Systematic reviews are secondary investigations that compile published results that have been obtained from studies involving human subjects. Meta-analysis is the term used to describe the carrying out of statistical analysis of the combination of the results of two or more original studies, which had to be selected from a systematic review. In this way, a meta-analysis cannot exist without a systematic review. Systematic reviews arise due to the exponential increase in the information; to provide all health personnel with a study that critically analyzes the results and discriminates those that may be useful in clinical practice. Systematic reviews are one of the fundamental tools in evidence-based medicine, in which two of the main steps refer to both the search and the critical analysis of the studies, which shall support medical decisions on aspects that are mainly related to diagnosis, treatment, or prognosis. On the other hand, systematic reviews have been essential for some time now when developing evidence-based clinical practice guidelines and they can be used to make decisions on health policies. The methodology for performing and interpreting systematic reviews and meta-analysis is described in this article.
Informar en los estudios sobre el tamaño del efecto de una intervención o del impacto de factor(es) sobre un desenlace, permite tomar mejores decisiones para la aplicación de los resultados a la práctica clínica. En este artículo se presenta la manera de analizar el tamaño del efecto, lo cual puede ser mediante métodos estadísticos directos o indirectos. Dentro de los métodos directos, se encuentra la diferencia de promedios entre grupos y la diferencia de frecuencias absolutas o relativas. Dentro de los métodos indirectos se muestran los índices de la familia de “d” de Cohen (que se basan en valores de desviación estándar), la familia de “r y R2”, medidas de asociación (RM, RR, HR) e impacto (NNT). La decisión del uso de cualquiera de los métodos descritos, depende de los objetivos del estudio, la escala de medición usada para evaluar los resultados y la distribución de los datos. Para facilitar la comprensión, se incluyen ejemplos y se resalta la necesidad de incluir los diferentes estadísticos con su nivel de precisión (ej. intervalos de confianza), junto con los umbrales clínicos de decisión, a fin de mejorar su interpretación.
Descriptive statistics is the branch of statistics that gives recommendations on how to summarize clearly and simply research data in tables, figures, charts, or graphs. Before performing a descriptive analysis it is paramount to summarize its goal or goals, and to identify the measurement scales of the different variables recorded in the study. Tables or charts aim to provide timely information on the results of an investigation. The graphs show trends and can be histograms, pie charts, “box and whiskers” plots, line graphs, or scatter plots. Images serve as examples to reinforce concepts or facts. The choice of a chart, graph, or image must be based on the study objectives. Usually it is not recommended to use more than seven in an article, also depending on its length.
4 March 2019 in International Journal for Equity in Health
Although acute lymphoblastic leukemia (ALL) 5 years survival in minors has reached 90%, socioeconomic differences have been reported among and within countries. Within countries, the difference has been related to the socioeconomic status of the parents, even in the context of public health services with universal coverage. In Mexico, differences in the mortality of children with cancer have been reported among sociodemographic zones. The Instituto Mexicano del Seguro Social (IMSS), the country’s main social security institution, has reported socioeconomic differences in life expectancy within its affiliated population. Here, the socioeconomic inequalities in the survival of children (< 15 years old) enrolled in the IMSS were analyzed. Five-year survival data were analyzed in cohorts of patients diagnosed with ALL during the period 2007–2009 in the two IMSS networks of medical services that serve 7 states of the central region of Mexico. A Cox proportional risk model was developed and adjusted for the socioeconomic characteristics of family, community of residence and for the clinical characteristics of the children. The slope of socioeconomic inequality of the probability of dying within five years after the diagnosis of ALL was estimated. For the 294 patients studied, the 5 years survival rate was 53.7%; the median survival was 4.06 years (4.9 years for standard-risk diagnosis; 2.5 years for high-risk diagnosis). The attrition rate was 12%. The Cox model showed that children who had been IMSS-insured for less than half their lives had more than double the risk of dying than those who had been insured for their entire lives. We did not find evidence of socioeconomic inequalities in the survival of children with ALL associated with family income, educational and occupational level of parents. However, we found a relevant gradient related social security protection: the longer children’s life insured by social security, the higher their probability of surviving ALL was. These results add evidence of the effectiveness of social security, as a mechanism of wealth redistribution and a promoter of social mobility. Extending these social security benefits to the entire Mexican population could promote better health outcomes.
Bayesian statistics is based on subjective probability. It works with evidence updating considering the knowledge acquired prior to an investigation, plus the evidence obtained thereof. Results’ interpretation requires for the hypotheses to be tested to be specified and their a priori probability to be estimated before the study. Study evidence is measured with the Bayes factor (compatibility ratio of the data under the proposed hypotheses). The conjunction of hypotheses a priori probabilities with the Bayes factor allows calculating the a posteriori probability of each one of them. The hypothesis with the highest degree of certainty at its update is the one that is accepted for decision making. In this review, three examples of hypothesis to be tested are shown: difference of means, correlation and association.