Computational protocol: Prevalence and factors associated with violence and abuse of older adults in Mexico’s 2012 National Health and Nutrition Survey

Similar protocols

Protocol publication

[…] We conducted a secondary data analysis using data from Mexico’s 2012 NHNS, which is a probabilistic survey with multistage and stratified sampling that allows for a representative sample of the population at the state and national level, as well as by urban and rural strata. The sample comprised of 8,894 adults who are 60 years of age or older. Due to the expanding factor calculated in the NHNS design, this sample is representative of 10,747,490 older adults. The response rate for completed surveys was 87 % [].Surveyors asked the older adults about episodes of violence such as being the victim of robbery, aggression, assault, abuse or other types of violence in the last 12 months. For the positive responses to these questions, an initial analysis was conducted to describe the variables of violence (by unknown party) and abuse (by member of the family). However, given the low frequency of older adults who reported abuse and the difficulty conducting a rigorous statistical analysis, we decided to group the two variables together, creating the dependent variable (binary) labeled “violence/abuse”. The independent variables included in analysis were: age, sex, living with five or more people in the same house, head of household’s condition, education, having a partner, diagnosed with a non-communicable chronic disease (diabetes, hypertension, cardiovascular disease, other heart disease, or cancer), being diagnosed with depression by a health professional, having limited function (at least one limitation to daily activities such as walking, bathing, lying down/standing up, getting dressed or having a limitation in instrumental daily activities such as preparing food, buying food, taking medication, or managing money), cognitive deterioration (being unable to draw a clock, remembering 1 or 2, or no words), low self-esteem (self-classification as valuable individuals or not) and loss of power within the family (not consulted for important decisions or household finances or no contribution to household income). We also used region and socioeconomic status (SES) as an independent variable, and used the NHNS household and SES classification described by Gutiérrez and colleagues in 2013 [] in line with the income deciles previously defined in the database, which corresponded to low (deciles 1 and 2), medium (deciles 3 to 6) and high (deciles 7 to 10). Finally, the country was divided into four geographic regions: North, Central-Western, Central and South-SoutheastWe conducted a descriptive analysis of the variables with confidence intervals of 95 %. Considering the sampling strategy utilized for the survey permitted unbiased estimates. To calculate the variance, we used the mean total score for strata with a single sampling unit. The sample size varied based on the analysis variable. For the analysis of the variables of most interest with the population of older adults, using both men and women, we utilized the STATA command “subpop” (subpopulation).We also evaluated the association of independent variables with the violence/abuse variable by using the chi-squared test or Fisher’s exact test, and crude odds ratios in the bivariate analysis. Finally, we used multiple logistic regression models to find adjusted estimates. This was done because of the complex survey design used for the NHNS; one for men and another for women. All statistical analysis was performed using STATA version 13.0 []. […]

Pipeline specifications

Software tools Stata, Subpop
Application Population genetic analysis
Organisms Martes pennanti, Homo sapiens