On the impact of disproportional samples in credit scoring models: An application to a Brazilian bank data

Statistical methods have been widely employed to assess the capabilities of credit scoring classification models in order to reduce the risk of wrong decisions when granting credit facilities to clients. The predictive quality of a classification model can be evaluated based on measures such as sensitivity, specificity, predictive values, accuracy, correlation coefficients and information theoretical measures, such as relative entropy and mutual information. In this paper we analyze the performance of a naive logistic regression model (Hosmer & Lemeshow, 1989) and a logistic regression with state-dependent sample selection model (Cramer, 2004) applied to simulated data. Also, as a case study, the methodology is illustrated on a data set extracted from a Brazilian bank portfolio. Our simulation results so far revealed that there is no statistically significant difference in terms of predictive capacity between the naive logistic regression models and the logistic regression with state-dependent sample selection models. However, there is strong difference between the distributions of the estimated default probabilities from these two statistical modeling techniques, with the naive logistic regression models always underestimating such probabilities, particularly in the presence of balanced samples. (C) 2012 Elsevier Ltd. All rights reserved.

Predicting smear negative pulmonary tuberculosis with classification trees and logistic regression: a cross-sectional study

Abstract Background Smear negative pulmonary tuberculosis (SNPT) accounts for 30% of pulmonary tuberculosis cases reported yearly in Brazil. This study aimed to develop a prediction model for SNPT for outpatients in areas with scarce resources. Methods The study enrolled 551 patients with clinical-radiological suspicion of SNPT, in Rio de Janeiro, Brazil. The original data was divided into two equivalent samples for generation and validation of the prediction models. Symptoms, physical signs and chest X-rays were used for constructing logistic regression and classification and regression tree models. From the logistic regression, we generated a clinical and radiological prediction score. The area under the receiver operator characteristic curve, sensitivity, and specificity were used to evaluate the model's performance in both generation and validation samples. Results It was possible to generate predictive models for SNPT with sensitivity ranging from 64% to 71% and specificity ranging from 58% to 76%. Conclusion The results suggest that those models might be useful as screening tools for estimating the risk of SNPT, optimizing the utilization of more expensive tests, and avoiding costs of unnecessary anti-tuberculosis treatment. Those models might be cost-effective tools in a health care network with hierarchical distribution of scarce resources.

Latent residual analysis in binary regression with skewed link

Model diagnostics is an integral part of model determination and an important part of the model diagnostics is residual analysis. We adapt and implement residuals considered in the literature for the probit, logistic and skew-probit links under binary regression. New latent residuals for the skew-probit link are proposed here. We have detected the presence of outliers using the residuals proposed here for different models in a simulated dataset and a real medical dataset.

Germline mutation analysis of Tpit in Poodle dogs with ACTH-dependent hyperadrenocorticism

There is a high incidence of pituitary-dependent hyperadrenocorticism (PDH) in Poodle dogs, with family members being affected by the disease, suggesting a genetic involvement. Tpit is an obligate transcription factor for the expression of pro-opiomelanocortingene and for corticotroph terminal differentiation. The aim of the present study was to screen the Tpit gene for germline mutations in Poodles with PDH. Fifty Poodle dogs (33 female, 8.71 +/- 2.8 years) with PDH and 50 healthy Poodle dogs (32 females, 9.4241 2.8 years) were studied. Genomic DNA was isolated from peripheral blood, amplified by PCR and submitted to automatic sequence. No mutation in the coding region of Tpit was found, whereas the new single nucleotide polymorphism p.S343G, in heterozygous state, was found in the same frequency in both PDH and control groups. We concluded that Tpit gain-of-function mutations are not involved in the etiology of PDH in Poodle dogs.

Computed tomographic dacryocystography in children undergoing balloon dacryoplasty

PURPOSE To ascertain whether the volume and circumference of the lacrimal sac and nasolacrimal duct as measured by contrast-enhanced computed tomographic dacryocystography (CT-DCG) before and after balloon dacryoplasty could be used to predict clinical success in children with congenital nasolacrimal obstruction. METHODS Nasolacrimal ducts of children aged 2 to 6 years with clinical signs of congenital nasolacrimal duct obstruction undergoing balloon dilation were imaged with contrast-enhanced CT-DCG before and 5 minutes after the procedure. The circumference of the most dilated portion of the lacrimal sac was measured on the axial plane. The volume of contrast within the nasolacrimal duct and sac was also measured before and after the procedure. Clinical success was defined as the disappearance of signs of epiphora. RESULTS A total of 18 nasolacrimal ducts of 13 children were included. The average circumference of the most dilated portion of the lacrimal sac was 1.30 +/- 0.45 cm (range, 0.64-2.50 cm) before the procedure. The average contrast volume was 0.12 +/- 0.08 cm(3) (range, 0.01-0.38 cm(3)) before and 0.07 +/- 0.06 cm(3) (range, 0.01-0.20 cm(3)) after (P = 0.01). Data were analyzed using multivariate logistic regression with a backward variable input model; a decrease in contrast volume before and after dilation (P = 0.04) was associated with clinical success, whereas the larger size of the most dilated portion of the lacrimal sac (P = 0.01) was associated with clinical failure. CONCLUSIONS Contrast-enhanced CT-DCG provides useful information about nasolacrimal anatomy in children with congenital nasolacrimal duct obstruction. The decrease in contrast volume before and after balloon dilation was predictive of success; A larger size of the most dilated portion of the lacrimal sac was associated with clinical failure. (J AAPOS 2012;16:464-467)

Instrumental activities of daily living performance in healthy and cognitively intact seniors from a Brazilian sample and its relation to age and other socio-demographic variables

Background: Studies on functional capacity in community-dwelling older people have shown associations between declines in instrumental activities of daily living (IADL) and several factors. Among these, age has been the most consistently related to functional capacity independent of other variables. We aimed at evaluating the performance of a sample of healthy and cognitively intact Brazilian older people on activities of daily living and to analyze its relation to social-demographic variables. Methods: We conducted a secondary analysis of data collected for previous epidemiological studies with community-dwelling subjects aged 60 years or more. We selected subjects who did not have dementia or depression, and with no history of neurological diseases, heart attack, HIV, hepatitis or arthritis (n = 1,111). Functional capacity was assessed using the Brazilian version of the Older American Resources and Services Questionnaire (BOMFAQ). ADL performance was analyzed according to age, gender, education, and marital status (Pearson's chi(2), logistic regression). Results: IADL difficulties were present in our sample, especially in subjects aged 80 years or more, with lower levels of education, or widowed. The logistic regression analysis results indicated that "higher age" and "lower education" (p <= 0.001) remained significantly associated with IADL difficulty. Conclusions: Functional decline was present in older subjects even in the absence of medical conditions and cognitive impairment. Clinicians and researchers could benefit from knowing what to expect from older people regarding IADL performance in the absence of medical conditions.

Factors associated with preterm birth in Campina Grande, Paraiba State, Brazil: a case-control study

A case-control study (2008-2009) analyzed risk factors for preterm birth in the city of Campina Grande, Paraiba State, Brazil. A total of 341 preterm births and 424 controls were included. A multiple logistic regression model was used. Risk factors for preterm birth were: previous history of preterm birth (OR = 2.32; 95% CI: 1.25-4.29), maternal age (OR = 2.00; 95% CI: 1.00-4.03), inadequate prenatal care (OR = 2.15; 95% CI: 1.40-3.27), inadequate maternal weight gain (OR = 2.33; 95% CI: 1.45-3.75), maternal physical injury (OR = 2.10; 95% CI: 1.22-3.60), hypertension with eclampsia (OR = 17.08; 95% CI: 3.67-79.43) and without eclampsia (OR = 6.42; 95% CI: 3.50-11.76), hospitalization (OR = 5.64; 95% CI: 3.47-9.15), altered amniotic fluid volume (OR = 2.28; 95% CI: 1.32-3.95), vaginal bleeding (OR = 1.54; 95% CI: 1.01-2.34), and multiple gestation (OR = 22.65; 95% CI: 6.22-82.46). High and homogeneous prevalence of poverty and low maternal schooling among both cases and controls may have contributed to the fact that socioeconomic variables did not remain significantly associated with preterm birth.

Clinical and sociodemographic factors in a sample of older subjects experiencing depressive symptoms

Objectives This study aims to determine the frequency of clinically significant depressive symptoms (CSDS) in a community sample of older Brazilians and to examine their relationship with sociodemographic factors, cognitive and functional impairment (CFI), and medical illness. Methods A total of 1145 subjects aged 60?years or older living in the City of Ribeirao Preto, State of Sao Paulo, Brazil, were interviewed. The following instruments were used: a 10-item scale for screening of depressive symptoms in older people, the mini mental state examination, the Fuld Object Memory Evaluation, the Informant Questionnaire on Cognitive Decline in the Elderly, the Bayer Activities of Daily Living Scale, and a sociodemographic and clinical questionnaire. Results The frequency of CSDS was 15.7%. Logistic regression analysis indicated that being previously depressed, having CFI, having lower level of education, using psychotropics, and not engaging in physical exercise were related to CSDS. On the other hand, being a woman, older, medically ill, employed, or married was not associated with CSDS. Conclusions Consistent with previous reports, lower education, lack of physical activity, and CFI were significantly associated with higher frequencies of CSDS. Further investigations are necessary to clarify the occurrence of depression and possible modifiable factors in developing countries such as Brazil. Copyright (C) 2011 John Wiley & Sons, Ltd.

MTP -493G/T gene polymorphism is associated with steatosis in hepatitis C-infected patients

The reduction of hepatic microsomal transfer protein (MTP) activity results in fatty liver, worsening hepatic steatosis and fibrosis in chronic hepatitis C (CHC). The G allele of the MTP gene promoter, -493G/T, has been associated with lower transcriptional activity than the T allele. We investigated this association with metabolic and histological variables in patients with CHC. A total of 174 untreated patients with CHC were genotyped for MTP -493G/T by direct sequencing using PCR. All patients were negative for markers of Wilson’s disease, hemochromatosis and autoimmune diseases and had current and past daily alcohol intake lower than 100 g/week. The sample distribution was in Hardy-Weinberg equilibrium. Among subjects with genotype 1, 56.8% of the patients with fibrosis grade 3+4 presented at least one G allele versus 34.3% of the patients with fibrosis grade 1+2 (OR = 1.8; 95%CI = 1.3-2.3). Logistic regression analysis with steatosis as the dependent variable identified genotypes GG+GT as independent protective factors against steatosis (OR = 0.4, 95%CI = 0.2-0.8; P = 0.01). The results suggest that the presence of the G allele of MTP -493G/T associated with lower hepatic MTP expression protects against steatosis in our CHC patients.

Role of Two Clusters of Male Alcoholics in Treatment Retention

Aims: This study aimed to classify alcohol-dependent outpatients on the basis of clinical factors and to verify if the resulting types show different treatment retention. Methods: The sample comprised 332 alcoholics that were enrolled in three different pharmacological trials carried out at Sao Paulo University, Brazil. Based on four clinical factors problem drinking onset age, familial alcoholism, alcohol dependence severity, and depression - K-means cluster analysis was performed by using the average silhouette width to determine the number of clusters. A direct logistic regression was performed to analyze the influence of clusters, medication groups, and Alcoholics Anonymous ( AA) attendance in treatment retention. Results: Two clusters were delineated. The cluster characterized by earlier onset age, more familial alcoholism, higher alcoholism severity, and less depression symptoms showed a higher chance of discontinuing the treatment, independently of medications used and AA attendance. Participation in AA was significantly related to treatment retention. Discussion: Health services should broaden the scope of services offered to meet heterogeneous needs of clients, and identify treatment practices and therapists which improve retention. Information about patients' characteristics linked to dropout should be used to make treatment programs more responsive and attractive, combining pharmacological agents with more intensive and diversified psychosocial interventions. Copyright (C) 2012 S. Karger AG, Basel

Prevalence and Predictors of Potential Drug-Drug Interactions in the Elderly: A Cross-Sectional Study in the Brazilian Primary Public Health System

Purpose. The primary objective of this study was to investigate the prevalence of clinically important potential drug-drug interactions (DDIs) in elderly patients attending the public primary health care system in Brazil. The secondary objective was to investigate possible predictors of potential DDIs. Methods. A cross-sectional study was carried out in 5 Brazilian cities located in the Ourinhos Micro-region, Sao Paulo State, between November 2010 and April 2011. The selected sample was divided according to the presence (exposed) or absence (unexposed) of one or more potential DDIs (defined as the presence of a minimum 5-day overlap in supply of an interacting drug pair). Data were collected from medical prescriptions and patients' medical records. Potential DDIs (rated major or moderate) were identified using 4 DDI-checker programs. Logistic regression analysis was used to study potential DDI predictors. Results. The prevalence of clinically important potential DDIs found during the study period was 47.4%. Female sex (OR = 2.49 [95% CI 2.29-2.75]), diagnosis of = 3 diseases (OR = 6.43 [95% CI 3.25-12.44]), and diagnosis of hypertension (OR = 1.68 [95% CI 1.23-2.41]) were associated with potential DDIs. The adjusted OR increased from 0.90 [95% CI 0.82-1.03] in patients aged 60 - 64 years to 4.03 [95% CI 3.79 - 4.28] in those aged 75 years or older. Drug therapy regimens involving = 2 prescribers (OR = 1.39 [95% CI 1.17-1.67]), = 3 drugs (OR = 3.21 [95% CI 2.78-3.59]), = 2 ATC codes (OR = 1.19 [95% CI 1.12-1.29]), = 2 drugs acting on cytochrome P450 (OR = 2.24 [95% CI 2.07-2.46]), and ATC codes B (OR = 1.89 [95% CI 1.05-2.08]) and C (OR = 4.01 [95% CI 3.55-4.57]) were associated with potential DDIs. Conclusion. Special care should be taken with the prescription and therapeutic follow-up of patients who present characteristics identified as predictors. Knowledge of potential DDI predictors could aid in developing preventive practices and policies that allow public health services to better manage this situation.

Consumo de alimentos de origem animal e câncer de boca e orofaringe

OBJETIVO: Avaliar a relação entre alimentos de origem animal e câncer de boca e orofaringe. MÉTODOS: Estudo caso-controle, de base hospitalar, pareado por sexo e idade (± 5 anos) com a coleta de dados realizada entre julho de 2006 e junho de 2008. A amostra foi composta por 296 pacientes com câncer de boca e orofaringe e 296 pacientes sem histórico de câncer atendidos em quatro hospitais da cidade de São Paulo (SP), Brasil. Foi aplicado um questionário semiestruturado, para a coleta de dados relativos à condição socioeconômica e aos hábitos deletérios (tabaco e bebidas alcoólicas). Para avaliação do consumo alimentar, utilizou-se um questionário de frequência alimentar qualitativo. A análise se deu por meio de modelos de regressão logística multivariada, que consideraram a hierarquia existente entre as características estudadas. RESULTADOS: Entre os alimentos de origem animal, o consumo frequente de carne bovina (OR = 2,73; IC95% = 1,27-5,87; P < 0,001), bacon (OR = 2,48; IC95% = 1,30-4,74; P < 0,001) e ovos (OR = 3,04; IC95% = 1,51-6,15; P < 0,001) estava relacionado ao aumento no risco de câncer de boca e orofaringe, tanto na análise univariada quanto na multivariada. Entre os laticínios, o leite apresentou efeito protetor contra a doença (OR = 0,41; IC95% = 0,21-0,82; P < 0,001). CONCLUSÕES: O presente estudo sustenta a hipótese de que alimentos de origem animal podem estar relacionados à etiologia do câncer de boca e orofaringe. Essa informação pode orientar políticas preventivas contra a doença, gerando benefícios para a saúde pública.

CYBA C242T polymorphism is associated with obesity and diabetes mellitus in Brazilian hypertensive patients

Diabet. Med. 29, e55e61 (2012) Abstract Aims The CYBA C242T polymorphism has been associated with cardiovascular phenotypes such as hypertension and atherosclerosis, but available data are conflicting. This report investigated the impact of this variant on hypertension and metabolic determinants of cardiovascular risk in a large Brazilian sample. Methods We cross-sectionally evaluated 1856 subjects (826 normotensive subjects and 1030 hypertensive patients) by clinical history, anthropometry, laboratory analysis and genotyping of the CYBA C242T polymorphism. Results Genotype frequencies in the whole population were consistent with the HardyWeinberg equilibrium and genotype distributions were not different between hypertensive and normotensive subjects. Hypertensive patients with the CC genotype presented lower fasting plasma glucose levels (5.9 +/- 0.1 vs. 6.2 +/- 0.1 mmol/l, P = 0.020) and waist circumference (94.5 +/- 0.6 vs. 96.3 +/- 0.6 cm, P = 0.028) than CT + TT ones. Similarly, the prevalence of diabetes mellitus and obesity was also lower in hypertensive patients carrying the CC genotype (16% vs. 21%, P = 0.041; 36% vs. 43%, P = 0.029, respectively). In addition, multiple and logistic regression analysis demonstrated that the CYBA C242T polymorphism was associated with glucose levels, waist circumference, obesity and diabetes mellitus in hypertensive patients independently of potential confounders. Conversely, in normotensive subjects, no significant difference in studied variables was detected between the genotype groups. Conclusions These data suggest that the T allele of the CYBA C242T polymorphism may be used as a marker for adverse metabolic features in Brazilian subjects with systemic hypertension.

Thioredoxin interacting protein genetic variation is associated with diabetes and hypertension in the Brazilian general population

Objective: To investigate the relationship between TXNIP polymorphisms, diabetes and hypertension phenotypes in the Brazilian general population. Methods: Five hundred seventy-six individuals randomly selected from the general urban population according to the MONICA-WHO project guidelines were phenotyped for cardiovascular risk factors. A second, independent, sample composed of 487 family-trios from a different site was also selected. Nine TXNIP polymorphisms were studied. The potential association between TXNIP variability and glucose-phenotypes in children was also explored. TXNIP expression was quantified by real-time PCR in 53 samples from human smooth muscle cells primary culture. Results: TXNIP rs7211 and rs7212 polymorphisms were significantly associated with glucose and blood pressure related phenotypes. In multivariate logistic regression models the studied markers remained associated with diabetes even after adjustment for covariates. TXNIP rs7211 T/rs7212 G haplotype (present in approximately 17% of individuals) was significantly associated to diabetes in both samples. In children, the TXNIP rs7211 T/rs7212 G haplotype was associated with fasting insulin concentrations. Finally, cells harboring TXNIP rs7212 G allele presented higher TXNIP expression levels compared with carriers of TXNIP rs7212 CC genotype (p = 0.02). Conclusion: Carriers of TXNIP genetic variants presented higher TXNIP expression, early signs of glucose homeostasis derangement and increased susceptibility to chronic metabolic conditions such as diabetes and hypertension. Our data suggest that genetic variation in the TXNIP gene may act as a "common ground" modulator of both traits: diabetes and hypertension. (C) 2011 Elsevier Ireland Ltd. All rights reserved.

Clinical features of obsessive-compulsive disorder with hoarding symptoms: A multicenter study

Background: Factor analyses indicate that hoarding symptoms constitute a distinctive dimension of obsessive-compulsive disorder (OCD), usually associated with higher severity and limited insight. The aim was to compare demographic and clinical features of OCD patients with and without hoarding symptoms. Method: A cross sectional study was conducted with 1001 DSM-IV OCD patients from the Brazilian Research Consortium of Obsessive-Compulsive Spectrum Disorders (CTOC), using several instruments. The presence and severity of hoarding symptoms were determined using the Dimensional Yale-Brown Obsessive-Compulsive Scale. Statistical univariate analyses comparing factors possibly associated with hoarding symptoms were conducted, followed by logistic regression to adjust the results for possible confounders. Results: Approximately half of the sample (52.7%, n = 528) presented hoarding symptoms, but only four patients presented solely the hoarding dimension. Hoarding was the least severe dimension in the total sample (mean score: 3.89). The most common lifetime hoarding symptom was the obsessive thought of needing to collect and keep things for the future (44.0%, n = 440). After logistic regression, the following variables remained independently associated with hoarding symptoms: being older, living alone, earlier age of symptoms onset, insidious onset of obsessions, higher anxiety scores, poorer insight and higher frequency of the symmetry-ordering symptom dimension. Concerning comorbidities, major depressive, posttraumatic stress and attention deficit/hyperactivity disorders, compulsive buying and tic disorders remained associated with the hoarding dimension. Conclusion: OCD hoarding patients are more likely to present certain clinical features, but further studies are needed to determine whether OCD patients with hoarding symptoms constitute an etiologically discrete subgroup. (C) 2012 Elsevier Ltd. All rights reserved.