Biostatistics calculators for medical researchers
Odds ratio, relative risk, NNT, sensitivity and specificity, ROC curve and AUC, sample size, Cohen's kappa, intraclass correlation, Bland-Altman analysis. Each calculator gives its result with a confidence interval, a sentence to paste into your manuscript, and the R code that reproduces the numbers.
Your data never leaves your browser. No sign-up, no ads.11 calculators are ready and 4 more are on the way. Pick one below.
Risk & odds
Compare two groups on a yes/no outcome.
Sample size
How many participants do you need?
Sample size for a proportion
Participants needed to estimate a prevalence with a chosen margin of error.
ReadySample size: two means or two proportions
Participants per group to compare two groups with a chosen power.
ReadySample size: survival analysis
Events and participants for a log-rank comparison.
Ready
Diagnostic tests
How well does a test tell disease from no disease?
Agreement
Do two raters or two methods agree?
Survival & models
Time-to-event data and regression. These need raw data, so they come last.
Kaplan-Meier and log-rank test
Survival curves from your follow-up data.
Coming soonHazard ratio and Cox regression
Adjusted time-to-event comparison.
Coming soonLogistic regression
Adjusted odds ratios from raw data.
Coming soonTable 1 by group
Baseline characteristics table for a paper.
Coming soon
Every result comes with three things
A sentence for your manuscript
Written in AMA style from your numbers, plus the methods wording and the reporting items (STROBE, CONSORT, STARD, GRRAS) you should cover.
R code that reproduces it
Base R, filled in with your numbers. Run it and compare. If a number differs, you will see where.
A checked answer
Before a calculator goes online, its results are compared with Python (SciPy and statsmodels) on reference cases, and they must match to 4 significant digits. Read the verification log.
Confidence intervals: Woolf (odds ratio), Katz (relative risk), Newcombe-Wilson (risk difference, NNT), Wilson score (proportions), log method (likelihood ratios), DeLong (area under the ROC curve), Fleiss-Cohen-Everitt standard error (kappa), F distribution (intraclass correlation), t distribution (Bland-Altman limits).
Frequently asked questions
Which agreement calculator do I need?
For two raters putting subjects into categories use Cohen's kappa. For numeric ratings by two or more raters use the intraclass correlation. To compare two measurement methods in their own unit use the Bland-Altman analysis.
Is BiostatKit free, and is my data safe?
The calculators are free to use. Every calculation runs in your browser; nothing you type is uploaded. After a page loads, it sends only anonymous visit counts, never your numbers or your results (see the privacy policy).
How do I know the numbers are right?
Each calculator is checked against Python (SciPy and statsmodels) on reference cases, and each page gives R code that reproduces its result so you can check it yourself.
Which calculator do I need for a 2x2 table?
For an association in a case-control study use the odds ratio. For a cohort study or a trial use the relative risk, and the NNT if you want a number of patients. For a new test against a reference standard use sensitivity and specificity.
Which calculator do I need for a test that gives a number?
If you have a measurement, such as a blood level, and a reference-standard label for each person, use the ROC curve calculator. It gives the area under the curve with a 95% CI and the best cut-off, with the sensitivity and specificity at that cut-off.
Which sample size calculator do I need?
To estimate a prevalence with a chosen precision use the proportion calculator. To compare two groups on a measured or a yes/no outcome use the two-group calculator. For a time-to-event outcome compared with the log-rank test use the survival calculator.