Statistics notes from The BMJ

• Interpreting diagnostic accuracy studies for patient care
• Brackets (parentheses) in formulas
• How to obtain the P value from a confidence interval
• Comparisons within randomised groups can be very misleading
• Correlation in restricted ranges of data
• Analysis of continuous data from small samples
• Parametric v non-parametric methods for data analysis
• Missing data
• The cost of dichotomising continuous variables
• Standard deviations and standard errors
• Treatment allocation by minimisation
• Diagnostic tests 4: likelihood ratios
• The logrank test
• Interaction revisited: the difference between two estimates
• Validating scales and indexes
• Analysing controlled trials with baseline and follow up measurements
• Concealing treatment allocation in randomised trials
• Blinding in clinical trials and other studies
• …