First, read the question stem thoroughly. You may cover up the answers to force yourself to read the stem first. Read every word. Don't skip ahead. Make a mental or physical note of key data.
Recognize terms and definitions (use your vocabulary).
Identify the point of the question. What is the disease or the scientific principle being tested?
Synthesize the data. Is there a connection?
Determine what the question is asking. By now you may have a good idea, even without having looked at the question foils. All that remains is uncovering the answers and picking the one you have already decided upon.
But I can't seem to get the point of the question, and nothing looks familiar! Let it go...let it go. It is better to move onto other questions than get stuck. You may be able to answer many more questions in the time wasted on a single question. After you have gotten into the rhythm of the exam and have encountered questions you can answer with ease, it is often the case that you can go back to difficult questions and find them easier. If you spend too much time on challenging questions, you will not have enough time to answer easier questions.
I can recognize some parts of the stem, but I still don't have a clear answer! In that case, make sure you have read all the information. It is best to try and work forwards from the data, but if stuck, work backwards from the foils. Try to exclude foils that you know are less probable. It is often the case that students are left with two choices that are difficult to separate. A 50/50 is still better than 20% or less. There is no magic formula at this point, just go with what you think is best. Mark one and move on.
I have time left and I am agonizing over questions I have already answered. Should I change an answer? Overall, students who are doing well (at 80% and above) who go back and change answers to challenging questions may come out ahead, but only slightly, maybe 55/45. Students doing poorly on the exam will generally do worse by changing answers. In general, the only reasons to change answers: (1) you did not thoroughly read the question the first time, or (2) you now suddenly remember information that you can apply.
There must be some trick to this! Don't go there. People under stress have a tendency to resort to magical thinking. Avoid that tendency. Continue to apply your knowledge. Physicians stay in control. Medicine is based upon science. Well constructed examinations assess knowledge, not gaming. If you want to play that game, the 'house' always wins.
Most improvements in test performance are incremental. It is harder to perform considerably better than one has already demonstrated, but it is far more likely that extraneous stresses will bring the score down. More of the same habits yields more of the same results. Something has to change.
Forcing yourself to learn more content in a short time doesn't work, because the brain has biochemical and bioelectric functions subject to the laws of physics. Trying to contain entropy (chaos) for extended, intense periods (cramming) is counterproductive, because synapses form at near zero-order kinetics, and because entropy will inevitably occur, possibly when you least want it (during the exam).
In general, the students who have significantly improved their performance are doing so, not because of academic reasons (studying harder) but because of elimination of non-academic problems, and they are performing at their real potential.
In reality, medical students are already performing at a high level, and differences are marginal. Well constructed examinations for medical licensure are designed to measure achievement of a standard curriculum and not produce a rank order. It is possible for everyone to score 100%, though that becomes harder with assessments of a broad knowledge base.