Journal of North Khorasan University of Medical Sciences
Year:2011 | Volume:3 | Issue:BIOSTATISTICS AND EPIDEMIOLOGY SUPPLEMENT|Papers Count:18

Background & Objectives One of the popular studies in medical sciences for finding risk factors and the reason of the disease are case control studies that the important index we can calculate is odds ratio.but some confounders which effect on response variable challenge the OR’s validity and present OR more or less than the real value.One way of omitting the effect of confounder is designing matched studies. Logistic regression is one of the general methods for modeling these studies. This article compares 3 logistic regression models: independence, marginal and conditional.Materials & Methods: This study has been conducted on correlated simulated data. Thus the data is simulated from bivariate normal distribution with the correlation coefficients (0, 0.2, 0.4, 0.6, and 0.8). Then with changing cut-off points at (0.05, 0.25), (0.25, 0.1), (0.25, 0.15), (0.25, 0.25) for their c.d.f, we convert continues distribution to categorical binary distribution which data are related together. Then 3 logistic regression model in independence, marginal and conditional version fit to data and calculate OR. With 10000 times iteration, we compare 2.5 and 97.5 percentiles values and the median OR percentile value at the above cut-off points for all their models.Results: When the correlation is zero, all three models have the same quantity for OR and also changing in point have the same coefficient.But with the increasing correlation between the observations, OR between marginal model and independence model is not different. But its value will vary with the conditional model. For example, the cut-off point (0.25, 0.1) and when the correlation is 0.6, median of OR that obtained in marginal model and independence model is 2.8, but in conditional model this quantity is 5 and it is twice of fitted value.Conclusion: When the correlation between observations is high, using of conditional model is more correctly method and with increasing this correlation, our error rate by using independence or marginal model rises. But when correlation between observations is negligible, using of the three models gives similar estimates.

Background & Objectives: Logistic regression is a general model to determine the relationship between covariates and binary response variables. Artificial neural network model is an alternative flexible model which can be used in these cases, too. This study aimed to make a comparison between the predictions of ANN and logistic regression model for binary outcome of medical data.Material & Methods: Data gathered from 639 registered gastric cancer patients between January 2002 and October 2007 at the Research Center for Gastroenterology and Liver Diseases of Shahid Beheshti University of Medical Sciences, Tehran, Iran. Stage of disease was considered as the dependent variable. Network performance was assessed by using of least square error of prediction and then concordance indexes and area under receiver operative characteristic curves (AUROC) were used to comparison of neural network and logistic regression models. Data analysis was performed by R 2.12 software.Results: Results showed that the concordance index of ANN and LR for drug treatment was calculated as 0.771 and 0.710; respectively.In addition AUROC for ANN and LR models were 0.725 and 0.699, respectively. The difference between the values of observed and predicted of the dependent variable by two models was significant (P=0.002).Conclusion: As a result, the total accuracy prediction of the ANN model is better than LR model, so this model is suggested to predict the stage of gastric cancer disease and also diagnostic goals.

Background & Objective: Infertility has several adverse effects on body, psychological and social aspects of women. Otherwise, body image could be a predictor of different behaviors including sexual behavior. Thus, in the present study we aimed to assess the association between body image with sexual function and marital adjustment in fertile and infertile women.Materials & Methods: One-hundred and thirty fertile and 130 infertile women were enrolled in this study. Statistical analysis was performed by using AMOS 18 with structural equation modeling.Results: In fertile women, the strongest path coefficient was related with the effect of body image on general health, while in infertile women, the strongest path coefficient was associated with the effect of body image on sexual function. In both occupying and non-occupying women the strongest path coefficient was related with the effect of body image on general health. In those infertile women who had the supports of their partners, the strongest path coefficient was associated with the effect of body image on sexual function, and in those infertile women who had not the supports of their partners; the strongest path coefficient belonged to the effect of sexual function on marital adjustment.Conclusion: Based on our results, it seems that using structural equation modeling in evaluating and recognition of direct, indirect and total effects of the similar model is absolutely necessary and can be a good alternative method instead of regression.

Background & Objective: Incidence and prevalence of cancer is increasing worldwide, especially in the low and middle income countries, including I.R o Iran. Valid data is needed to measure burden of cancer and take appropriate decision towards cancer control. Due to limitation of cancer registrations it is hard to make an inference about cancer in Iran. It is important to measure how complete and valid is the report of cancer incidence and mortality.Material & Method: We prepared mortality registry (source 1) and cancer registry (source 2) data from the health deputy in Fars province. In addition, we collected cancer cases from different hospitals (source3) between 2000 and 2005. We followed the cancer patients from the date of diagnosis up to the date of death.. We selected 10 most common cancer types in this study. If any of the patients died during 2003-2005 enter to this study. We compared mortality data from the three sources using capture recapture method using log linear modeling.Result: We collected 44863 data in which 9401was evaluated and finally 2232 enter to the study. Sensitivity of the mortality registry was 58% and based on the best model estimated number of cases was 3309 instead of the observed number.Conclusion: Coverage of the mortality registry for cancer death was weak leading to an underestimation in the mortality rate of the Fars Province. Improvement of the mortality registry is needed before making any firm decision about burden of cancer.

Background & Objectives: Cox proportional hazard regression model is the most common and standard method to study the effect of potential covariates on the survival times. However, under certain conditions such as availability of a good information about the system under study, parametric models are known to be more efficient than the Cox models. In this study, we aim to evaluate the efficiency of the semi-parametric Cox models and alternative parametric models and nonparametric Kaplan-Meier estimates in analyzing the survival times of patients suffering from acute myocardial infraction Material & Methods: This study is a prospective-cohort research with the total number of 650 participant's experienced acute myocardial infarction. The patients followed up for one year in order to analyze their survival times. Three approaches; non-parametric, semi-parametric and parametric models, were used to model the survival times. To evaluate the relative goodness of fit of each procedures, Akaike information criterion (AIC) was used.Results: During one year of the following up, the total number of 67 deaths were reported. Out of 460 initial patients, 48% were smoking, 33.4% suffered from hypertension, 17.7% diabetes, 15.3 Hyperlipidemia, 30% airtime, and 10.3% had block. A broad agreement was observed between parametric and Cox models. Both models associated similar covariates as significant factors. However, the non-parametric model suggested high age , non-used streptokinase and block significantly increased the risk of death.Conclusion: However, the hazard ratio in Cox model and parametric models were similar and results of nonparametric Kaplan Meier approach were largely similar but, The AIC suggested the parametric Weibull model as the most appropriate model for modeling the survival times of patients with acute myocardial infarction.

Background & objectives: In the common Cox proportional hazards model one of the basic assumptions is independence between censoring time and event time. In clinical studies, when censoring is caused by competing risks or patient withdrawal, there is always a concern about the validity of treatment effect estimates that are obtained under the assumption of independent censoring.Introduction a solution for checking the assumption of independent and extending the Cox model for dependent censoring is useful.Material & Methods: To achieve this goal, we use copula function and extend the Cox model. This model can also perform a sensitivity analysis for checking the assumption of independent. We generalize the likelihood function in R software and estimate the parameters by using an iteration algorithm. We apply the proposed method to the data of breast cancer patients at Ghaem and Omid hospitals and check the independence assumption.Results: We show by Simulation that this algorithm works well and show the effect of dependent censoring on estimated parameters. We apply the proposed method to the data of breast cancer patients at Ghaem and Omid hospitals and show that the independence assumption is true.Conclusion: Not assumption of independence causes bias in estimation in Cox model and the bias is dependent on the degree of association between event and censoring times. In dataset of breast cancer patients the primary diagnosis is an important role in Success in the treatment.

Background & Objective: The use of clustering methods for the discovery of cancer subtypes has drawn a great deal of attention in the scientific community. While bioinformaticians have proposed new clustering methods that take advantage of characteristics of the gene expression data, the medical community has a preference for using "classic" clustering methods. There have been no studies thus far performing a large scale evaluation of different clustering methods in this context.Method & Material: We present CCK index for assessing clustering result of gene expression data. This index was made by combining two arbitrary classification and clustering algorithms result and finally.the first large-scale analysis of nine different clustering methods, Hierarchical clustering with Single, Average, Complete and Ward linkages, UPGMA, Diana, K-means, PAM and CLARA methods for the analysis of 5 cancer gene expression data sets.Afterward we use Margin Trees method for assessing quality of result of clustering methods. Ultimately we calculate quality of result of clustering methods via Kappa coefficient between result of clustering methods and result of Margin Tree method for each clustering methods.Results: Our results reveal that the PAM, followed closely by CLARA, exhibited the best performance in terms of recovering the true structure of the data sets. Also we found that Partitioning clustering methods (PAM, CLARA and K-means) have better performance than Hierarchical clustering methods (Hierarchical clustering with Single, Average, Complete and Ward linkages, UPGMA and Diana).Conclusion: The validation technique was used in this paper (Margin Trees) can aid in the selection of an optimal algorithm, for a given data set, from a collection of available clustering algorithms.

Background & Objectives: Numbers of older people rapidly increase. Therefore it is important to the health of the elderly.Social capital by providing emotional and mental supports plays an effective role in promoting mental health in older people. So this study examines the factors affecting mental health and its relationship with social capital in the older people of Tehran 9 district.Materials and Methods: This study is a cross-sectional study that was conducted in Tehran 9 district on the older people over 60 years in 2011.320 participants in this study were selected by multistage random sampling. To assess mental, health General Health Questionnaire (GHQ-28) and to assess social capital, Urban HEART Questionnaire were used. To determine components of social capital, used from exploratory factor analysis.Results: factor analysis determined three factors include individual trust, social cohesion–support and social cohesion support.That these three components, explained 68.1 percent of the total variance of social capital. There was relationship between mental health and individual trust and social cohesion and support. But there was not relationship between mental health and social trust and collective relations.Conclusion: Whatever the social capital is more, mental health is also better. On the other hand individuals with better mental health, qualified to social responsibility and role in society. This shows that there is a mutual relationship between mental health and social capital. So the promotion each of them can effect on other.

Background & Objectives: Technology advances in this century, especially, in molecular generics yields high volume, high dimensional data. This creates many unprecedented challenges for statisticians who are responsible for analysis of such data. Although logistic regression method is quite popular in association analysis in medical researches but it has some serious limitations in handling high dimensional data. In present study, our goal is introduce a modern model-free statistical method called random forest that we believe is able to overcome difficulties of the classical statistical methods in finding association between predictors and a trait.Material & Methods: In this study, the nonparametric random forest technique was employed to determine the important factors associated with ankylosing spondylitis (AS) disease. Genetic materials including information on HLA-B27 status (positive/negative) and 12 polymorphisms of the ERAP-1 gene were collected on 401 patients and 316 healthy controls. The data were analyzed both with the logistic regression method and random forests technique and the results were compared.Results: Based on a stepwise logistic regression, HLA-B27 and rs28096 polymorphism were significantly associated with the disease. However, using the random forests technique, we found that HLA-B27 and rs1065407 were the main factors associated with diseases and in fact rs28096 polymorphism becomes the third in importance ranking.Conclusion: The results from our study indicate some discrepancies between logistic regression and random forest analyses of high-dimensional data such as the genetic data that we are dealing here. Although logistic regression is quite popular, easy to employ, and is a predominant statistical method among researchers, but it has some serious limitations. On the other hand, more modern statistical such random forest enjoy a more methodological sophistication and yield more accurate and reliable results.Therefore, researchers should be aware of such alternatives and should use these alternatives accordingly and as situation arise in screening tests especially in genetic data analyses.

Background & Objectives: Multilevel (hierarchical) modeling is effective method for analyzing medicals data that set in more than one level. Multilevel Modeling is generalization of linear modeling in which regression coefficients are themselves given a model, whose parameters are also estimated from data. In this paper we want to illustrate about theoretical aspects and estimating method in three level modeling and application of this method in longitudinal Blood Pressure (BP) data.Material & methods: Data of this longitudinal study were extracted from annual observations of the male workers of Isfahan’s Mobarakeh Steel Company (IMSC), collected in the Health and Safety Executive office of the company between 2003 and 2009. In this research, we assessed the effect of Shift Work (SW) on Diastolic BP (DPB) with controlling BMI and age. In this paper, MLwiN and SPSS software were used to apply a Multilevel Modeling.Results: This study consists of 6713 workers (45.2% regular day worker, 5.8% weekly rotating shift worker, and 49% routine rotating shift worker). In this study, after controlling confounding factor, SW not showed a significant association with DBP.Conclusion: High speed and high ability to fit models with high sample size are benefits of Iterative method rather likelihood method. And also suitable model of IMSC can be used to control the effect on BP on SW.

Background & Objectives: In in vitro fertilization techniques it is always better to transform a high quality embryo into uterus to enhance the chance of the success. Therefore, a comprehensive research was conducted to identify important factors related to the embryo quality. We considered potential correlation between embryos derived from an infertile couple. In addition, possible nonlinear relationships between the factors and the embryo quality were investigated.Materials& Methods: In this study, the statistical method of generalized additive models (GAM) has been employed to determine any nonlinear relationship between quantitative exploratory variables and the response.Then, the results were used to analyze the data using the Generalized Estimating Equation (GEE) technique.Result: Based on results of the GEE, linear relationships of drugs, type of infertility, and polycystic ovary syndrome was statistically significant on embryo quality. Interestingly, after examining the nonlinear relationships with GAM, in addition to the above variables, smoking status, sperm motility, and uterine factor were also shown to be significant.Conclusion: Since observations on a single subject in fertilization techniques are correlated and the relationship between factors and outcomes of fertilization may not be linear, the usual methods of GAM should incorporate the potential intra-cluster correlation using models such as the GEE.

Background & Objectives: Logistic Regression is a general model for medical and epidemiological data analysis. Recently few researchers have directed their studies to analysis of Logistic Regression with missing value at covariate variable. While the missing is a major threat in results authenticity of data set, in many studies the researchers face data with missing value and it is difficult to avoid such a case in studies.Material & Methods: Satten and Carroll, in the case of completely observed value of covariate variable and some covariate variable with missing at random mechanism (MAR), introduced a special likelihood function for parameters estimation of Logistic Regression model. In this research the above- mentioned likelihood function has been used in Bayesian analysis for parameters estimation of Logistic Regression model and the conclusions are compared with the Multiple Imputation method and Complete Case method.Results: The above-mentioned methods were applied on both simulation data and dentistry data and concluded that The parameters estimation from SCMCMC method had less variance in comparison with parameters estimation from Multiple Imputation and Complete Case methods.Conclusion: After comparison of the three mentioned methods results it had been concluded that if the mechanism is of missing at random the application of Bayesian analysis with MCMC causes to more accurate estimation and shorter Confidence Intervals than the Multiple Imputation method and Complete Case.

Background& Objective: These days, in many societies, fair and equality are the most significant concepts among the governments and peoples. The concept of fairness in the health systems, describes equal expenses of households in this area. The aim of this paper is to investigate the health expense inequality of households in the Iran provinces, with Gini index of inequality measurement and statistical methods.Material & Methods: In this paper we use the 2008 health expense data from statistical center of Iran. The process of preparing data and estimating the statistical indexes, (like mean and coefficient of variation, Lorenz curve and evaluating the Gini index) and making plots have been done with the use of S-PLUS software. For making the index comparing maps we have used ArcGIS 9.2 software.Results: By estimating the Gini index, plots and maps we found out that there is a remarkable inequality of health expenses in the provinces such as the Khorasan Razavi and Hormozgan by 0.71 and 0.92 who had the least and the most expenses between the provinces. On the other hand the mean of these expenses varies from 90000 to 630000 Rials and the poor provinces that are in the east borders of Iran have the least mean of this expense.Conclusions: The mean expenses of health in households decreases between states by going from east to the west of country. The household expenses of poor states have the least incomes and, the least facilities in health. On the other hand inequality in this expenses distribution is noticeable.Governments should decrease the inequality by increasing health facilities and financial aids in poor areas and in different ways such as different types of health insurance.

Background & Objectives: In medical researchers, there are lots of correlated data which cannot be analyzed using the usual classical statistical methods because the assumption of independency between observations is not met. Data from cluster sampling, longitudinal studies, observations on paired organs and matched studies are examples of such data.Materials & Methods: Two statistical methods that can be used to correctly handle these kinds of data are marginal and mixed models. These models are different in considering correlation between subjects and Interpretation of the regression coefficients.These models were compared in this paper.Results: The regression coefficients in marginal models with nonidentity link functions show the change in population whereas in mixed models they represent changes within a subject or a cluster.Conclusion: In result, in nonlinear models, application of these two kinds of models depends on the areas of their usage. While the marginal models are more attractive to the Health policy makers who are considering the potential effects of a variable on the population as a whole, the mixed model will be of most interest to a physician in a physician/patients context.

Background & Objective: In most countries, heart diseases are known as first cause of death among non-contagious diseases and in Iran it is one of the most important causes of death. Therefore assessing risk factors associated with survival of these patients is necessary. In this study, effect of some risk factors on hazard of death of subjects with acute myocardial infarction (AMI) is investigated.Materials & Methods: In this historical cohort study, 197 patients that experience AMI for the first time was studied which were admitted in Bushehr hospitals during 1997-2001. Aalen’s additive hazard model used in this study to assess the relation of death hazard rate after first AMI and risk factors like age at time of infarction, patient’s sex, systolic and diastolic blood pressure, serum total cholesterol, LDL and HDL.Results: Average survival time of patients in the present study was more than 4 years. The result showed that there is no significant relation between sex and survival time of patient (p>0.05) however age at time of infarction has a significant relation with survival of patient (p<0.05).Conclusion: According to Aalen’s additive hazard model as age at time of infarction increases, survival risk of AMI patients increases.

Background & Objectives: The analysis of age and survival probabilities, today, is several decades old. The calculation of age average in different groups of human society and analysis of factors which are effective on it to access methods for increase of individual’s health and age is one the daily concerns of scientists in different branches of biology and medicine. The objectives of this study were to assess the effects of prognostic factors on survival of patients with breast cancer using the Aalen additive hazards model, and to illustrate the advantage of Aalen’s plot.Materials & Methods: during 2005-2007, 133 women with breast cancer were treated in Fayazbakhsh hospital, Tehran. All patients followed until 2011 April. To analysis data, Aalen’s additive hazard model used. The graph of estimated cumulative regression function (Aalen s graph) was draw to analyze the effect of independent variable during time.Results: Significant factors in univariate analysis was entered into multivariable analysis of Allen`s additive hazard model.HER2, third tumor malignancy grade and the number of positive lymph nodes more than10 were recognized as significant factors in patient`s survival in multivariable model.Conclusion: Additive hazard model is used as a suitable substitution for Cox model. In which standard deviation of estimations will be smaller than Cox model.

Background & Objectives: The increase of mortality rate of gastric cancer in Iran and World in recent years, reveal necessity of studies on this disease. In this study, Hazard Function for gastric cancer patients was estimated using Wavelet and Kernel methods to compare, and some related factors with the Hazard Function were assessed.Material & Methods: Ninety-five gastric cancer patients in Fayazbakhsh Hospital between 1996 and 2003 were studied. The effects of age of patients, gender, stage of disease and treatment method on patients’ lifetime were assessed. For data analyses, survival analyses using Wavelet method and Log-rank test in R software were used.Results: Nearly 25.3% of patients were female and the rest were male. Fourteen percent had surgery treatment and the rest had treatment without surgery. Three forth of patients died and the rest were censored. Almost 9.5% of patients were in early stage of disease, 53.7% in Locally Advance stage and 36.8% in Metastatic stage. The Hazard Function estimation in Wavelet method showed significant difference for stage of disease (P<0.001) and did not reveal any significant difference for age, gender and treatment method.Conclusion: Only stage of disease had effect on hazard and most of patients diagnosed in the late stages of disease, which is possibly one of the most reasons for high hazard rate and low survival of these patients. Therefore, it seems to be necessary a public education about symptoms of disease by media and regular tests and screening for early diagnosis of this disease.

Background & Objectives: This study was performed to surveillance visible goiter. Iodine deficiency disorders (IDD) are among the most important health-nutrition problems. In the most area where exposed with iodine deficiency the only visible Complication is goiter.Material & Method: This study was performed with the data of the national study of health survey (1999) in Iran. In this survey 53633 people were selected and examined thyroid gland. In order to study risk factors of visible goiter Chi-square test and multiple mixed logistic regressions have been used.Result: Based on this study it has been shown that 32.1 percent of people have a kind of goiter. And 4.5 percent of people had visible goiter. Frequency of goiter was significantly different in the age categories, different sex and in rural and urban (p-value<0.001).The 6.1 percent of females have goiter and odds ratio of the goiter for women was 2.29 higher than males (CI95%=(2.09, 2.50)). Also 5.4 percent of rural people have goiter and adjusted odds ratio of the goiter for rural people was 1.39 higher than urban people (CI95%=(1.28, 1.50)). In this study 6 percent of 13 to 18 years old people have goiter which was the most prevalence of goiter among the age categories.Conclusion: Consider the result of this study, there should be more consideration on the nutrition of 13 to 18 age category, female and rural to prevent the goiter