When I first got interested in medicine, I also got interested in health policy. I was already out of college and so I decided to educate myself – not a bad discussion topic for med school interviews, after all – and found it much, much harder than expected. There were so many topics, so many opinions, and so little that seemed to connect up to give me the big picture. I had wanted a short, wide-reaching, neutral book and was surprised not to find one. Thus, a year and a half ago, I partnered with another medical student to write the book we wish we’d had.

In my journey from naïve pre-med then medical student and now to health care systems author, I’ve learned a lot about the American health care system, and wanted to share a few of the major points with you.

1. There is no American health care system. In fact, there are lots of them.  There’s the Department of Defense – a single payer, universal health care delivery system with employed providers similar to the UK.  There’s Medicare – a public insurance program that reimburses independent physicians and hospitals, similar to the system in Australia.  A big chunk of the population has no insurance at all – a situation usually found in the third world.  Then consider employer-sponsored insurance, Medicaid, individual markets, the Veterans Affairs system, HMOs, consumer-driven health plans, TRIcare, and a new system of government-subsidized health insurance exchanges.  We haven’t even started with the different arrangements seen in hospitals, clinics, physician groups, and with other health care providers.

This means that whenever anyone talks about “the American health care system” they are really talking about a mish-mash of systems that organically grew to meet our needs over time, not a planned organization. And keep in mind that any suggestions for reform must account for this decentralized mish-mash.

2. Figuring out the facts can be very tough. Anyone who has tried to search for physician salary averages has an inkling of what I mean here. Different sources give different numbers, and you’re left unsure of what to believe. As I researched topics for the book, I was surprised to find that there’s not just disagreement about what the facts mean, or how the facts should influence policy – there’s often disagreement about what the facts even are.

For example, we had a hard time ascertaining how Emergency Department (ED) usage has changed in Massachusetts when their health reform law was implemented in 2006. This was a major issue, since one of the commonly cited reasons for universal insurance is that it will reduce ED usage by uninsured patients, as these patients will (a) receive more preventive care, and (b) go to their primary care provider when sick instead of the ED. Question was: did this “common sense” view of behavior actually happen?

Well, it depends on the study. In the past year, one study said there was no change in ED usage, and another said there was a 3% decrease. The problem may be that we need to wait longer for the data to make more sense, but still – most people would be surprised that there is disagreement about what seems like it should be a simple count.  Issues like this arise in medicine and in health care all the time, and what should be cleared up by evidence continues as a debate about the issue and the evidence.

3. Comparisons can only go so far. Comparative effectiveness research pits treatments against each other to see which is more effective in treating patients. Whichever does better can then be used to form clinical practice guidelines that all doctors use to treat their patients. Sounds great, right? But what about if your patient is a 65 year old female with diabetes and cardiovascular disease, but the clinical practice guideline suggests a treatment that was tested mostly on men and women ages 40-60 who have never had cardiovascular problems? Does the same effectiveness apply?

And then there’s Pay For Performance (P4P), a recent innovation in how physicians are paid. Rather than paying a salary or a set fee for each procedure, P4P seeks to pay physicians a low base salary and then add bonuses for those providers whose patients are healthier or get healthier. Sounds great! Except how do you account for providers who see patients who are already sicker? Who are affected by poverty, or who won’t be active in their care, or who came to the doctor because none of the others she’d seen could help her and this one was supposed to be the best? Did that doctor really perform “worse” than her peers?

Then how about this. The USA clearly has some problems to fix in our health care system, yet countries like Germany and France have, by all measures, excellent health systems. Why not implement some of their systems here? Sounds great! But those countries have relatively more homogeneous populations, larger social welfare programs, denser environments, different medical education…and the list goes on. Point #2 shows us that even tiny details are important, and these are huge details. So how can we know that a program that works in France will work here, in a quite different context?

Comparisons may give us ideas, be illustrative, or seriously help guide policy and practice. But we should never assume they are equivalences, and we should always note the exceptions.

4. Everything is more complicated than you think. The above examples barely scratch the surface of the complications in learning about health care and how to make it better. Yes, it is frustrating to run into so many obstacles, to know so much yet know so little. The fact is that very little in health care is known. The way forward is instead based on our best guess about what works. So my recommendation is to approach research, policy, and opinion with a healthy dose of questioning, skepticism, and eye for detail.  This will put you a step ahead of the game, and, hopefully make your best guesses a little better than the rest.

Elisabeth Askin is a medical student and co-author of The Health Care Handbook: A Clear and Concise Guide to the U.S. Health Care System, also available in paperback.