How global surgery research should be done

The following two tabs change content below.
Mark Shrime
Mark is a head and neck cancer and reconstructive surgeon in Boston, currently getting PhD in health policy from Harvard. He also works regularly in West Africa and writes on topics in global surgery.

Cross-posted from here.

This month, Adam Kushner and his group at Surgeons Overseas have published a spectacular pair of articles, in which they accomplish what so little other surgical disease burden research does.

Here’s what I mean:  Let’s say you wanted to figure out how many people in a country have a goiter. You could do this in a couple of ways.  You could ask every single person in the country—this would be the most accurate but would also be a horrifically silly waste of your time. Or you could find a way to create a random sample of people from the country which is representative of everyone else (this is the key), and then ask them.

The thing is: it’s really hard to create representative samples of an entire country—doing it the most strictly theoretically correct way would still require you to have access to each and every person in the country and then to pick from them at absolute random.  This is, clearly, insanely difficult in countries with well-developed infrastructures.  It’s impossible in countries without.

So, a lot of people cheat.  Instead of finding a random sample, many authors create a “convenience” sample of patients.  That is just what it sounds like—they pick the patients that they have the easiest access to.  For example, authors often survey patients who come to the hospital for treatment of surgical conditions, and then extrapolate from them to the disease burden in the entire country.

As you can imagine, this introduces terrible amounts of bias:  patients who seek treatment are more likely to live nearer to a hospital than further away, more likely to be wealthier and have better access to care than poorer, and potentially more likely also to have worse disease.  This convenience sample of patients tells you nothing about the country as a whole.

Unfortunately, as I said, this is how the bulk of the literature on surgical disease burden is done.

The method that Kushner and colleagues used—cluster randomized sampling—gets around a lot of these issues, and doesn’t require you to find every man, woman, and child in the country.  But it still takes time and money (and tedium) to do.  So props to Groen, Kushner, and the rest for doing this in Sierra Leone.

And even bigger props for combining it with the supply side of the equation.  These guys didn’t stop at determining just how much burden there was, but they also tried to figure out how good Sierra Leone’s infrastructure was at meeting at least part of that burden. If burden is the demand, then the ability to avert that burden is the supply.

The results—well, you can probably guess.  Surgical conditions are terribly prevalent, and the ability to treat any but the most minor of them, terribly absent. Kushner and I have not always seen eye-to-eye on the optimal way to tackle this unmet need (generally, and in West Africa in specific), but he and his group have done well to bring it to the fore.

Others are doing the same (see, for example, Sarika Bansal’s piece in the New York Times last week).  This is the sort of thing that we should do more of.

Here’s hoping this is just the start.