Job Description

It's been a fantastic first week at the WHO, and it is hard to believe an eighth of the internship has already gone by. I have been learning and soaking in so much, both from conversations and meetings with the staff here and from my own daily work. Here's what I've been up to:

As I mentioned in my previous post, the GLC approves projects in countries around the world burdened by MDR-TB in order to implement treatment programs to manage the disease. In just eight years, the GLC has expanded from a couple projects to more than 60 projects in 52 countries. Each year, the GLC requests patient outcome data from each project to analyze where treatment has been effective, where it has not, and why it has not. For instance, a project may have provided less hospital-based care, may have run out of the GLC-recommended drug and decided to use the "next best," or countless other reasons. It is vital that the data be submitted to the GLC punctually and accurately, so that problems can be corrected and policies can be adapted.

But the first problem is that the WHO cannot force it's member countries to do anything, and likewise the GLC cannot force its constituents to reveal MDR-TB patient outcome data that may not be very flattering (side note: during one year, one country indeed decided it did not want to submit any of its MDR-TB data because there was so high a disease burden that it would allegedly have made the country look bad). So there are lots of politics involved. Where do I fit into the picture? This week, I have been corresponding with the project overseers for entire countries to request their projects' annual patient outcome data. Personal emails to project managers and the WHO heads of TB management in China, Philippines, and Vietnam, for example.

The second problem is that projects sometimes submit data to the GLC that contains discrepancies. For starters, the GLC requests each project to monitor its patients longitudinally throughout the 2 year treatment, classifying each patient under categories like "still on treatment," "defaulted," "cured," and "died," and without changing the number of patients in each cohort. Each year, a new cohort of patients begin treatment; and each year, every cohort is reevaluated. But sometimes discrepancies occur. For example, what if a cohort of 164 patients who start treatment in 2007 suddenly contains 170 patients in the 2008 data? The difference of six patients means these six patients should have been put in a new 2008 cohort when they started treatment, but were instead placed in the 2007 one. This distorts cure rates, which in turn affects the measured efficacy of that project's treatment program. That's a big problem. As another example, what if the number of patients "still on treatment" rises while the number who are recorded to have "defaulted" on their treatment drops? Once you've defaulted, you've defaulted, so the only way for a defaulter to resume treatment should be by starting from scratch with the next year's cohort of patients rather than with his original cohort. The rise in "still on treatment" and the drop in "defaulted" means that patients who stopped taking their necessary drugs and dosages at some point (due to really bad side effects, etc.) are put back into a category where they resume taking their medications with patients who have adhered all along, instead of (correctly) starting treatment from scratch. What are you left with? Improper treatment, which exacerbates drug resistance and messes with the measured efficacy of the treatment program. Again, big problems. So where do I fit in here? I'm finding these discrepancies and contacting the corresponding country projects to inquire why these discrepancies occurred.

Then, once I get all the data, I'll be figuring out just how effective these projects really are. What if a country's data reveals a cure rate of 50 percent? Sweet! An entire half of the patients will be totally free of MDR-TB if they undergo the GLC's programmatic treatment! That's great, isn't it? No. The cure rate for MDR-TB patients if you simply give them food, clean water, and sleep is 30 to 40 percent. In other words, a 50 percent cure rate means that programmatic treatment is only slightly better than doing nothing at all. That's a problem. If the GLC's getting these projects access to high-quality drugs, then other on-the-ground factors are clearly contributing to the poor cure rate (e.g. less hospital care, doctors giving patients multiple low-quality drugs because they run out of the high-quality one, etc.). But what about if the country's cure rate is 60 percent? Now surely that's an awesome cure rate, no? No. That's only par with the world average--a world that is accustomed to late detection, low-quality drugs, and low-quality monitoring. Once again, if the GLC's getting these projects access to high-quality drugs, then other on-the-ground factors are clearly at play. My job, once I get to that point, will be to investigate these factors by asking project managers the right questions. If we find something that could be done more in line with WHO recommendations and guidelines, then more patients are cured. If the project data shows somehow that some guidelines (e.g. drug recommendations, drug forecasting procedures, etc.) aren't actually optimal, then policies are adapted. Either way, it's a win-win.