As my main project with Aravind’s eye bank wraps up, I think it’s time I blog about it. I have been hesitant to write about my actual work at the hospital because throughout much of my time working on the eye bank study, I was frustrated with my progress (or seemingly lack there of). While I have been having a valuable learning experience seeing the groundwork of a health initiative not unlike those I have studied in my courses at Penn, I felt like I myself was not giving back anything meaningful to the hospital.
Let me back up.
Aravind Pondicherry has an eye bank within its hospital which collects corneal donations from a few sources: volunteer organizations, other hospitals, and trained Aravind doctors who do home retrievals themselves. Upon acquiring these donations, eye bank specialists evaluate the donation quality and determine if there are any surgeries for which it may be used. Those tissues suitable for surgery are then stored until an appropriate recipient is available. Those which are not suitable for surgery or whose storage time is exhausted are typically used for research and training. All records of donations as of May 2017 were done with pen and paper and stored in carefully labeled binders.
After I arrived on May 25th and told the hospital about my statistical background, my supervisors proposed I work with the eye bank. They explained to me how, as of then, they hadn’t considered any data about the eye donations, but they suspected there were potentially avoidable gaps in the number of tissues approved for surgery and the number actual used for surgery. In other words, perhaps the utilization rate of cornea donations could be improved, but without the data there was no way of knowing.
This all sounded exciting to me—delving into the different parameters which could be affecting utilization. The possible pertinent factors were the cause of donor death, the difference in time of death and enucleation (eye removal), difference in enucleation and preservation, duration of preservation before surgery, age of the donor, sex of the donor, which organization collected the donor tissue…this list goes on. I was ready to get to work on the data set.
That is, until I learned the data set didn’t exist. And because Aravind Pondicherry doesn’t have a statistics department, even if the data did exist I wouldn’t have had access to the Stata package I would’ve needed to begin analysis anyway. Alas, it seemed like a minor roadblock. Two weeks of manual data entry loomed ahead of me, but I hoped during that time the hospital would be able to secure Stata software for me to use once finished.
I’ll spare you the details, but essentially what I had thought would take me 12 work days to enter (working 6 days a week) ended up taking nearly 4 weeks due to some miscommunication. Further, once I finally finished the tedious task of entering all the donor information in the spreadsheet, I found out obtaining Stata was, in fact, a problem (despite initially being assured otherwise). It would be impossible to install the expensive software in Pondicherry for free, and it would be useless after I left because no one here is familiar with the language. It would be a wasted investment for the cost-conscious eye hospital.
I felt like I had just spent the past month doing perhaps the worst part of data work— the entry, a mindless task anyone could complete— only to be told I wouldn’t be able to participate in the analysis which I had looked forward to the most. My data set would just have to be sent off the Madurai statistics team.
When I met with my supervisor to show him the final excel file, we decided it would be best to additionally use Madurai’s eye bank data before giving it to a statistician there to investigate. I knew it would be better to have more observations, but I couldn’t help but wonder who in the main tertiary care center would sit down and do the painfully slow data entry I had done? Who would take on that task, given there was no one here in Pondicherry to do it before I came? I feared the excel file I had slaved over (okay fine, the work wasn’t laborious, but it did nearly bore me to tears and was very time consuming) would be emailed to Madurai, only to gather digital dust in someone’s inbox.
As I began woefully putting together a formal research proposal to send with my data, the thought occurred to me that the person who would receive the half-finished data set might be more compelled to see it completed if the numbers crunched so far were intriguing. I didn’t have Stata to do a preliminary analysis, but I did have R, a free program.
I am less familiar with the open source software, but with the reliable help of google.com I managed to come up with some (unconcise) code and interesting findings after all. The statistic which stuck out to me the most was that for the donations deemed suitable for surgery and never used, 70% did not have a donor blood sample collected. Because an operation cannot be completed without a blood sample, presumably the cause for the non-utilization was not the lack of waiting recipients, but rather the lack of serum availability. This indicates taking a blood sample is missing from the protocols of some of Aravind’s donation collectors. The next step would be determining what it would take to change those protocols so as to improve utilization rates.
When I showed the preliminary stats I found to my supervisor and the doctor I was working with in the cornea clinic before we sent the research proposal to the Madurai team, they were enthusiastic about the possibilities that lie ahead with the project. For the first time, four or five weeks in, I felt a sense of purpose in my work.
Since then, we’ve done a phone meeting with the manager of the eye bank in Madurai, and I am much more confident this research will continue after I leave. It took a while before I felt I was making any progress here, but truth be told it wasn’t all bad along the way. Throughout my weeks in front of my computer and 600 donor records, as gloomy as the featured picture looks, the women who worked in the eye bank gave me something to look forward to when I went in every day. Not only did they retrieve the records I required, translate the papers written only in Tamil, and clarify handwriting when needed, they kept me company. Without the data entry, I wouldn’t have those friendships.
Looking back, it’s easy to say I wish I hadn’t gotten so discouraged at times. I wish I could have avoided the miscommunication which resulted in more data entry time. I wish I hadn’t gotten so hung up on using Stata, and considered using R earlier. I’ve been lucky enough with my job doing data analysis at Penn that I haven’t needed to do large-scale entry before. But I realize that, as an undergraduate, this may not be the last time I’m tasked with it. Nevertheless, I now have the foresight to keep an open mind should I come face to face a 3-foot stack of handwritten records again.