Thursday, May 05, 2005

Dissertation Progress

In order to make myself think about my dissertation project regularly and to force myself to construct tangible ideas, I have decided to use this space for updates on my thesis.

When I left Michigan for medical school last summer, I took a data set and a dissertation question along with me. The ideas was to simply finish up the analysis of that data on my own, write up and send in chapters via email to my advisor/committee.

Coming off months of thurough training in SAS, SUDAAN, and SPSS throgh coursework and summer seminars at the UM-School of Public Health, I felt fairly confident about the statistical analyses that would be needed to answer my questions regarding cultral variation in diet and obesity and its link to the Glycemic Index within the United States.

The only caveat was that the dataset (NHANES) did not actually include Glycemic Index values. But, the February 2004 publication by Brand-Miller (Journal of Clinical Nutrition) listed the values for thousands of food items, using the USDA food codes which are also used in the NHANES diet survey. Therefore, I didn't think it would be a big deal to have the computer calculate the glycemic index values for the food entries in NHANES, providing me with the necessary data.

Well, it didn't take long for me to figure out that this was simply not going to work, at least I wasn't going to get it to work. Therefore, I began exploring the next option - hiring a software engineer to go through the data set and perform the necessary conversions, etc. While thinking about pursuing this option, however, the Willet group and others, continued publishing on the irrelevance of single-food glycemic index values since foods were eaten in combination with other foods providing a mixed-meal glycemic value. In other words, the glycemic values for each individual food that is part of a meal are not necessarily additive. Willet demonstrated this by publishing the average values for a number of mixed meals. However, the almost infinite combinations of mixed meals in NHANES makes it virtually impossible (and fully impossible at this time because the data doesn't exist) to calculate the mixed-meal values for each dietary entry for every sample.

In other words, the glycemic index question had become all but irrelevant by Fall of 2004 and the calculations and management of the data set had become much more difficult then originally envisaged.

The notion of completing my dissertation was being pushed to a backburner - a very far back back burner - as the daily business of medical school took over, coupled with the questionable value of a PhD for a physician. Meaning, now that my data and project had fallen through, did it really matter? Would there be any affect of this on my career as a physician?

Well, the basic answer to this question is that if I was going to do family practice in some South Tampa group or emergency medicine in a busy urban hospital ,then, no, of course it would not matter. However, if I was going to continue exploring my interests in epidemiology, infectious disease, and preventive medicine, then a PhD (particulary from a "brand" name university like Michigan) would be quite valuable on the policy/research side of life.

I let these ideas stew in my head as I continued through Winter and Spring coursework of my first year in medical school, growing more distant from anthropology due to the clinically oriented environment of my everyday routines. However, the past few weeks have allowed me to refocus and reexplore the possibility and value of completing my dissertation due to some fortunate events.

First of all, the heavy coursework period has ended and we have a little more time in this block to make plans for a 2 month summer break beginning in June. This has given me the chance to read up on some anthro journals, look back at some of my ideas and talk with some graduate school colleagues.

Secondly, I have been able to put the past 9 months into perspective thanks to the help of Sumaiya. These past 9 months have been the first period in my life for the past 5 years where I didn't DO anthropology everyday, where I didn't feel like an anthropologist - whether that be teaching, research, courses, etc. The lack of contact with anthropology and anthropologists makes the 9 month period seem longer than it is. I had started feeling too far away from my time at Michigan to finish my dissertation. However, some time for reflection and some conversations with Sumaiya remind me that this 9 months is no big deal for a number of reasons:
1. I left with the encouragement of my advisor as part of a bigger plan
2. I am only in my 5th year of the PhD program - well below the average of 9 years to completion
3. Many bio-anthro students have been in the "field" for long periods (1-2 years, or more) or simply taken time to be in other places doing other things. Granted my "field" work is a little different, it doesn't really change things too much if my plans remain the same.
4. Rackham policy gives a doctoral candidate up to three years to complete the thesis

Finally, I have accepted a research position with Dr. John Sinnott who heads up the Infectious Disease unit at Tampa General Hospital, and is a clinical professor. This position, ultimately, may be the thing that allows me to realistically complete my thesis. Sinnott is a clinician, academic, and researcher, allowing allowing him to understand my unique circumstances (MD/PhD student at two different places in a program I created). In addition, and just as important, Dr. Sinnott has access to valuable data sets that are of clinical and anthropological interest. For example, the VRACS dataset is a large sample survey collected in India in order to assess risk factors for HIV. The group working on this dataset has produced papers previously with an anthropological question (SEE ABSTRACT) and there are many great questions stilil waiting to be answered. Fortunately for me, Dr. Sinnott has agreed to let me spend the summer using this data to complete my dissertation (!). I don't think I could have imagined a better opportunity.

For completing my PhD to be a reality, I set up a couple of regulations that had to met. These regulation appear IN ORDER - from most difficult to fulfill, to easiest to fulfill.
1. I needed to find a dataset that was already available (e.g. no fieldwork)
2. I needed to be able to do analysis and writing while in Tampa
3. I needed to find someone here who could act as a committee member/co-advisor
4. I needed to get approval from Frisancho

I think have definitely met requirements 1-3. Currently, I am in discussion with Frisancho to meet #4. Basically, he just wants to know my question - to know if its relevant to anthropology, to him, etc. I am not concerned about this - if anything I am concerned about his advancing age and desire to retire in the near-future. In order to get some insight into all of this, I have contacted Susan Tanner (a Frisancho student very near completion). Hopefully she can shed some light on things as she tends to be very down to earth and observant.

I am currently awaiting the VRACS data set, which is probably going to be a huge cumbersome EXCEL file that I will need to convert to SAS. The plan is that I will get to know the VRACS data set by teaching a little mini-SAS course to the residents and fellows at TGH (um, I hope I remember how to use SAS!). This will allow me to develop my question a bit while reading up on the literature and 2 previous papers that were presented at conferences from this data. Oh, speaking of literature - apparently my life is going to be one step easier - 2 high school aged volunteers with the infectious disease groups will be able to help me do literature searches, editing, etc. Let's get this thing moving fast before the data falls apart or HIV, India, or something else becomes irrelevant (ha).


Adam said...

What aspects of this new data set are of particular interest to you and what questions have already been asked of it?

a living hominid said...

Previous questions have concerned basic epidemiological statistics for the most part.

However, the data is from non-urban Indian populations - therefore, there is some potential for examining urban HIV vs. tribal group HIV infection rates in rural areas and asking questions concerning globalization and its relationship to communicable disease education and prevention - meaning, in areas of India where people might know a little bit more about communicable disease due to a greater impact of market forces, etc, does this actually translate into any decrease in rates of infection? What do the infection rates look like in these outlying groups? If they are higher/lower, why? What types of culturally centered promotion/prevention approaches could be focused towards these groups.