Parameterized Link Functions in Generalized Linear Random Effect Models: a Case Study on Breast Cancer Treatment

In non-linear random effects some attention has been very recently devoted to the analysis ofsuitable transformation of the response variables separately (Taylor 1996) or not (Oberg and Davidian 2000) from the transformations of the covariates and, as far as we know, no investigation has been carried out on the choice of link function in such models. In our study we consider the use of a random effect model when a parameterized family of links (Aranda-Ordaz 1981, Prentice 1996, Pregibon 1980, Stukel 1988 and Czado 1997) is introduced. We point out the advantages and the drawbacks associated with the choice of this data-driven kind of modeling. Difficulties in the interpretation of regression parameters, and therefore in understanding the influence of covariates, as well as problems related to loss of efficiency of estimates and overfitting, are discussed. A case study on radiotherapy usage in breast cancer treatment is discussed.

Estimation for the general sample selection models

Consider a general regression model with an arbitrary and unknown link function and a stochastic selection variable that determines whether the outcome variable is observable or missing. The paper proposes U-statistics that are based on kernel functions as estimators for the directions of the parameter vectors in the link function and the selection equation, and shows that these estimators are consistent and asymptotically normal.

Temporal trends (2000–2011) and influences on fishery-independent catch rates for loggerhead sea turtles (Caretta caretta) at an important coastal foraging region in the southeastern United States

Seasonal trawling was conducted randomly in coastal (depths of 4.6–17 m) waters from St. Augustine, Florida, (29.9°N) to Winyah Bay, South Carolina (33.1°N), during 2000–03, 2008–09, and 2011 to assess annual trends in the relative abundance of sea turtles. A total of 1262 loggerhead sea turtles (Caretta caretta) were captured in 23% (951) of 4207 sampling events. Capture rates (overall and among prevalent 5-cm size classes) were analyzed through the use of a generalized linear model with log link function for the 4097 events that had complete observations for all 25 model parameters. Final models explained 6.6% (70.1–75.0 cm minimum straight-line carapace length [SCLmin]) to 14.9% (75.1–80.0 cm SCLmin) of deviance in the data set. Sampling year, geographic subregion, and distance from shore were retained as significant terms in all final models, and these terms collectively accounted for 6.2% of overall model deviance (range: 4.5–11.7% of variance among 5-cm size classes). We retained 18 parameters only in a subset of final models: 4 as exclusively significant terms, 5 as a mixture of significant or nonsignificant terms, and 9 as exclusively nonsignificant terms. Four parameters also were dropped completely from all final models. The generalized linear model proved appropriate for monitoring trends for this data set that was laden with zero values for catches and was compiled for a globally protected species. Because we could not account for much model deviance, metrics other than those examined in our study may better explain catch variability and, once elucidated, their inclusion in the generalized linear model should improve model fits.

Étude du comportement et des contextes associés à la consommation d’aliments sucrés dans une communauté montréalaise originaire du Moyen-Orient

L’Organisation mondiale de la Santé recommande aux individus de limiter leur consommation d’aliments sucrés dans le but de prévenir le développement des maladies chroniques. En santé publique, peu de recherches ont tenté d’identifier les facteurs individuels et contextuels qui peuvent influencer conjointement la consommation de ces aliments. Or, de telles connaissances seraient utiles pour guider les interventions nutritionnelles visant à en réduire la consommation. L’objectif de cette thèse est d'étudier les facteurs reliés au comportement et les contextes associés à la consommation quotidienne d’aliments sucrés chez des adultes vivant dans un milieu urbain occidental. Cette étude a été menée auprès d'une communauté moyen-orientale établie dans la Ville de Montréal. Les aliments sucrés ont été définis comme étant les glucides raffinés dont la teneur en sucres totaux dépasse 20 % de l’énergie totale. Lors de l’étape exploratoire (N = 42), un rappel de 24 heures a permis d’identifier les sources d’aliments sucrés et de déterminer l’apport quotidien en sucres totaux de cette communauté. Une étude qualitative descriptive a été privilégiée et un cadre écologique a guidé la réalisation d’entrevues semi-dirigées sur les contextes de consommation (N = 42). Une analyse de contenu employant des procédures de codage initial et focus a mené à l’élaboration d’un instrument de mesure quantitatif sur les contextes de consommation. Cet instrument a été soumis à un pré-test (N = 20), puis administré à l’échantillon principal (N = 192). Une analyse factorielle exploratoire a permis de préciser les contextes de consommation. Les facteurs individuels mesurés incluent les données sociodémographiques, les symptômes dépressifs, la maîtrise de soi, l’assoupissement de jour, les perceptions ainsi que l’hémoglobine glycosylée. La consommation quotidienne de sucres totaux a été mesurée par un questionnaire de fréquence alimentaire (N = 192). Une analyse de régression multivariée employant le modèle linéaire généralisé (distribution de type gamma et lien logarithmique) a été effectuée pour mesurer les relations entre les contextes de consommation, les facteurs individuels et la consommation de sucres totaux, en contrôlant l’âge et le sexe. L’apport quotidien en sucres totaux de l'échantillon est de 20,3 %, ce qui s’apparente aux apports des Canadiens et des Québécois. La consommation quotidienne moyenne est de 76 g/j. Les analyses qualitative et factorielle ont permis d’identifier un ensemble de 42 contextes de consommation regroupés en sept domaines (Actes et situations de grignotage, Stimuli visuels, Besoins énergétiques, Besoins émotionnels, Indulgence, Contraintes, Socialisation). La consommation quotidienne de sucres totaux est supérieure chez les hommes (B = 0,204, ES = 0,094, p = 0,03). Les facteurs positivement associés à la consommation sont le grignotage (B = 0,225, ES = 0,091, p = 0,01), la prise de dessert (B = 0,105, ES = 0,036, p = 0,001) ainsi que les symptômes dépressifs (B = 0,017, ES = 0,094, p = 0,03). L’âge (B = -0,01, ES = 0,004, p = 0,02), l’indulgence (B = -0,103, ES = 0,052, p = 0,05) et l’auto-modération (B = -0,121, ES = 0,042, p = 0,001) montrent, pour leur part, une association négative. Cette étude a privilégié une méthodologie mixte et a permis de développer une mesure innovatrice pour étudier les facteurs contextuels associés à la consommation d’aliments sucrés. Ceux-ci ont été analysés conjointement avec les facteurs individuels. Afin d'encourager les individus à réduire leur consommation de sucres totaux lorsque nécessaire, les initiatives en santé publique devraient en effet cibler les contextes de consommation de même que les facteurs individuels.

A practical approach to a multi-level analysis with a sparse binary outcome within a large surgical trial

Objectives: To assess the potential source of variation that surgeon may add to patient outcome in a clinical trial of surgical procedures. Methods: Two large (n = 1380) parallel multicentre randomized surgical trials were undertaken to compare laparoscopically assisted hysterectomy with conventional methods of abdominal and vaginal hysterectomy; involving 43 surgeons. The primary end point of the trial was the occurrence of at least one major complication. Patients were nested within surgeons giving the data set a hierarchical structure. A total of 10% of patients had at least one major complication, that is, a sparse binary outcome variable. A linear mixed logistic regression model (with logit link function) was used to model the probability of a major complication, with surgeon fitted as a random effect. Models were fitted using the method of maximum likelihood in SAS((R)). Results: There were many convergence problems. These were resolved using a variety of approaches including; treating all effects as fixed for the initial model building; modelling the variance of a parameter on a logarithmic scale and centring of continuous covariates. The initial model building process indicated no significant 'type of operation' across surgeon interaction effect in either trial, the 'type of operation' term was highly significant in the abdominal trial, and the 'surgeon' term was not significant in either trial. Conclusions: The analysis did not find a surgeon effect but it is difficult to conclude that there was not a difference between surgeons. The statistical test may have lacked sufficient power, the variance estimates were small with large standard errors, indicating that the precision of the variance estimates may be questionable.

Generalized linear mixed models for the genetic evaluation of binary reproductive traits: a simulation study

The objective of this study was to evaluate the use of probit and logit link functions for the genetic evaluation of early pregnancy using simulated data. The following simulation/analysis structures were constructed: logit/logit, logit/probit, probit/logit, and probit/probit. The percentages of precocious females were 5, 10, 15, 20, 25 and 30% and were adjusted based on a change in the mean of the latent variable. The parametric heritability (h²) was 0.40. Simulation and genetic evaluation were implemented in the R software. Heritability estimates (ĥ²) were compared with h² using the mean squared error. Pearson correlations between predicted and true breeding values and the percentage of coincidence between true and predicted ranking, considering the 10% of bulls with the highest breeding values (TOP10) were calculated. The mean ĥ² values were under- and overestimated for all percentages of precocious females when logit/probit and probit/logit models used. In addition, the mean squared errors of these models were high when compared with those obtained with the probit/probit and logit/logit models. Considering ĥ², probit/probit and logit/logit were also superior to logit/probit and probit/logit, providing values close to the parametric heritability. Logit/probit and probit/logit presented low Pearson correlations, whereas the correlations obtained with probit/probit and logit/logit ranged from moderate to high. With respect to the TOP10 bulls, logit/probit and probit/logit presented much lower percentages than probit/probit and logit/logit. The genetic parameter estimates and predictions of breeding values of the animals obtained with the logit/logit and probit/probit models were similar. In contrast, the results obtained with probit/logit and logit/probit were not satisfactory. There is need to compare the estimation and prediction ability of logit and probit link functions.

Choosing between Cox proportional hazards and logistic models for interval-censored data via bootstrap

This work develops a new methodology in order to discriminate models for interval-censored data based on bootstrap residual simulation by observing the deviance difference from one model in relation to another, according to Hinde (1992). Generally, this sort of data can generate a large number of tied observations and, in this case, survival time can be regarded as discrete. Therefore, the Cox proportional hazards model for grouped data (Prentice & Gloeckler, 1978) and the logistic model (Lawless, 1982) can befitted by means of generalized linear models. Whitehead (1989) considered censoring to be an indicative variable with a binomial distribution and fitted the Cox proportional hazards model using complementary log-log as a link function. In addition, a logistic model can be fitted using logit as a link function. The proposed methodology arises as an alternative to the score tests developed by Colosimo et al. (2000), where such models can be obtained for discrete binary data as particular cases from the Aranda-Ordaz distribution asymmetric family. These tests are thus developed with a basis on link functions to generate such a fit. The example that motivates this study was the dataset from an experiment carried out on a flax cultivar planted on four substrata susceptible to the pathogen Fusarium oxysoprum. The response variable, which is the time until blighting, was observed in intervals during 52 days. The results were compared with the model fit and the AIC values.

Late administration of a specific COX-2 inhibitor does not treat and/or prevent progression of gastric tumors in rats submitted to duodenogastric reflux procedure

PURPOSE:To assess whether late introduction of a specific COX-2 inhibitor (Meloxicam) can treat and/or prevent the progression of tumors in the stomach of rats submitted to duodenogastric reflux. METHODS: Seventy five male Wistar rats, weighing 150 grams, were submitted to the induction of duodenogastric reflux through the pylorus. At 36 weeks of follow-up were established three experimental groups: DGR36 sacrificed immediately, DGR54 and DGR54MLX both sacrificed at 54th week of follow-up . The animals of the latter group were fed with a rat chow premixed with Meloxicam (2.0 mg/ kg feed; 0.3 mg / kg bw / day) and the other two with standard rat chow. The lesions found in the pyloric mucosa and gastrojejunal anastomosis were analyzed macroscopically and histologically. For statistical analysis was adjusted a generalized linear model assuming a binomial distribution with LOGIT link function. RESULTS: No significant differences were found when comparing the incidences of benign tumor lesions (Adenomatous Hyperplasia), p=0.4915, or malignant (Mucinous Adenocarcinoma), p=0.2731, among groups. CONCLUSION: Late introduction of specific COX-2 inhibitor (Meloxicam) did not treat and was not able to prevent the progression of tumoral lesions induced by duodenogastric reflux in the rat stomachs.

Modelos lineares generalizados mistos na avaliação genética da prenhez precoce na raça Nelore

This study aimed to use the generalized linear models with probit and logit link function to evaluate early pregnancy, and to observe the effects on genetic variability and on sire selection when different ages are adopted in the definition of this trait. Early pregnancy was studied at 15 (EP15), and 21 (EP21) months. The analysis was done in R software. Pearson correlations (PC), between genetic predicted values and percentage of bulls in common considering only 10% of bulls with higher genetic values (TOP 10), between classification by logit and probit models and in each model among EP15 and EP21, were calculated. The heritability for EP15 and EP21 were close between models, except for EP15 using probit link function. PC and TOP10 among models were high. The Akaike and Bayesian criteria reported was similar between models. TOP10, considering the same model, among EP15-EP21 were moderated between EP15-EP21.

Detection of cavitated approximal surfaces using cone beam CT and intraoral receptors

Objectives: The aim of this study was to compare cone beam CT (CBCT) in a small field of view (FOV) with a solid-state sensor and a photostimulable phosphor plate system for detection of cavitated approximal surfaces. Methods: 257 non-filled approximal surfaces from human permanent premolars and molars were recorded by two intraoral digital receptors, a storage phosphor plate (Digora Optime, Soredex) and a solid-state CMOS sensor (Digora Toto, Soredex), and scanned in a cone beam CT unit (3D Accuitomo FPD80, Morita) with a FOV of 4 cm and a voxel size of 0.08 mm. Image sections were carried out in the axial and mesiodistal tooth planes. Six observers recorded surface cavitation in all images. Validation of the true absence or presence of surface cavitation was performed by inspecting the surfaces under strong light with the naked eye. Differences in sensitivity, specificity and agreement were estimated by analysing the binary data in a generalized linear model using an identity link function. Results: A significantly higher sensitivity was obtained by all observers with CBCT (p,0.001), which was not compromised by a lower specificity. Therefore, a significantly higher overall agreement was obtained with CBCT (p,0.001). There were no significant differences between the Digora Optime phosphor plate system and the Digora Toto CMOS sensor for any parameter. Conclusions: CBCT was much more accurate in the detection of surface cavitation in approximal surfaces than intraoral receptors. The differences are interpreted as clinically significant. A CBCT examination performed for other reasons should also be assessed for approximal surface cavities in teeth without restorations. © 2013 The British Institute of Radiology.

Efeito antinociceptivo, gastrointestinal e na atividade locomotora espontânea da administração intravenosa e metadona em equinos

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Modelagem estatística do frete da madeira no Uruguai

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

Estrutura da população e epidemiologia de Moniliophthora roreri no Magdalena Medio Colombiano

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

On Wald Residuals in Generalized Linear Models

A rigorous asymptotic theory for Wald residuals in generalized linear models is not yet available. The authors provide matrix formulae of order O(n(-1)), where n is the sample size, for the first two moments of these residuals. The formulae can be applied to many regression models widely used in practice. The authors suggest adjusted Wald residuals to these models with approximately zero mean and unit variance. The expressions were used to analyze a real dataset. Some simulation results indicate that the adjusted Wald residuals are better approximated by the standard normal distribution than the Wald residuals.

Impact of weather and climate on the incidence of acute coronary syndromes

BACKGROUND: First investigations of the interactions between weather and the incidence of acute myocardial infarctions date back to 1938. The early observation of a higher incidence of myocardial infarctions in the cold season could be confirmed in very different geographical regions and cohorts. While the influence of seasonal variations on the incidence of myocardial infarctions has been extensively documented, the impact of individual meteorological parameters on the disease has so far not been investigated systematically. Hence the present study intended to assess the impact of the essential variables of weather and climate on the incidence of myocardial infarctions. METHODS: The daily incidence of myocardial infarctions was calculated from a national hospitalization survey. The hourly weather and climate data were provided by the database of the national weather forecast. The epidemiological and meteorological data were correlated by multivariate analysis based on a generalized linear model assuming a log-link-function and a Poisson distribution. RESULTS: High ambient pressure, high pressure gradients, and heavy wind activity were associated with an increase in the incidence of the totally 6560 hospitalizations for myocardial infarction irrespective of the geographical region. Snow- and rainfall had inconsistent effects. Temperature, Foehn, and lightning showed no statistically significant impact. CONCLUSIONS: Ambient pressure, pressure gradient, and wind activity had a statistical impact on the incidence of myocardial infarctions in Switzerland from 1990 to 1994. To establish a cause-and-effect relationship more data are needed on the interaction between the pathophysiological mechanisms of the acute coronary syndrome and weather and climate variables.