Natural experiment. You take a whole bunch of people who have naturally received wounds, classify them as best you can by severity of wound, some people get suction, some people merely get bandages, and keep track of how long it takes people to heal. With enough people in the study (probably not all that many if the effect is as large as the article makes it out to be) any difference in individual healing time or minor problems in the mischaracterization of wounds will average out. Problems like this are exactly why statistical tests take into account things like standard deviation.
