If there were no differences between humans, scientific research would be extremely easy. Any one subject could be exposed to an intervention, and that one subject’s intervention outcome could be extrapolated to the general population. Of course this is not the case, and we are faced with numerous challenges on how to infer study findings from our sample to our target population. Hence, the foundation for the development of biostatistics stems from the challenges of extrapolating useful real world data from a sample which is inherently biologically variable.

The transfer of “truth” from a study sample to the general population requires that the study has optimized both internal and external validity. The categorization of study criteria into internal and external validity provides a conceptual frame-work to organize data. They are by no means mutually exclusive, and factors among them often overlap. Internal validity refers to the degree to which results are true for the patients being studied. In general, factors related to internal validity are study protocol and the statistics utilized. Common detractors to internal validity for ED quantification include varied definitions of ED, inconsistent use of ED measurement tools that have been validated, time-frame of when to measure ED, patient report versus researcher acquisition of ED data, the use and reporting of phosphodiesterase inhibitors, study completion rates, and the role of chance. External validity refers to the degree to which study results can be generalized to the population outside the study.

Factors related to external validity include eligibility and exclusion criteria. Some of the most likely sources to decrease external validity in ED quantification include patient interest and partner availability; patient age; baseline sexual function; physician factors; the ability to spare nerves with CAP treatment; and comorbidities for ED.

The importance of study design and execution cannot be overstated. Others have advocated the use of a qualified biostatistician in study design and the analysis for trials in ED for specific issues of sample power, statistical modeling and design, covariate subgroup analyses, and effect size calculation. It is crucial to remember, however, that no amount of statistical methodology or power can compensate for biased data (garbage in = garbage out), or where necessary information was failed to be recorded. Graphically depicts the interplay of ED quantification data, biologic variability, and study validity.