Análisis mediante variables instrumentales para establecer causalidad en investigaciones observacionales / Analysis through instrumental variables to establish causality in observational research

La identificación de relaciones causales es uno de los problemas fundamentales de la investigación científica en medicina y es necesaria para poder ejercerla en forma efectiva. Sin embargo, desde el punto de vista práctico es difícil establecer la existencia de relaciones causales en estudios de carácter observacional, en gran parte por la presencia de factores de confusión. El análisis a través de variables instrumentales es una de las estrategias que permite controlar el efecto confundidor y documentar la presencia de relaciones causa-efecto en estas situaciones. En este artículo, el autor resume los principales supuestos del análisis a través de variables instrumentales, haciendo foco en la aleatorización mendeliana. (AU)

The identification of causal relationships is one of the fundamental challenges in scientific research in medicine and is necessary for its effective practice. However, from a practical standpoint, establishing the existence of causal relationships in observational studies is difficult, largely due to the presence of confounding factors. Analysis through instrumental variables is one of the strategies that allows to control the confounding effect and documenting the presence of cause-and-effect relationships in these situations. In this article, the author summarizes the main assumptions of analysis through instrumental variables, with a focus on Mendelian randomization. (AU)

Effect of periodontitis on diabetes: a review of cohort studies / 口腔疾病防治

@#Periodontitis is an inflammation that occurs in the supporting tissues around teeth with plaque biofilm as the starting factor. Periodontitis is closely related to many systemic diseases, among which the relationship between periodontitis and diabetes is the most widely reported. A cohort study is an essential clinical research method to explore the etiology. Large, well-conducted prospective cohort studies have high power, which can provide important clinical evidence for the impact of periodontitis on blood sugar control, incidence rate and complications of diabetes mellitus. Periodontitis is associated with the deterioration of glycemic control. At present, there is moderate evidence that nonsurgical periodontal treatment can significantly improve the blood sugar level of diabetes patients with periodontitis compared with no periodontal treatment. Studies on the impact of periodontitis on the incidence rate of diabetes lack consistent conclusions because of different population backgrounds. The evidence regarding whether periodontitis affects the incidence rate of diabetes complications is relatively limited. Therefore, well-designed cohort studies are needed to provide high-quality clinical evidence.

How to use analysis of variance correctly——an analysis of variance for the univariate quantitative data collected from the fractional factorial design / 四川精神卫生

The purpose of this paper was to introduce the fractional factorial design and its quantitative data analysis of variance and the SAS implementation. The fractional factorial designs were very similar to the factorial designs and the orthogonal designs， but they had some differences. The fractional factorial design required significantly fewer combinations of levels than the factorial design of the same size， and even saved sample size than the orthogonal design of the same size. In general， the precision of the results obtained by a fractional factorial design was lower than an orthogonal design and much lower than a factorial design. The fractional factorial design was suitable for the trial tests with many experimental factors， and its main purpose was to explore experimental factors that had a greater impact on the quantitative experimental results. When performing ANOVA and regression analysis on quantitative data with a fractional factorial design， it should be clear which factors or interactions had confounded effects.

Role of covariates in the analysis of causal mediation effects / 四川精神卫生

The purpose of this paper was to introduce the theoretical basis of the causal mediation effect analysis and the specific method to realize an example by the causal mediation effect analysis with SAS. The theoretical basis of the causal mediation effect analysis included the following two aspects， the basic concept and defining the counterfactual framework of the causal mediation effect. The example was about whether the encouraging environment provided by parents would affect the cognitive development of children. The traditional multiple linear regression analysis， the causal mediation effect analysis without considering covariates and with considering covariates were used， respectively. By comparing the results obtained by the three analysis methods， the following conclusions were drawn： ① when there were the mediation variables in the data， it was not suitable to use traditional multiple linear regression analysis to replace the causal mediation effect analysis； ② when there were covariates in the data， it was not suitable to conduct causal mediation analysis under the condition of ignoring covariates.

Gender differences on the WAIS-IV in patients with schizophrenia / Diferencias de género en la WAIS-IV en pacientes con esquizofrenia

Abstract Introduction Results from studies that have investigated gender differences in neuropsychological functioning in schizophrenia have been inconsistent. Differences in the illness stage, in the demographic and clinical characteristics of the samples, and the instruments used to measure cognition may have contributed to the heterogeneity in the results. Objective Investigate the heterogeneity in the results comparing cognitive functioning in chronically ill male and female patients with schizophrenia. Method Twenty-five women and twenty-five men chronically ill patients with schizophrenia matched on age, age at illness onset, and years of education were evaluated in cognitive functioning using the WAIS-IV. Results Men showed higher scores than women on the two global measures, on the perceptual reasoning and working memory indices, and on the information, visual puzzles, digit span, and arithmetic subtests of the WAIS-IV. Cohen's d effect sizes were high for the two global measures and the two indices (d > .68). Discussion and conclusion Overall, in chronically stable patients with diagnosis of schizophrenia, women's performance on cognitive functioning was below men's when assessed with the WAIS-IV, except in the case of processing speed. This pattern of gender differences is similar to the pattern observed in healthy populations. Our results can help to clarify the heterogeneity in the results from studies on gender differences in cognitive functioning in schizophrenia and may be valuable in designing cognitive-targeted interventions for schizophrenia.

Resumen Introducción Los resultados de los estudios que han investigado diferencias de género en funcionamiento neuropsicológico en la esquizofrenia han sido inconsistentes. Diferencias en la fase de la enfermedad, en las características demográficas y clínicas de las muestras y en los instrumentos utilizados podrían explicar esa heterogeneidad. Objetivo Investigar la heterogeneidad en los resultados comparando el funcionamiento cognitivo de pacientes con diagnostico de esquizofrenia. Método Veinticinco mujeres y veinticinco hombres pacientes con diagnóstico de esquizofrenia equiparados en edad, edad al inicio de la enfermedad y nivel educativo se evaluaron en funcionamiento cognitivo utilizando la WAIS-IV. Resultados Los hombres mostraron puntuaciones más altas que las mujeres en las dos medidas globales, en los índices de razonamiento perceptivo y de memoria de trabajo y en las subtests de la WAIS-IV de información, puzles visuales, amplitud de dígitos y aritmética. Los tamaños de efecto d de Cohen fueron altos en las dos medidas globales y en los dos índices (d > .68). Discusión y conclusión En conjunto, la ejecución de las mujeres en funcionamiento cognitivo está por debajo de la de los hombres cuando se mide con la WAIS-IV, excepto en el caso de la velocidad de procesamiento. Este patrón de diferencias de género es similar al patrón observado en población sana. Nuestros resultados pueden ayudar a clarificar la heterogeneidad de resultados en los estudios sobre diferencias de género en el funcionamiento cognitivo en la esquizofrenia y podrían ser útiles en el diseño de intervenciones centradas en la cognición.

COVID-19: The Ivermectin African Enigma / COVID-19: El enigma de la Ivermectina en Africa

Abstract Introduction: The low frequency of cases and deaths from the SARS-CoV-2 COVID-19 virus in some countries of Africa has called our attention about the unusual behavior of this disease. The ivermectin is considered a drug of choice for various parasitic and viral diseases and shown to have in vitro effects against SARS-CoV-2. Aims: Our study aimed to describe SARS-CoV2 infection and death rates in African countries that participated in an intensive Ivermectin mass campaign carried out to control onchocerciasis and compare them with those of countries that did not participate. Methods: Data from 19 countries that participated in the World Health Organization (WHO) sponsored African Programme for Onchocerciasis Control (APOC), from 1995 until 2015, were compared with thirty-five (Non-APOC), countries that were not included. Information was obtained from https://www.worldometers.info/coronavirus/ database. Generalized Poisson regression models were used to obtain estimates of the effect of APOC status on cumulative SARS-CoV-2 infection and mortality rates. Results: After controlling for different factors, including the Human Development Index (HDI), APOC countries (vs. non-APOC), show 28% lower mortality (0.72; 95% CI: 0.67-0.78) and 8% lower rate of infection (0.92; 95% CI: 0.91-0.93) due to COVID-19. Conclusions: The incidence in mortality rates and number of cases is significantly lower among the APOC countries compared to non-APOC countries. That a mass public health preventive campaign against COVID-19 may have taken place, inadvertently, in some African countries with massive community ivermectin use is an attractive hypothesis. Additional studies are needed to confirm it.

Resumen Introducción: La baja frecuencia de casos y muertes por el virus SARS-CoV-2 COVID-19 en algunos países de África llamó nuestra atención sobre el comportamiento inusual de esta enfermedad. La ivermectina se considera un fármaco de elección para diversas enfermedades parasitarias y virales, y se ha demostrado que tiene efectos in vitro contra el SARS-CoV-2. Objetivos: Nuestro estudio tiene el objetivo describir las tasas de infección y mortalidad del SARS-CoV-2 en los países africanos que participaron en una campaña intensiva masiva de ivermectina para el control de la oncocercosis y compararlas con las de los países que no participaron. Métodos: Los datos de 19 países que participaron en el Programa Africano para el Control de la Oncocercosis (APOC) patrocinado por la Organización Mundial de la Salud (OMS), desde 1995 hasta 2015, se compararon con 35 países que no fueron incluidos (NO APOC). La información sobre casos y muertes por COVID-19 se obtuvo de la base de datos https://www.worldometers.info/coronavirus/. Se utilizaron modelos de regresión de Poisson para obtener estimaciones del efecto del estado APOC sobre las tasas acumuladas de infección y mortalidad por SARS-CoV-2. Resultados: Después de controlar diferentes factores, incluido el Índice de Desarrollo Humano (IDH), los países APOC (frente a los no APOC) mostraron una mortalidad 28% menor (razón de tasas ajustada: RR= 0.72, IC 95%: 0.67-0.78) y una tasa de infección 8% menor (RR= 0.92, IC 95%: 0.91-0.93) por COVID-19. Conclusiones: Las tasas de mortalidad e infección son significativamente más bajas en países APOC en comparación con los países no APOC. Una campaña preventiva masiva de salud pública contra el COVID-19 pudo haber tenido lugar inadvertidamente en algunos países africanos con un uso masivo de ivermectina en la comunidad es una hipótesis atractiva. Se necesitan estudios adicionales para confirmarlo.

A comprehensive review of the healthy worker effect in occupational epidemiological studies

The reduction of mortality and morbidity rates among occupational cohort studies may be attributed to the presence of the healthy worker effect (HWE). Occupational epidemiologic studies investigating worker’s health are prone to the risk of having the HWE phenomenon and this special form of bias has been debated over the years. Hence, it’s imperative to explore in-depth the magnitude and sources of HWE, and further, elucidate the factors that may affect HWE and strategies reducing HWE. The HWE should be considered as a mixed bias between selection and confounding bias. The validity threats due to the HWE among morbidity studies are the same as the mortality studies. The consequent reduction due to the HWE in the association between the exposure and outcome may lead to underestimating some harmful exposures in the workplace or occupational settings. Healthy hire effect and healthy worker survivor effect are the main sources of HWE. Several factors can increase or decrease the probability of HWE; therefore, the investigators should consider them among future occupational epidemiological studies. Many strategies can help in reducing the impact of HWE, but each strategy has its weaknesses and strengths. Not all strategies can be applied among all occupational epidemiological studies. Mathematical procedures still need further investigations to be validated. HWE is a consequence of inappropriate comparison groups in nature. The usage of the general population as a reference group is not an appropriate choice. By considering the HWE sources and factors and using appropriate strategies, the impact of HWE may be reduced.

Metodologías Cuantitativas 2: Sesgo de confusión y cómo controlar un confusor / Quantitative Methodologies 2: Confusion bias and how to control a confounding factor

Identificar el sesgo de confusión y cómo controlarlo es uno de los desafíos metodológicos más importantes en el diseño de estudios que buscan identificar la causalidad. Este sesgo está presente en cualquier análisis de la asociación entre una exposición y un resultado de interés, una asociación que puede estar sesgada o no por una tercera variable llamada confusor. Podemos diagnosticar un confusor en todos los casos en los que este crea una asociación espuria entre una variable de exposición o variable independiente y la variable de resultado o variable dependiente. Para controlar el sesgo de confusión, podemos usar diferentes métodos. Estos incluyen aquellas técnicas aplicadas en el diseño del estudio, tales como restricción, aleatorización y coincidencia, y aquellas técnicas empleadas en el análisis de datos, como la estratificación, el análisis multivariado, la estandarización, los puntajes de propensión, el análisis de sensibilidad y el inverso ponderación de probabilidad. En esta revisión, analizamos cómo identificar una variable de confusión y las principales técnicas para controlar el sesgo de confusión.

Addressing confounding bias is one of the challenges when conducting causality studies. This occurs when we report a causal association between an exposure and an outcome, when in fact it could be result of the effect of a third factor called confounding variable. That is, when a confounder variable creates a spurious relationship between the exposure or independent variable and the outcome of interest or dependent variable. By knowing the confounding variables and their association with the exposure of interest, the confounding bias could be controlled. To control for confounding bias, we can use different methods. These include techniques applied in study design, such as restriction, randomization, and coincidence, and techniques used in data analysis, such as stratification, multivariate analysis, standardization, propensity scores, analysis sensitivity and the inverse probability weighting. In this review, we discuss how to identify a confounding variable and the main techniques for controlling for confounding bias.

Serie de Redacción Científica: Estudios Trasversales / Scientific Writing Series: Cross-Sectional studies

Los estudios transversales se caracterizan por la medición simultánea de la exposición y el outcome de interés. Son el diseño idóneo para estimar prevalencias, analizar la precisión diagnóstica de una prueba y validar instrumentos, para lo cual es esencial controlar los sesgos de información, selección y confusión ya sea por diseño o por análisis. Asimismo, es crucial escoger la medida de asociación idónea para cada outcome de interés, llámese el odds ratio para eventos raros y la razón de prevalencias para los eventos frecuentes. Finalmente, para su redacción y publicación se recomienda revisar las guías STROBE y STARD

Cross-sectional studies are characterized by the simultaneous measurement of exposure and the outcome of interest. They are the ideal design to estimate prevalence, analyze the diagnostic accuracy of a test and validate instruments, for which it is essential to control information bias, selection, and confusion either by design or by analysis. It is also crucial to choose the appropriate association measure for each outcome of interest, call the odds ratio for rare events, and the prevalence ratio for frequent events. Finally, for its writing and publication, it is recommended to review the STROBE and STARD guidelines

Lost in Translation: Sesgos y la información en la pandemia de COVID-19 / Lost in Translation: Biases and information in the COVID-19 pandemic

The COVID-19 pandemic has produced an endless stream of information in relation to data, analysis and projections of all aspects of this disease. One of the characteristics of scientific information is that it is full of uncertainties, which are not correctly disclosed to the general population. In this essay, we describe the main cognitive biases that cause people to modify their perception of risk, where the most important are memory, novelty, and emotionality or affection. In addition, we review frequent errors that have been made in terms of collection, dissemination and analysis of information by scientific communicators, experts in other areas and the media in general. Ideally, these biases and confounders should be known to all who participate in the flow of information on the health problems of this pandemic, in order to integrate the uncertainty inherent in the data and the critical analysis of the information received.

La pandemia COVID-19 ha producido un caudal interminable de información en relación a datos, análisis y proyecciones de todos los aspectos de esta enfermedad. Una de las características de la información científica es que está llena de incertidumbres, las que no se divulgan correctamente a la población general. En este ensayo, hacemos una descripción de los principales sesgos cognitivos que hacen que las personas modifiquen su percepción de riesgo donde los más importantes son el de memoria, de novedad y emocionalidad o afecto. Además, revisamos errores frecuentes que se pueden cometer en términos de recolección, difusión y análisis de la información de parte de comunicadores científicos, expertos en otras áreas y medios de comunicación en general. Idealmente, estos sesgos y confundentes deben ser conocidos por todos los que participan en el flujo de información de los problemas de salud de esta pandemia, de manera de poder integrar la incertidumbre inherente a los datos y el análisis crítico de la información que se recibe.

Common bias and its control in real-world study / 中国临床药理学与治疗学

The widespread availability real-world study (RWS) offers valuable insights into disease treatment, disease management, and socio-economic status in routine practice, as well as cautionary tales and methodological challenges such as the discovery of sample heterogeneity and bias of data and its correction. This paper summarizes the common bias and its control in the process of design, implementation and analysis for RWS in order to promote the standardization and rationality of the implementation of RWS.

Compreendendo os métodos de escore de propensão na pesquisa clínica / Understanding the methods of propensity score in clinical research

Os métodos de escore de propensão são a probabilidade de um sujeito receber um tratamento condicional em um conjunto de características de base (confundidores), sendo usado para comparar pacientes com distribuição similar de fatores de confusão, de modo que a diferença nos resultados forneça estimativa imparcial do efeito do tratamento. Esta revisão mostra os conceitos básicos dos escore de propensão e fornece orientação na implementação de métodos de propensão, além de outros, como estratificação, ponderação e ajuste de covariáveis, tornando-se uma guia prático para o clínico

The propensity score methods are the probability of a subject receiving conditional treatment on a set of baseline characteristics (confounders), and are used to compare patients with similar confounding distributions, so that the difference in results provides an unbiased estimate of the treatment effect. This review shows the basic concepts of propensity scores, and provides guidelines for the implementation of propensity methods, and others based on it, such as stratification, weighting, and adjustment of covariables, becoming a practical guide for the clinician

Mendelian Randomization Analysis in Observational Epidemiology

Mendelian randomization (MR) in epidemiology is the use of genetic variants as instrumental variables (IVs) in non-experimental design to make causality of a modifiable exposure on an outcome or disease. It assesses the causal effect between risk factor and a clinical outcome. The main reason to approach MR is to avoid the problem of residual confounding. There is no association between the genotype of early pregnancy and the disease, and the genotype of an individual cannot be changed. For this reason, it results with randomly assigned case-control studies can be set by regressing the measurements. IVs in MR are used genetic variants for estimating the causality. Usually an outcome is a disease and an exposure is risk factor, intermediate phenotype which may be a biomarker. The choice of the genetic variable as IV (Z) is essential to a successful in MR analysis. MR is named ‘Mendelian deconfounding’ as it gives to estimate of the causality free from biases due to confounding (C). To estimate unbiased estimation of the causality of the exposure (X) on the clinically relevant outcome (Y), Z has the 3 core assumptions (A1-A3). A1) Z is independent of C; A2) Z is associated with X; and A3) Z is independent of Y given X and C; The purpose of this review provides an overview of the MR analysis and is to explain that using an IV is proposed as an alternative statistical method to estimate causal effect of exposure and outcome under controlling for a confounder.

Association between atherogenic index of plasma and renal function impairment in patients with gout / 中华内分泌代谢杂志

Objective@#To study the relationship between atherogenic index of plasma (AIP)and renal impairment in male patients with gout.@*Methods@#A retrospective analysis of 821 male subjects was conducted to measure the relevant biochemical indicators and to calculate the AIP, endogenous creatinine-clearance rate (Ccr), and estimated glomerular filtration rate (eGFR). EpiData 3.1 software was used for data entry, SPSS21.0 was used for statistical analysis, and GraphPad Prism 6.0 software was used for charts.@*Results@#Compared with control group, AIP, serum uric acid, triglyceride in gout group were significantly higher (all P<0.01), while eGFR and high density lipoprotein-cholesterol were significantly lower (both P<0.05). The composition ratio of renal function impairment in gout group was significantly higher (P<0.01). With the increase of AIP level, eGFR level decreased and serum creatinine level increased, but the overall difference was not statistically significant (P>0.05), while Ccr and serum uric acid levels gradually increased (P<0.05). Logistic regression analysis after adjusting for various confounding factors showed that AIP, triglyceride, and serum uric acid were risk factors for renal function damage in patients with gout (P<0.05), the relevant risk were 7.030, 1.291, 1.004 respectively. After adjusting confounding factors, the associations between triglyceride, serum uric acid and renal function injury risk changed little, while AIP showed more evident, the OR value increased from 2.629 to 6.265 and 7.030.@*Conclusions@#(1)AIP is closely related to the renal function damage of patients with gout. After adjusting various confounding factors, AIP can better reflect the renal function damage than other indicators, which is of great significance to predict the renal function damage of patients with gout. (2)That patients with gout with high uric acid level may suffer from renal atherosclerosis and have a higher risk of renal impairment. (3)Dynamic observation of AIP in gout patients is helpful for early identification of the risk of renal failure in such patients.

Association between atherogenic index of plasma and renal function impairment in patients with gout / 中华内分泌代谢杂志

Objective To study the relationship between atherogenic index of plasma ( AIP ) and renal impairment in male patients with gout. Methods A retrospective analysis of 821 male subjects was conducted to measure the relevant biochemical indicators and to calculate the AIP, endogenous creatinine-clearance rate (Ccr), and estimated glomerular filtration rate ( eGFR) . EpiData 3.1 software was used for data entry, SPSS21.0 was used for statistical analysis, and GraphPad Prism 6. 0 software was used for charts. Results Compared with control group, AIP, serum uric acid, triglyceride in gout group were significantly higher (all P<0.01), while eGFR and high density lipoprotein-cholesterol were significantly lower ( both P<0.05) . The composition ratio of renal function impairment in gout group was significantly higher (P<0.01). With the increase of AIP level, eGFR level decreased and serum creatinine level increased, but the overall difference was not statistically significant ( P>0.05) , while Ccr and serum uric acid levels gradually increased (P<0.05). Logistic regression analysis after adjusting for various confounding factors showed that AIP, triacylglycerol, and serum uric acid were risk factors for renal function damage in patients with gout (P<0.05), the relevant risk were 7.030, 1.291, 1.004 respectively, After adjusting corfounding factors, the associafions betwees triglyceride, serum uric acid and with renal function injury risk changed little, while AIP show more evident, the OR value increased from 2.629 to 6.265 and 7.030. Conclusions (1)AIP is closely related to the renal function damage of patients with gout. After adjusting various confounding factors, AIP can better reflect the renal function damage than other indicators, which is of great significance to predict the renal function damage of patients with gout. ( 2) That patients with gout with high uric acid level may suffer from renal atherosclerosis and have a higher risk of renal impairment. ( 3) Dynamic observation of AIP in gout patients is helpful for early identification of the risk of renal failure in such patients.

Application of parametric g-formula in causal analysis / 中华流行病学杂志

At present, traditional methods on statistics have limitations in controlling time- varying confounding. This paper introduces an analysis method, parametric g-formula, which would adjust time-varying confounding, and also exemplifies the steps of its implementation for purpose to provide a new reference for researchers to deal with long-term observational data.

The propensity score method and it′s application / 中华预防医学杂志

Propensity score method, as an analytical strategy for adjusting multiple covariates, has been widely used in observational comparative effectiveness research. This paper introduces this method covered basic principles, case illustration and software implementation, in order to help readers understand propensity score method, apply it correctly in their researches, improve the efficiency of data utilization, and enhance the level of statistical analysis.

Sensitivity analysis method for unmeasured confounding interference in observational study / 中华流行病学杂志

Objective@#To introduce the methods for sensitivity analysis, discuss and compare the advantages and disadvantages of different methods.@*Methods@#The difference between confounding function method and bounding factor method in accuracy of identifying unmeasured confounding factors in observational studies through simulation trials and actual clinical data was compared.@*Results@#The results of simulation trials and actual clinical data showed that when there was unmeasured confounding between exposure (X) and outcome (Y), the results of confounding function and the bounding factor analysis were similar in terms of the effect of unmeasured confounding factor to lead to the complete change of the magnitude and direction of the observed effect value. However, the confounding function method needed smaller confounding effect to fully interpret the observed effect value than the bounding factor needed. In addition, the bounding factor method needed to analyze two confounding parameters, while only one parameter was needed in the confounding function method. The confounding function method was simpler and more sensitive than the bounding factor method.@*Conclusion@#For real-world observational data, the sensitivity analysis process is essential in analyzing the causal effects between exposure (X) and outcome (Y). In terms of the calculation process and result interpretation the sensitivity analysis method of confounding function is worth to recommend.

Aplicación de grafos acíclicos dirigidos en la evaluación de un set mínimo de ajuste de confusores: un complemento al modelamiento estadístico en estudios epidemiológicos observacionales / Directed acyclic graphs in statistical modelling of epidemiological studies

Background: Confusion in observational epidemiological studies distorts the relationship between exposure and event. "Step by step" regression models, diverts the decision to a statistical algorithm with little causal basis. Directed Acyclic Graphs (DAGs), qualitatively and visually assess the confusion. They can complement the decision on confounder control during statistical modeling. Aim: To evaluate the minimum set of confounders to be controlled in a cause-effect relationship with the use of "step-by-step regression" and DAGs, in a study of arsenic exposure. Material and Methods: We worked with data from Cáceres et al., 2010 in 66 individuals from northern Chile. The interindividual variability in the urinary excretion of dimethyl arsenic acid attributable to the GSTT1 polymorphism was estimated. A causal DAG was constructed using DAGitty v2.3 with the list of variables. A multiple linear regression model with the step-by-step backwards methodology was carried out. Results: The causal diagram included 12 non-causal open pathways. The minimum adjustment set corresponded to the variables "sex", "body mass index" and "fish and seafood ingest". Confusion retention of the multivariate model included normal and overweight status, gender and the interaction between "water intake" and GSTT1. Conclusions: The use of DAG prior to the modeling would allow a more comprehensive, coherent and biologically plausible analysis of causal relationships in public health.

Survey of the use of statistical methods in Journal of the Korean Association of Oral and Maxillofacial Surgeons

OBJECTIVES: This study aimed to describe recent patterns in the types of statistical test used in original articles that were published in Journal of the Korean Association of Oral and Maxillofacial Surgeons. MATERIALS AND METHODS: Thirty-six original articles published in the Journal in 2015 and 2016 were ascertained. The type of statistical test was identified by one researcher. Descriptive statistics, such as frequency, rank, and proportion, were calculated. Graphical statistics, such as a histogram, were constructed to reveal the overall utilization pattern of statistical test types. RESULTS: Twenty-two types of statistical test were used. Statistical test type was not reported in four original articles and classified as unclear in 5%. The four most frequently used statistical tests constituted 47% of the total tests and these were the chi-square test, Student's t-test, Fisher's exact test, and Mann-Whitney test in descending order. Regression models, such as the Cox proportional hazard model and multiple logistic regression to adjust for potential confounding variables, were used in only 6% of the studies. Normality tests, including the Kolmogorov-Smirnov test, Levene test, Shapiro-Wilk test, and Scheffé's test, were used diversely but in only 10% of the studies. CONCLUSION: A total of 22 statistical tests were identified, with four tests occupying almost half of the results. Adoption of a nonparametric test is recommended when the status of normality is vague. Adjustment for confounding variables should be pursued using a multiple regression model when the number of potential confounding variables is numerous.