Stata 12 statistical software

Descriptive analyses were undertaken to describe the status and trends of access to improved water sources and sanitation facilities over time and by location, demographic, and socio-economic factors. For analytical statistics, chi-squared tests with Bonferroni corrections were performed to examine the differences in the access to improved water sources and improved sanitation facilities. Three sets of binary logistic regressions were also conducted on data from 2000, 2006, and 2011 to examine the associations between the access to improved water sources and improved sanitation facilities and household characteristics (geographical regions, wealth index, ethnicity of household head, sex of household head, and educational level of household head).
A dichotomous variable, ‘water sources and sanitation facilities in combination’, was derived whereby: 1 if a household had both ‘improved water sources AND improved sanitation facilities’, and 0, if a household had ‘unimproved water source OR unimproved sanitation facilities’ and ‘unimproved water sources AND unimproved sanitation facilities’. The results of these binary logistic regressions were reported as odds ratios (OR) with 95% confidence intervals (CI). STATA statistical software 12.0 (Stata Corporation) was used to perform all statistical analyses. The level of statistical significance was set at 0.05.

First, Student’s t test was used to compare differences in PTPN22 and CSK mRNA expression between: patients and controls, patients and controls stratified according to their reference or risk allele for each SNP, as well as patients and controls stratified according to PTPN22 and CSK haplotypes. Differences in PTPN22 and CSK serum levels between RA patients and controls were also assessed by Student’s t test. Results were expressed as mean ± standard deviation for each study group. Next, covariance analysis was performed in order to adjust the results by potential confounding factors, including sex, age at time of study, and cardiovascular (CV) risk factors (hypertension, dyslipidemia, smoking, diabetes and obesity)^{33 (link)}. The association between PTPN22 and CSK mRNA expression in RA patients and their clinical characteristics was evaluated using Student’s t test or Pearson partial correlation coefficients (r) as required, after adjusting for the confounding factors commented above. In all cases, p-values < 0.05 were considered statistically significant. Regarding genotyping, Hardy-Weinberg equilibrium (HWE) was checked using Χ² test. The statistical power of the study is displayed in Supplementary Table 3. All these analysis were performed with STATA statistical software 12.0 (Stata Corp., College Station, TX, USA).

In all models, age, gender, BMI, smoking status, origin and education were considered potential confounders. Manual forward selection was used to develop fully adjusted models. Variables remained in the multivariable model when they were either significantly associated with the outcome (p ≤ 0.05) or confounded the association of the disease (dependent) with the outcome (i.e. ≥10% change in the coefficient). All analyses were performed on complete cases using STATA statistical software 12.0 [17 (link)].