Main
Authors: Pietro Ravani, Fabio De Gennaro, Fabio Malberti (MD, Cremona Hospital, Italy); Paolo Cogliati e Daniele Marcelli (MD, FMC Italy and Germany).
PREMISE.
The authors are aware that there is room for improvement, in order to make easier the use of the software and more understandable the meaning of the output. Therefore, any suggestion, comment, concern are welcome.
OBJECTIVES.
This calculator means to be a tool for an easier updating and not a SW for statistical analysis of raw data. It can be helpful for critically reviewing observational and interventional studies, as well as for evaluating the accuracy of a screening test.
The comparison of event-rates between randomised groups differing only in the experimental factor under study (all else being equal), should produce results similar to those reported by authors. Differences can be made by the univariable analyses of the calculator and by the lack of the effect of time-to-event.
Overt discrepancies can be caused by multivariable analyses needed for covariates adjustment between non randomised groups. In epidemiological studies, even opposite results can be produced for this reason. Therefore, in non-randomised interventional trials, in case of randomisation failure and for observational trials, the difference between the SW output results and those reported by authors’ analysis, can be explained by multivariable approaches.
USE.
Patients and events N need to enter into a 2X2 contingency table.
From mean follow-up of each group patient-years at risk and event rates and odds/100 pt-years are calculated. However, if the follow-up is not specified 12 months must be assumed.
According to the end-point of interest, different results can be obtained:
- negative end-point reduction measures;
- indicators of benefit increase;
NNT (benefit and harm) are calculated, with the 95% CI around each estimate. N refer to 1 year of treatment, independently of the trial follow-up.
Also for the diagnostic test performances analysis 4 data are required: the results of gold standard and those of the screening or diagnostic test. Efficacy indicators are calculated, with their 95% CI bounds.