that arrest (P = 0.0251), and mtDNA/gDNA, combined with morphological grading, has the potential to predict blastulation better than morphology alone (P = 0.02). Moreover, mtDNA/gDNA ratio was higher in the media from good-quality embryos that reached the full blastocyst stage on Day 5 compared with those that developed more slowly (P < 0.0001). With respect to blastocyst morphology, higher trophectoderm quality was associated with a higher mtDNA/gDNA ratio in the culture medium. Finally, a high mtDNA/gDNA ratio in spent medium was associated with successful implantation outcome (P = 0.0452) of good-quality embryos. In summary, the mtDNA/gDNA ratio in the Day 3 embryo secretome, in combination with morphological grading, may be a novel, non-invasive, early biomarker to improve identification of viable embryos with high developmental potential.

Monitoring dry-curing of S. Daniele ham by magnetic resonance imaging

S. Daniele hams were collected at different stages during dry-curing and submitted to magnetic resonance imaging (MRI) according to the acquisition Spin-Echo sequences T1 and T2. The intensity of the MR signals in the images of the Semimembranosus, Semitendinosus, Rectus femoris and Biceps femoris muscles of the hams was computed and expressed in grey levels. Muscles were also submitted to traditional analyses, including aw, soluble solids, sodium chloride, total and water soluble nitrogen. T1 and T2 MR signals well described the evolution of the phenomena occurring in the different muscles during dry-curing. MR signal acquired in T2 mode well correlated with traditional indicators in Semitendinosus, Rectus femoris and Biceps femoris muscles. Predictive models estimating the value of aw, moisture, salt content and proteolysis extent on the basis of the MR signal intensity were proposed.

Non-parametric maximum likelihood estimators for disease mapping

A Non-Parametric Maximum Likelihood approach to the estimation of relative risks in the context of disease mapping is discussed and a NPML approximation to conditional autoregressive models is proposed. NPML estimates have been compared to other proposed solutions (Maximum Likelihood via Monte Carlo Scoring, Hierarchical Bayesian models) using real examples. Overall, the NPML autoregressive estimates (with weighted term) were closer to the Bayesian estimates. The exchangeable NPML model ranked immediately after, even if it implied a greater shrinkage, while the truncated auto-Poisson showed inadequate for disease mapping. The coefficients of the autoregressive term for the different mixtures have clear interpretations: in the breast cancer example, the larger cities in the region showed high rates and very low correlation with the neighbouring areas, while the less populated rural areas with low rates were strongly positively correlated each other. This pattern is expected since breast cancer is strongly correlated with parity and age at first birth, and the female population of the rural areas experienced a decline in fertility much later than those living in the larger cities. The leukemia example highlighted the failure of the Poisson-Gamma model and other general overdispersion tests to detect high risk areas under specific conditions. The NPML approach in Aitkin is very general, simple and flexible. However the user should be warned against the possibility of local maxima and the difficulty in detecting the optimal number of components. Special software (such as CAMAN or DismapWin) had been developed and should be recommended mainly to not experienced users.

Covariate selection in multivariate spatial analysis of ovine parasitic infection.

Gastrointestinal (GI) strongyle and fluke infections remain one of the main constraints on health and productivity in sheep dairy production. A cross-sectional survey was conducted in 2004-2005 on ovine farms in the Campania region of southern Italy in order to evaluate the prevalence of Haemonchus contortus, Fasciola hepatica, Dicrocoelium dendriticum and Calicophoron daubneyi from among other parasitic infections. In the present work, we focused on the role of the ecological characteristics of the pasture environment while accounting for the underlying long range geographical risk pattern. Bayesian multivariate spatial statistical analysis was used. A systematic grid (10 km×10 km) sampling approach was used. Laboratory procedures were based on the FLOTAC technique to detect and count eggs of helminths. A Geographical Information System (GIS) was constructed by using environmental data layers. Data on each of these layers were then extracted for pasturing areas that were previously digitalized aerial images of the ovine farms. Bayesian multivariate statistical analyses, including improper multivariate conditional autoregressive models, were used to select covariates on a multivariate spatially structured risk surface. Out of the 121 tested farms, 109 were positive for H. contortus, 81 for D. dendriticum, 17 for C. daubneyi and 15 for F. hepatica. The statistical analysis highlighted a north-south long range spatially structured pattern. This geographical pattern is treated here as a confounder, because the main interest was in the causal role of ecological covariates at the level of each pasturing area. A high percentage of pasture and impermeable soil were strong predictors of F. hepatica risk and a high percentage of wood was a strong predictor of C. daubneyi. A high percentage of wood, rocks and arable soil with sparse trees explained the spatial distribution of D. dendriticum. Sparse vegetation, river, mixed soil and permeable soil explained the spatial distribution of the H. contortus. Bayesian multivariate spatial analysis of parasitic infections with covariates from remote sensing at a very small geographical level allowed us to identify relevant risk predictors. All the covariates selected are consistent with the life cycles of the helminths investigated. This research showed the utility of appropriate GIS-driven surveillance systems. Moreover, spatial features can be used to tailor sampling design where the sampling fraction can be a function of remote sensing covariables.

A hierarchical Bayesian approach to multiple testing in disease mapping.

We propose a Bayesian approach to multiple testing in disease mapping. This study was motivated by a real example regarding the mortality rate for lung cancer, males, in the Tuscan region (Italy). The data are relative to the period 1995–1999 for 287 municipalities. We develop a tri-level hierarchical Bayesian model to estimate for each area the posterior classification probability that is the posterior probability that the municipality belongs to the set of non-divergent areas. We show also the connections of our model with the false discovery rate approach. Posterior classification probabilities are used to explore areas at divergent risk from the reference while controlling for multiple testing. We consider both the Poisson-Gamma and the Besag, York and Mollié model to account for extra Poisson variability in our Bayesian formulation. Posterior inference on classification probabilities is highly dependent on the choice of the prior. We perform a sensitivity analysis and suggest how to rely on subject-specific information to derive informative a priori distributions. Hierarchical Bayesian models provide a sensible way to model classification probabilities in the context of disease mapping.

Parametric and semi-parametric approaches in the analysis of short-term effects of air pollution on health

Generalized additive models (GAMs) have become the standard tool for the analysis of short-term effects of air pollution on human health. Usually, the confounding effect of seasonality and long-term trend is described by flexible parametric or non-parametric functions of calendar time. Two different modeling strategies, i.e. GAM with penalized regression splines and GAM with regression splines, were compared by means of a simulation study, addressing attention to the inference on air pollutant effect. Simulation results indicated that GAM with regression splines provides negligibly biased estimates of air pollutant effect and it is robust to misspecification of the degrees of freedom of the spline. GAM with penalized regression splines requires a certain amount of undersmoothing in order to reduce the bias of the estimates and to improve the coverage of confidence intervals. These findings agree with asymptotic results developed in the context of partially splined models.

A transitional non-parametric maximum pseudo-likelihood estimator for disease mapping

Non-parametric maximum likelihood estimators of relative risk have been proposed as an alternative to empirical Bayes or full Bayes approaches to disease mapping. They have the advantage of being relatively simple, the EM algorithm assures convergence and area classification is straightforward. However, they do not take into account spatial autocorrelation and have higher mean square error when the true underlying risk pattern is strongly spatially structured. Furthermore, the EM algorithm is sensible to starting values and could converge to local maxima. We review the transitional generalized linear models and propose a transitional non-parametric maximum pseudo-likelihood estimator for disease mapping. The usual kernel likelihood of the mixture models is replaced by the conditional density of the observed response for a single area given the values observed in adjacent areas. The estimation of the parameters is based on the EM algorithm, appropriately modified to handle the problem of local maxima and to estimate the number of components of the mixture. A simulation study shows that the transitional non-parametric maximum pseudo-likelihood estimator performs similarly to full Bayes estimators.